d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
4046851367
freebsd-dev/sys/dev/mmc/mmc.c

2584 lines
74 KiB
C
Raw Normal View History

Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
/*-
sys/dev: further adoption of SPDX licensing ID tags. Mainly focus on files that use BSD 2-Clause license, however the tool I was using misidentified many licenses so this was mostly a manual - error prone - task. The Software Package Data Exchange (SPDX) group provides a specification to make it easier for automated tools to detect and summarize well known opensource licenses. We are gradually adopting the specification, noting that the tags are considered only advisory and do not, in any way, superceed or replace the license texts.
2017-11-27 14:52:40 +00:00
* SPDX-License-Identifier: BSD-2-Clause-FreeBSD
*
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
* Copyright (c) 2006 Bernd Walter. All rights reserved.
* Copyright (c) 2006 M. Warner Losh. All rights reserved.
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
* Copyright (c) 2017 Marius Strobl <marius@FreeBSD.org>
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
* IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
* NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
* THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
A careful reading of the disclaimer that is required to download the SD Simplified specification, as well as other SD and SDIO implemenations I've examined, suggest this disclaimer may be required. It is unclear to me exactly what the license would be for, or why it might be required. Err on the side of caution and include this disclaimer so anybody deploying this code can judge for themselves. I have no further unformation about the details.
2007-05-26 05:23:36 +00:00
*
* Portions of this software may have been developed with reference to
* the SD Simplified Specification. The following disclaimer may apply:
*
* The following conditions apply to the release of the simplified
* specification ("Simplified Specification") by the SD Card Association and
* the SD Group. The Simplified Specification is a subset of the complete SD
* Specification which is owned by the SD Card Association and the SD
* Group. This Simplified Specification is provided on a non-confidential
* basis subject to the disclaimers below. Any implementation of the
* Simplified Specification may require a license from the SD Card
* Association, SD Group, SD-3C LLC or other third parties.
*
* Disclaimers:
*
* The information contained in the Simplified Specification is presented only
* as a standard specification for SD Cards and SD Host/Ancillary products and
* is provided "AS-IS" without any representations or warranties of any
* kind. No responsibility is assumed by the SD Group, SD-3C LLC or the SD
* Card Association for any damages, any infringements of patents or other
* right of the SD Group, SD-3C LLC, the SD Card Association or any third
* parties, which may result from its use. No license is granted by
* implication, estoppel or otherwise under any patent or other rights of the
* SD Group, SD-3C LLC, the SD Card Association or any third party. Nothing
* herein shall be construed as an obligation by the SD Group, the SD-3C LLC
* or the SD Card Association to disclose or distribute any technical
* information, know-how or other confidential information to any third party.
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/kernel.h>
#include <sys/malloc.h>
#include <sys/lock.h>
#include <sys/module.h>
#include <sys/mutex.h>
#include <sys/bus.h>
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
#include <sys/endian.h>
Improve mmc driver verbose logging. Make requests logging controllable by hw.mmc.debug sysctl.
2009-01-28 22:09:00 +00:00
#include <sys/sysctl.h>
Log mmc and sd command failures. Reporting of routine expected errors, such as timeouts while probing a bus or testing for a feature, is squelched. Also, error reporting is limited to 5 events per second, because when an sdcard goes bad on a low-end embedded board, flooding the console at high speed isn't helpful. Original logging code contributed by Michal Meloun, but then I fancied it up with squelching and ppsratecheck.
2014-12-20 04:24:40 +00:00
#include <sys/time.h>
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
#include <dev/mmc/bridge.h>
#include <dev/mmc/mmc_private.h>
#include <dev/mmc/mmc_subr.h>
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
#include <dev/mmc/mmcreg.h>
#include <dev/mmc/mmcbrvar.h>
#include <dev/mmc/mmcvar.h>
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
#include "mmcbr_if.h"
#include "mmcbus_if.h"
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
CTASSERT(bus_timing_max <= sizeof(uint32_t) * NBBY);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
/*
* Per-card data
*/
struct mmc_ivars {
uint32_t raw_cid[4]; /* Raw bits of the CID */
uint32_t raw_csd[4]; /* Raw bits of the CSD */
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
uint32_t raw_scr[2]; /* Raw bits of the SCR */
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
uint8_t raw_ext_csd[MMC_EXTCSD_SIZE]; /* Raw bits of the EXT_CSD */
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
uint32_t raw_sd_status[16]; /* Raw bits of the SD_STATUS */
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
uint16_t rca;
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
u_char read_only; /* True when the device is read-only */
u_char high_cap; /* High Capacity device (block addressed) */
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
enum mmc_card_mode mode;
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
enum mmc_bus_width bus_width; /* Bus width to use */
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
struct mmc_cid cid; /* cid decoded */
struct mmc_csd csd; /* csd decoded */
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
struct mmc_scr scr; /* scr decoded */
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
struct mmc_sd_status sd_status; /* SD_STATUS decoded */
Add high capacity MMC cards support.
2008-10-09 20:09:56 +00:00
uint32_t sec_count; /* Card capacity in 512byte blocks */
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
uint32_t timings; /* Mask of bus timings supported */
uint32_t vccq_120; /* Mask of bus timings at VCCQ of 1.2 V */
uint32_t vccq_180; /* Mask of bus timings at VCCQ of 1.8 V */
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
uint32_t tran_speed; /* Max speed in normal mode */
uint32_t hs_tran_speed; /* Max speed in high speed mode */
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
uint32_t erase_sector; /* Card native erase sector size */
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
uint32_t cmd6_time; /* Generic switch timeout [us] */
- If available, use TRIM instead of ERASE for implementing BIO_DELETE. This also involves adding a quirk table as TRIM is broken for some Kingston eMMC devices, though. Compared to ERASE (declared "legacy" in the eMMC specification v5.1), TRIM has the advantage of operating on write sectors rather than on erase sectors, which typically are of a much larger size. Thus, employing TRIM, we don't need to fiddle with coalescing BIO_DELETE requests that are also of (write) sector units into erase sectors, which might not even add up in all cases. - For some SanDisk iNAND devices, the CMD38 argument, e. g. ERASE, TRIM etc., has to be specified via EXT_CSD[113], which now is also handled via a quirk. - My initial understanding was that for eMMC partitions, the granularity should be used as erase sector size, e. g. 128 KB for boot partitions. However, rereading the relevant parts of the eMMC specification v5.1, this isn't actually correct. So drop the code which used partition granularities for delmaxsize and stripesize. For the most part, this change is a NOP, though, because a) for ERASE, mmcsd_delete() used the erase sector size unconditionally for all partitions anyway and b) g_disk_limit() doesn't actually take the stripesize into account. - Take some more advantage of mmcsd_errmsg() in mmcsd(4) for making error codes human readable.
2017-08-07 23:33:05 +00:00
uint32_t quirks; /* Quirks as per mmc_quirk->quirks */
o Fixes: - When switching to 4-bit operation, send a SET_CLR_CARD_DETECT command to disconnect the card-detect pull-up resistor from the DAT3 line before sending the SET_BUS_WIDTH command. - Add the missing "reserved" zero entry to the mantissa table used to decode various CSD fields. This was causing SD cards to report that they could run at 30 MHz instead of the maximum 25 MHz mandated in the spec. o Enhancements: - At the MMC layer, format various info from the CID into a string that uniquely identifies the card instance (manufacturer number, serial number, product name and revision, etc). Export it as an instance variable. - At the MMCSD layer, display the formatted card ID string, and also report the clock speed of the hardware (not the card's max speed), and the number of bits and number of blocks per transfer. It comes out like this now: mmcsd0: 968MB <SD SD01G 8.0 SN 276886905 MFG 08/2008 by 3 SD> at mmc0 22.5MHz/4bit/128-block o Use DEVMETHOD_END. o Use NULL instead of 0 for pointers. PR: 156496 Submitted by: Ian Lepore MFC after: 1 week
2012-04-21 01:51:16 +00:00
char card_id_string[64];/* Formatted CID info (serial, MFG, etc) */
Export an mmc or sd card's serial number from the mmc layer as an ivar. In the mmcsd layer use this value to populate disk->d_ident. Also set disk->d_descr to the full set of card identification info (includes vendor, model, manufacturing date, etc).
2014-07-31 16:54:54 +00:00
char card_sn_string[16];/* Formatted serial # for disk->d_ident */
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
};
Fix some more overly long lines, whitespace and other bugs according to style(9) as well as spelling in comments.
2017-02-04 19:35:38 +00:00
#define CMD_RETRIES 3
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
- If available, use TRIM instead of ERASE for implementing BIO_DELETE. This also involves adding a quirk table as TRIM is broken for some Kingston eMMC devices, though. Compared to ERASE (declared "legacy" in the eMMC specification v5.1), TRIM has the advantage of operating on write sectors rather than on erase sectors, which typically are of a much larger size. Thus, employing TRIM, we don't need to fiddle with coalescing BIO_DELETE requests that are also of (write) sector units into erase sectors, which might not even add up in all cases. - For some SanDisk iNAND devices, the CMD38 argument, e. g. ERASE, TRIM etc., has to be specified via EXT_CSD[113], which now is also handled via a quirk. - My initial understanding was that for eMMC partitions, the granularity should be used as erase sector size, e. g. 128 KB for boot partitions. However, rereading the relevant parts of the eMMC specification v5.1, this isn't actually correct. So drop the code which used partition granularities for delmaxsize and stripesize. For the most part, this change is a NOP, though, because a) for ERASE, mmcsd_delete() used the erase sector size unconditionally for all partitions anyway and b) g_disk_limit() doesn't actually take the stripesize into account. - Take some more advantage of mmcsd_errmsg() in mmcsd(4) for making error codes human readable.
2017-08-07 23:33:05 +00:00
static const struct mmc_quirk mmc_quirks[] = {
/*
* For some SanDisk iNAND devices, the CMD38 argument needs to be
* provided in EXT_CSD[113].
*/
{ 0x2, 0x100, "SEM02G", MMC_QUIRK_INAND_CMD38 },
{ 0x2, 0x100, "SEM04G", MMC_QUIRK_INAND_CMD38 },
{ 0x2, 0x100, "SEM08G", MMC_QUIRK_INAND_CMD38 },
{ 0x2, 0x100, "SEM16G", MMC_QUIRK_INAND_CMD38 },
{ 0x2, 0x100, "SEM32G", MMC_QUIRK_INAND_CMD38 },
/*
* Disable TRIM for Kingston eMMCs where a firmware bug can lead to
* unrecoverable data corruption.
*/
{ 0x70, MMC_QUIRK_OID_ANY, "V10008", MMC_QUIRK_BROKEN_TRIM },
{ 0x70, MMC_QUIRK_OID_ANY, "V10016", MMC_QUIRK_BROKEN_TRIM },
{ 0x0, 0x0, NULL, 0x0 }
};
Mark all SYSCTL_NODEs static that have no corresponding SYSCTL_DECLs. The SYSCTL_NODE macro defines a list that stores all child-elements of that node. If there's no SYSCTL_DECL macro anywhere else, there's no reason why it shouldn't be static.
2011-11-07 15:43:11 +00:00
static SYSCTL_NODE(_hw, OID_AUTO, mmc, CTLFLAG_RD, NULL, "mmc driver");
Improve mmc driver verbose logging. Make requests logging controllable by hw.mmc.debug sysctl.
2009-01-28 22:09:00 +00:00
o Define some symbols for a few items that are bare constants in the code. o Use NULL in preference to 0 for a few pointers. o default to bus timing normal, like we default to bus_width_1.
2009-02-03 04:28:45 +00:00
static int mmc_debug;
Fix some more overly long lines, whitespace and other bugs according to style(9) as well as spelling in comments.
2017-02-04 19:35:38 +00:00
SYSCTL_INT(_hw_mmc, OID_AUTO, debug, CTLFLAG_RWTUN, &mmc_debug, 0,
"Debug level");
Improve mmc driver verbose logging. Make requests logging controllable by hw.mmc.debug sysctl.
2009-01-28 22:09:00 +00:00
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
/* bus entry points */
Add missing prototypes. While at it, sort them alphabetically. MFC after: 3 days
2012-06-02 20:47:00 +00:00
static int mmc_acquire_bus(device_t busdev, device_t dev);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
static int mmc_attach(device_t dev);
Add missing prototypes. While at it, sort them alphabetically. MFC after: 3 days
2012-06-02 20:47:00 +00:00
static int mmc_child_location_str(device_t dev, device_t child, char *buf,
size_t buflen);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
static int mmc_detach(device_t dev);
Add missing prototypes. While at it, sort them alphabetically. MFC after: 3 days
2012-06-02 20:47:00 +00:00
static int mmc_probe(device_t dev);
static int mmc_read_ivar(device_t bus, device_t child, int which,
uintptr_t *result);
static int mmc_release_bus(device_t busdev, device_t dev);
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
static int mmc_resume(device_t dev);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
static void mmc_retune_pause(device_t busdev, device_t dev, bool retune);
static void mmc_retune_unpause(device_t busdev, device_t dev);
Add missing prototypes. While at it, sort them alphabetically. MFC after: 3 days
2012-06-02 20:47:00 +00:00
static int mmc_suspend(device_t dev);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
static int mmc_wait_for_request(device_t busdev, device_t dev,
Add missing prototypes. While at it, sort them alphabetically. MFC after: 3 days
2012-06-02 20:47:00 +00:00
struct mmc_request *req);
static int mmc_write_ivar(device_t bus, device_t child, int which,
uintptr_t value);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
Fix some more overly long lines, whitespace and other bugs according to style(9) as well as spelling in comments.
2017-02-04 19:35:38 +00:00
#define MMC_LOCK(_sc) mtx_lock(&(_sc)->sc_mtx)
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
#define MMC_UNLOCK(_sc) mtx_unlock(&(_sc)->sc_mtx)
Fix some more overly long lines, whitespace and other bugs according to style(9) as well as spelling in comments.
2017-02-04 19:35:38 +00:00
#define MMC_LOCK_INIT(_sc) \
mtx_init(&(_sc)->sc_mtx, device_get_nameunit((_sc)->dev), \
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
"mmc", MTX_DEF)
Fix some more overly long lines, whitespace and other bugs according to style(9) as well as spelling in comments.
2017-02-04 19:35:38 +00:00
#define MMC_LOCK_DESTROY(_sc) mtx_destroy(&(_sc)->sc_mtx);
#define MMC_ASSERT_LOCKED(_sc) mtx_assert(&(_sc)->sc_mtx, MA_OWNED);
#define MMC_ASSERT_UNLOCKED(_sc) mtx_assert(&(_sc)->sc_mtx, MA_NOTOWNED);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
Add missing prototypes. While at it, sort them alphabetically. MFC after: 3 days
2012-06-02 20:47:00 +00:00
static int mmc_all_send_cid(struct mmc_softc *sc, uint32_t *rawcid);
static void mmc_app_decode_scr(uint32_t *raw_scr, struct mmc_scr *scr);
static void mmc_app_decode_sd_status(uint32_t *raw_sd_status,
struct mmc_sd_status *sd_status);
static int mmc_app_sd_status(struct mmc_softc *sc, uint16_t rca,
uint32_t *rawsdstatus);
static int mmc_app_send_scr(struct mmc_softc *sc, uint16_t rca,
uint32_t *rawscr);
Give mmcsd driver a bit more information about card. It allows to reorganize log message in a way a bit more common for disk devices. Also it will allow mmcsd driver to use MMC/SD specific commands when needed.
2008-10-11 13:05:13 +00:00
static int mmc_calculate_clock(struct mmc_softc *sc);
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
static void mmc_decode_cid_mmc(uint32_t *raw_cid, struct mmc_cid *cid,
bool is_4_41p);
Add missing prototypes. While at it, sort them alphabetically. MFC after: 3 days
2012-06-02 20:47:00 +00:00
static void mmc_decode_cid_sd(uint32_t *raw_cid, struct mmc_cid *cid);
static void mmc_decode_csd_mmc(uint32_t *raw_csd, struct mmc_csd *csd);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
static int mmc_decode_csd_sd(uint32_t *raw_csd, struct mmc_csd *csd);
Add missing prototypes. While at it, sort them alphabetically. MFC after: 3 days
2012-06-02 20:47:00 +00:00
static void mmc_delayed_attach(void *xsc);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
static int mmc_delete_cards(struct mmc_softc *sc, bool final);
Add missing prototypes. While at it, sort them alphabetically. MFC after: 3 days
2012-06-02 20:47:00 +00:00
static void mmc_discover_cards(struct mmc_softc *sc);
static void mmc_format_card_id_string(struct mmc_ivars *ivar);
static void mmc_go_discovery(struct mmc_softc *sc);
static uint32_t mmc_get_bits(uint32_t *bits, int bit_len, int start,
int size);
static int mmc_highest_voltage(uint32_t ocr);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
static bool mmc_host_timing(device_t dev, enum mmc_bus_timing timing);
Add missing prototypes. While at it, sort them alphabetically. MFC after: 3 days
2012-06-02 20:47:00 +00:00
static void mmc_idle_cards(struct mmc_softc *sc);
static void mmc_ms_delay(int ms);
static void mmc_log_card(device_t dev, struct mmc_ivars *ivar, int newcard);
Implement power down, and power down the bus on detach. Submitted by: mav@
2008-09-28 23:24:52 +00:00
static void mmc_power_down(struct mmc_softc *sc);
Add missing prototypes. While at it, sort them alphabetically. MFC after: 3 days
2012-06-02 20:47:00 +00:00
static void mmc_power_up(struct mmc_softc *sc);
static void mmc_rescan_cards(struct mmc_softc *sc);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
static int mmc_retune(device_t busdev, device_t dev, bool reset);
Add missing prototypes. While at it, sort them alphabetically. MFC after: 3 days
2012-06-02 20:47:00 +00:00
static void mmc_scan(struct mmc_softc *sc);
static int mmc_sd_switch(struct mmc_softc *sc, uint8_t mode, uint8_t grp,
uint8_t value, uint8_t *res);
static int mmc_select_card(struct mmc_softc *sc, uint16_t rca);
static uint32_t mmc_select_vdd(struct mmc_softc *sc, uint32_t ocr);
static int mmc_send_app_op_cond(struct mmc_softc *sc, uint32_t ocr,
uint32_t *rocr);
static int mmc_send_csd(struct mmc_softc *sc, uint16_t rca, uint32_t *rawcsd);
static int mmc_send_if_cond(struct mmc_softc *sc, uint8_t vhs);
static int mmc_send_op_cond(struct mmc_softc *sc, uint32_t ocr,
uint32_t *rocr);
static int mmc_send_relative_addr(struct mmc_softc *sc, uint32_t *resp);
static int mmc_set_blocklen(struct mmc_softc *sc, uint32_t len);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
static int mmc_set_card_bus_width(struct mmc_softc *sc, struct mmc_ivars *ivar,
enum mmc_bus_timing timing);
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
static int mmc_set_power_class(struct mmc_softc *sc, struct mmc_ivars *ivar);
Add missing prototypes. While at it, sort them alphabetically. MFC after: 3 days
2012-06-02 20:47:00 +00:00
static int mmc_set_relative_addr(struct mmc_softc *sc, uint16_t resp);
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
static int mmc_set_timing(struct mmc_softc *sc, struct mmc_ivars *ivar,
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
enum mmc_bus_timing timing);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
static int mmc_set_vccq(struct mmc_softc *sc, struct mmc_ivars *ivar,
enum mmc_bus_timing timing);
static int mmc_switch_to_hs200(struct mmc_softc *sc, struct mmc_ivars *ivar,
uint32_t clock);
static int mmc_switch_to_hs400(struct mmc_softc *sc, struct mmc_ivars *ivar,
uint32_t max_dtr, enum mmc_bus_timing max_timing);
Add missing prototypes. While at it, sort them alphabetically. MFC after: 3 days
2012-06-02 20:47:00 +00:00
static int mmc_test_bus_width(struct mmc_softc *sc);
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
static uint32_t mmc_timing_to_dtr(struct mmc_ivars *ivar,
enum mmc_bus_timing timing);
static const char *mmc_timing_to_string(enum mmc_bus_timing timing);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
static void mmc_update_child_list(struct mmc_softc *sc);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
static int mmc_wait_for_command(struct mmc_softc *sc, uint32_t opcode,
uint32_t arg, uint32_t flags, uint32_t *resp, int retries);
Add missing prototypes. While at it, sort them alphabetically. MFC after: 3 days
2012-06-02 20:47:00 +00:00
static int mmc_wait_for_req(struct mmc_softc *sc, struct mmc_request *req);
static void mmc_wakeup(struct mmc_request *req);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
static void
mmc_ms_delay(int ms)
{
Add missing prototypes. While at it, sort them alphabetically. MFC after: 3 days
2012-06-02 20:47:00 +00:00
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
DELAY(1000 * ms); /* XXX BAD */
}
static int
mmc_probe(device_t dev)
{
Other busses on the system use the official capitalization for the technology. Make mmc conform to that. Submitted by: mav@
2008-09-28 22:23:06 +00:00
device_set_desc(dev, "MMC/SD bus");
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
return (0);
}
static int
mmc_attach(device_t dev)
{
struct mmc_softc *sc;
sc = device_get_softc(dev);
sc->dev = dev;
MMC_LOCK_INIT(sc);
/* We'll probe and attach our children later, but before / mount */
sc->config_intrhook.ich_func = mmc_delayed_attach;
sc->config_intrhook.ich_arg = sc;
if (config_intrhook_establish(&sc->config_intrhook) != 0)
device_printf(dev, "config_intrhook_establish failed\n");
return (0);
}
static int
mmc_detach(device_t dev)
{
First cut at making detach work. also add sdh as a possible mmc bridge. Submitted by: Andrea Bittau (Andrea may have updated patches, but I've tested these)
2007-05-15 05:49:14 +00:00
struct mmc_softc *sc = device_get_softc(dev);
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
int err;
First cut at making detach work. also add sdh as a possible mmc bridge. Submitted by: Andrea Bittau (Andrea may have updated patches, but I've tested these)
2007-05-15 05:49:14 +00:00
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
err = mmc_delete_cards(sc, true);
if (err != 0)
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
return (err);
Implement power down, and power down the bus on detach. Submitted by: mav@
2008-09-28 23:24:52 +00:00
mmc_power_down(sc);
First cut at making detach work. also add sdh as a possible mmc bridge. Submitted by: Andrea Bittau (Andrea may have updated patches, but I've tested these)
2007-05-15 05:49:14 +00:00
MMC_LOCK_DESTROY(sc);
Conform to style(9) for return (foo); The files were a mix before. Submitted by: mav@
2008-09-29 18:05:26 +00:00
return (0);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
}
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
static int
mmc_suspend(device_t dev)
{
struct mmc_softc *sc = device_get_softc(dev);
int err;
err = bus_generic_suspend(dev);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
if (err != 0)
return (err);
/*
* We power down with the bus acquired here, mainly so that no device
* is selected any longer and sc->last_rca gets set to 0. Otherwise,
* the deselect as part of the bus acquisition in mmc_scan() may fail
* during resume, as the bus isn't powered up again before later in
* mmc_go_discovery().
*/
err = mmc_acquire_bus(dev, dev);
if (err != 0)
Fix overly long lines, whitespace and other bugs according to style(9).
2017-01-29 00:05:49 +00:00
return (err);
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
mmc_power_down(sc);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
err = mmc_release_bus(dev, dev);
return (err);
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
}
static int
mmc_resume(device_t dev)
{
struct mmc_softc *sc = device_get_softc(dev);
mmc_scan(sc);
return (bus_generic_resume(dev));
}
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
static int
mmc_acquire_bus(device_t busdev, device_t dev)
{
struct mmc_softc *sc;
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
struct mmc_ivars *ivar;
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
int err;
uint16_t rca;
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
enum mmc_bus_timing timing;
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
The parameters to the MMCBR_ACQUIRE_HOST and MMCBR_RELEAES_HOST were the device in question, rather than the bus doing the requesting. Fix it so that it is the bus. Submitted by: mav@
2008-09-29 01:28:30 +00:00
err = MMCBR_ACQUIRE_HOST(device_get_parent(busdev), busdev);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
if (err)
return (err);
sc = device_get_softc(busdev);
MMC_LOCK(sc);
if (sc->owner)
- Fix some typos in mmc_acquire_bus() and mmc_send_csd(). - Fix some math errors in mmc_decode_csd_sd(). - Fix incorrect arguments to mmc_send_app_op_cond() in mmc_go_discovery(). - Add reporting of CSD for debug purposes. - Add detection (and skipping) of password-locked cards. - Add setting of block length on card if necessary. Submitted by: Patrick Kelsey MFC after: 3 days
2012-05-27 16:08:58 +00:00
panic("mmc: host bridge didn't serialize us.");
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
sc->owner = dev;
MMC_UNLOCK(sc);
if (busdev != dev) {
Convert all C++ comments to C style. There's lots of folks that don't like them and they don't add enough to be worth standing out.
2008-09-29 01:32:21 +00:00
/*
* Keep track of the last rca that we've selected. If
* we're asked to do it again, don't. We never
* unselect unless the bus code itself wants the mmc
* bus, and constantly reselecting causes problems.
*/
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
ivar = device_get_ivars(dev);
rca = ivar->rca;
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
if (sc->last_rca != rca) {
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
if (mmc_select_card(sc, rca) != MMC_ERR_NONE) {
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
device_printf(busdev, "Card at relative "
"address %d failed to select\n", rca);
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
return (ENXIO);
}
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
sc->last_rca = rca;
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
timing = mmcbr_get_timing(busdev);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
/*
* For eMMC modes, setting/updating bus width and VCCQ
* only really is necessary if there actually is more
* than one device on the bus as generally that already
* had to be done by mmc_calculate_clock() or one of
* its calees. Moreover, setting the bus width anew
* can trigger re-tuning (via a CRC error on the next
* CMD), even if not switching between devices an the
* previously selected one is still tuned. Obviously,
* we need to re-tune the host controller if devices
* are actually switched, though.
*/
if (timing >= bus_timing_mmc_ddr52 &&
sc->child_count == 1)
return (0);
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
/* Prepare bus width for the new card. */
Improve mmc driver verbose logging. Make requests logging controllable by hw.mmc.debug sysctl.
2009-01-28 22:09:00 +00:00
if (bootverbose || mmc_debug) {
Give mmcsd driver a bit more information about card. It allows to reorganize log message in a way a bit more common for disk devices. Also it will allow mmcsd driver to use MMC/SD specific commands when needed.
2008-10-11 13:05:13 +00:00
device_printf(busdev,
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
"setting bus width to %d bits %s timing\n",
style(9): spaces around operators.
2008-10-12 07:30:05 +00:00
(ivar->bus_width == bus_width_4) ? 4 :
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
(ivar->bus_width == bus_width_8) ? 8 : 1,
mmc_timing_to_string(timing));
}
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
if (mmc_set_card_bus_width(sc, ivar, timing) !=
MMC_ERR_NONE) {
device_printf(busdev, "Card at relative "
"address %d failed to set bus width\n",
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
rca);
return (ENXIO);
}
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
mmcbr_set_bus_width(busdev, ivar->bus_width);
mmcbr_update_ios(busdev);
if (mmc_set_vccq(sc, ivar, timing) != MMC_ERR_NONE) {
device_printf(busdev, "Failed to set VCCQ "
"for card at relative address %d\n", rca);
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
return (ENXIO);
}
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
if (timing >= bus_timing_mmc_hs200 &&
mmc_retune(busdev, dev, true) != 0) {
device_printf(busdev, "Card at relative "
"address %d failed to re-tune\n", rca);
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
return (ENXIO);
Give mmcsd driver a bit more information about card. It allows to reorganize log message in a way a bit more common for disk devices. Also it will allow mmcsd driver to use MMC/SD specific commands when needed.
2008-10-11 13:05:13 +00:00
}
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
}
} else {
Convert all C++ comments to C style. There's lots of folks that don't like them and they don't add enough to be worth standing out.
2008-09-29 01:32:21 +00:00
/*
* If there's a card selected, stand down.
*/
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
if (sc->last_rca != 0) {
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
if (mmc_select_card(sc, 0) != MMC_ERR_NONE)
return (ENXIO);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
sc->last_rca = 0;
}
}
return (0);
}
static int
mmc_release_bus(device_t busdev, device_t dev)
{
struct mmc_softc *sc;
int err;
sc = device_get_softc(busdev);
MMC_LOCK(sc);
if (!sc->owner)
panic("mmc: releasing unowned bus.");
if (sc->owner != dev)
panic("mmc: you don't own the bus. game over.");
MMC_UNLOCK(sc);
The parameters to the MMCBR_ACQUIRE_HOST and MMCBR_RELEAES_HOST were the device in question, rather than the bus doing the requesting. Fix it so that it is the bus. Submitted by: mav@
2008-09-29 01:28:30 +00:00
err = MMCBR_RELEASE_HOST(device_get_parent(busdev), busdev);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
if (err)
return (err);
MMC_LOCK(sc);
sc->owner = NULL;
MMC_UNLOCK(sc);
return (0);
}
static uint32_t
mmc_select_vdd(struct mmc_softc *sc, uint32_t ocr)
{
The OCR register defines both acceptable voltage bits, as well as bits for other things. Mask out the voltage only bits when returning the Vdd mask for voltage computation. Submitted by: mav@
2008-09-28 22:27:29 +00:00
Conform to style(9) for return (foo); The files were a mix before. Submitted by: mav@
2008-09-29 18:05:26 +00:00
return (ocr & MMC_OCR_VOLTAGE);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
}
static int
mmc_highest_voltage(uint32_t ocr)
{
int i;
Fix highest voltage detection Submitted by: Daisuke Aoyama <aoyama at peach.ne.jp> Approved by: imp@
2013-01-21 23:24:16 +00:00
for (i = MMC_OCR_MAX_VOLTAGE_SHIFT;
i >= MMC_OCR_MIN_VOLTAGE_SHIFT; i--)
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
if (ocr & (1 << i))
Conform to style(9) for return (foo); The files were a mix before. Submitted by: mav@
2008-09-29 18:05:26 +00:00
return (i);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
return (-1);
}
static void
mmc_wakeup(struct mmc_request *req)
{
struct mmc_softc *sc;
sc = (struct mmc_softc *)req->done_data;
MMC_LOCK(sc);
req->flags |= MMC_REQ_DONE;
MMC_UNLOCK(sc);
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
wakeup(req);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
}
static int
mmc_wait_for_req(struct mmc_softc *sc, struct mmc_request *req)
{
req->done = mmc_wakeup;
req->done_data = sc;
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
if (__predict_false(mmc_debug > 1)) {
Improve mmc driver verbose logging. Make requests logging controllable by hw.mmc.debug sysctl.
2009-01-28 22:09:00 +00:00
device_printf(sc->dev, "REQUEST: CMD%d arg %#x flags %#x",
req->cmd->opcode, req->cmd->arg, req->cmd->flags);
if (req->cmd->data) {
printf(" data %d\n", (int)req->cmd->data->len);
} else
printf("\n");
}
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
MMCBR_REQUEST(device_get_parent(sc->dev), sc->dev, req);
MMC_LOCK(sc);
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
while ((req->flags & MMC_REQ_DONE) == 0)
msleep(req, &sc->sc_mtx, 0, "mmcreq", 0);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
MMC_UNLOCK(sc);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
if (__predict_false(mmc_debug > 2 || (mmc_debug > 0 &&
req->cmd->error != MMC_ERR_NONE)))
sdhci/mmc: Minor whitespace cleanups No functional change. Submitted by: Johannes Lundberg <yohanesu75 at gmail.com>
2016-12-20 03:38:14 +00:00
device_printf(sc->dev, "CMD%d RESULT: %d\n",
Handle command retries for commands originating at the mmc layer, and ensure that all such commands have a non-zero retry count except for those that are expected to fail (for example, because they are used to probe for feature support). While it is possible to pass a retry count down to the hardware driver in the command request structure, no hardware driver currently implements any retry logic. The hardware doesn't know much about the context of a single request, so it makes more sense to handle retries at a layer that does. This adds retry loops to the mmc_wait_for_cmd() and mmc_wait_for_app_cmd() functions. These functions are the gateway from other code within mmc.c to the hardware. App commands are a sequence of two commands and a retry has to rerun both of them in order, so it needs its own retry loop. Retry looping is specifically NOT implemented in mmc_wait_for_request() because it is the gateway for children on the bus, and they have to implement their own retry logic depending on what makes sense for them.
2013-08-16 23:05:34 +00:00
req->cmd->opcode, req->cmd->error);
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
return (0);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
}
static int
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
mmc_wait_for_request(device_t busdev, device_t dev, struct mmc_request *req)
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
{
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
struct mmc_softc *sc;
struct mmc_ivars *ivar;
int err, i;
enum mmc_retune_req retune_req;
sc = device_get_softc(busdev);
KASSERT(sc->owner != NULL,
("%s: Request from %s without bus being acquired.", __func__,
device_get_nameunit(dev)));
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
/*
* Unless no device is selected or re-tuning is already ongoing,
* execute re-tuning if a) the bridge is requesting to do so and
* re-tuning hasn't been otherwise paused, or b) if a child asked
* to be re-tuned prior to pausing (see also mmc_retune_pause()).
*/
if (__predict_false(sc->last_rca != 0 && sc->retune_ongoing == 0 &&
(((retune_req = mmcbr_get_retune_req(busdev)) != retune_req_none &&
sc->retune_paused == 0) || sc->retune_needed == 1))) {
if (__predict_false(mmc_debug > 1)) {
device_printf(busdev,
"Re-tuning with%s circuit reset required\n",
retune_req == retune_req_reset ? "" : "out");
}
if (device_get_parent(dev) == busdev)
ivar = device_get_ivars(dev);
else {
for (i = 0; i < sc->child_count; i++) {
ivar = device_get_ivars(sc->child_list[i]);
if (ivar->rca == sc->last_rca)
break;
}
if (ivar->rca != sc->last_rca)
return (EINVAL);
}
sc->retune_ongoing = 1;
err = mmc_retune(busdev, dev, retune_req == retune_req_reset);
sc->retune_ongoing = 0;
switch (err) {
case MMC_ERR_NONE:
case MMC_ERR_FAILED: /* Re-tune error but still might work */
break;
case MMC_ERR_BADCRC: /* Switch failure on HS400 recovery */
return (ENXIO);
case MMC_ERR_INVALID: /* Driver implementation b0rken */
default: /* Unknown error, should not happen */
return (EINVAL);
}
sc->retune_needed = 0;
}
Conform to style(9) for return (foo); The files were a mix before. Submitted by: mav@
2008-09-29 18:05:26 +00:00
return (mmc_wait_for_req(sc, req));
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
}
static int
mmc_wait_for_command(struct mmc_softc *sc, uint32_t opcode,
uint32_t arg, uint32_t flags, uint32_t *resp, int retries)
{
struct mmc_command cmd;
int err;
memset(&cmd, 0, sizeof(cmd));
cmd.opcode = opcode;
cmd.arg = arg;
cmd.flags = flags;
Explicitly set data to NULL rather than relying mmc_wait_for_cmd() to do it for us. There may be commands we have to wait for that need to set data. Submitted by: mav@
2008-09-29 18:33:08 +00:00
cmd.data = NULL;
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
err = mmc_wait_for_cmd(sc->dev, sc->dev, &cmd, retries);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
if (err)
return (err);
if (resp) {
if (flags & MMC_RSP_136)
memcpy(resp, cmd.resp, 4 * sizeof(uint32_t));
else
*resp = cmd.resp[0];
}
return (0);
}
static void
mmc_idle_cards(struct mmc_softc *sc)
{
device_t dev;
struct mmc_command cmd;
sdhci/mmc: Minor whitespace cleanups No functional change. Submitted by: Johannes Lundberg <yohanesu75 at gmail.com>
2016-12-20 03:38:14 +00:00
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
dev = sc->dev;
mmcbr_set_chip_select(dev, cs_high);
mmcbr_update_ios(dev);
mmc_ms_delay(1);
memset(&cmd, 0, sizeof(cmd));
cmd.opcode = MMC_GO_IDLE_STATE;
cmd.arg = 0;
cmd.flags = MMC_RSP_NONE | MMC_CMD_BC;
Explicitly set data to NULL rather than relying mmc_wait_for_cmd() to do it for us. There may be commands we have to wait for that need to set data. Submitted by: mav@
2008-09-29 18:33:08 +00:00
cmd.data = NULL;
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
mmc_wait_for_cmd(sc->dev, sc->dev, &cmd, CMD_RETRIES);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
mmc_ms_delay(1);
mmcbr_set_chip_select(dev, cs_dontcare);
mmcbr_update_ios(dev);
mmc_ms_delay(1);
}
static int
mmc_send_app_op_cond(struct mmc_softc *sc, uint32_t ocr, uint32_t *rocr)
{
struct mmc_command cmd;
int err = MMC_ERR_NONE, i;
memset(&cmd, 0, sizeof(cmd));
cmd.opcode = ACMD_SD_SEND_OP_COND;
cmd.arg = ocr;
cmd.flags = MMC_RSP_R3 | MMC_CMD_BCR;
Explicitly set data to NULL rather than relying mmc_wait_for_cmd() to do it for us. There may be commands we have to wait for that need to set data. Submitted by: mav@
2008-09-29 18:33:08 +00:00
cmd.data = NULL;
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
Wait for commands to complete 10 times longer. This makes my A-DATA 32GB SDHC card being detected. Reviewed by: imp MFC after: 2 weeks
2011-01-04 00:08:39 +00:00
for (i = 0; i < 1000; i++) {
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
err = mmc_wait_for_app_cmd(sc->dev, sc->dev, 0, &cmd,
CMD_RETRIES);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
if (err != MMC_ERR_NONE)
break;
Only voltage bits should be zero in send_op_cond argument to ignore busy.
2008-10-08 21:10:55 +00:00
if ((cmd.resp[0] & MMC_OCR_CARD_BUSY) ||
(ocr & MMC_OCR_VOLTAGE) == 0)
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
break;
err = MMC_ERR_TIMEOUT;
mmc_ms_delay(10);
}
if (rocr && err == MMC_ERR_NONE)
*rocr = cmd.resp[0];
Conform to style(9) for return (foo); The files were a mix before. Submitted by: mav@
2008-09-29 18:05:26 +00:00
return (err);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
}
static int
mmc_send_op_cond(struct mmc_softc *sc, uint32_t ocr, uint32_t *rocr)
{
struct mmc_command cmd;
int err = MMC_ERR_NONE, i;
memset(&cmd, 0, sizeof(cmd));
cmd.opcode = MMC_SEND_OP_COND;
cmd.arg = ocr;
cmd.flags = MMC_RSP_R3 | MMC_CMD_BCR;
Explicitly set data to NULL rather than relying mmc_wait_for_cmd() to do it for us. There may be commands we have to wait for that need to set data. Submitted by: mav@
2008-09-29 18:33:08 +00:00
cmd.data = NULL;
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
Wait for commands to complete 10 times longer. This makes my A-DATA 32GB SDHC card being detected. Reviewed by: imp MFC after: 2 weeks
2011-01-04 00:08:39 +00:00
for (i = 0; i < 1000; i++) {
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
err = mmc_wait_for_cmd(sc->dev, sc->dev, &cmd, CMD_RETRIES);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
if (err != MMC_ERR_NONE)
break;
Only voltage bits should be zero in send_op_cond argument to ignore busy.
2008-10-08 21:10:55 +00:00
if ((cmd.resp[0] & MMC_OCR_CARD_BUSY) ||
(ocr & MMC_OCR_VOLTAGE) == 0)
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
break;
err = MMC_ERR_TIMEOUT;
mmc_ms_delay(10);
}
if (rocr && err == MMC_ERR_NONE)
*rocr = cmd.resp[0];
Conform to style(9) for return (foo); The files were a mix before. Submitted by: mav@
2008-09-29 18:05:26 +00:00
return (err);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
}
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
static int
mmc_send_if_cond(struct mmc_softc *sc, uint8_t vhs)
{
struct mmc_command cmd;
int err;
memset(&cmd, 0, sizeof(cmd));
cmd.opcode = SD_SEND_IF_COND;
cmd.arg = (vhs << 8) + 0xAA;
cmd.flags = MMC_RSP_R7 | MMC_CMD_BCR;
cmd.data = NULL;
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
err = mmc_wait_for_cmd(sc->dev, sc->dev, &cmd, CMD_RETRIES);
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
return (err);
}
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
static void
mmc_power_up(struct mmc_softc *sc)
{
device_t dev;
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
enum mmc_vccq vccq;
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
dev = sc->dev;
mmcbr_set_vdd(dev, mmc_highest_voltage(mmcbr_get_host_ocr(dev)));
mmcbr_set_bus_mode(dev, opendrain);
mmcbr_set_chip_select(dev, cs_dontcare);
mmcbr_set_bus_width(dev, bus_width_1);
mmcbr_set_power_mode(dev, power_up);
mmcbr_set_clock(dev, 0);
mmcbr_update_ios(dev);
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
for (vccq = vccq_330; ; vccq--) {
mmcbr_set_vccq(dev, vccq);
if (mmcbr_switch_vccq(dev) == 0 || vccq == vccq_120)
break;
}
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
mmc_ms_delay(1);
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
mmcbr_set_clock(dev, SD_MMC_CARD_ID_FREQUENCY);
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
mmcbr_set_timing(dev, bus_timing_normal);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
mmcbr_set_power_mode(dev, power_on);
mmcbr_update_ios(dev);
mmc_ms_delay(2);
}
Implement power down, and power down the bus on detach. Submitted by: mav@
2008-09-28 23:24:52 +00:00
static void
mmc_power_down(struct mmc_softc *sc)
{
device_t dev = sc->dev;
mmcbr_set_bus_mode(dev, opendrain);
mmcbr_set_chip_select(dev, cs_dontcare);
mmcbr_set_bus_width(dev, bus_width_1);
mmcbr_set_power_mode(dev, power_off);
mmcbr_set_clock(dev, 0);
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
mmcbr_set_timing(dev, bus_timing_normal);
Implement power down, and power down the bus on detach. Submitted by: mav@
2008-09-28 23:24:52 +00:00
mmcbr_update_ios(dev);
}
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
static int
mmc_select_card(struct mmc_softc *sc, uint16_t rca)
{
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
int err, flags;
style(9): spaces around operators.
2008-10-12 07:30:05 +00:00
flags = (rca ? MMC_RSP_R1B : MMC_RSP_NONE) | MMC_CMD_AC;
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
sc->retune_paused++;
err = mmc_wait_for_command(sc, MMC_SELECT_CARD, (uint32_t)rca << 16,
flags, NULL, CMD_RETRIES);
sc->retune_paused--;
return (err);
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
}
static int
o Define some symbols for a few items that are bare constants in the code. o Use NULL in preference to 0 for a few pointers. o default to bus timing normal, like we default to bus_width_1.
2009-02-03 04:28:45 +00:00
mmc_sd_switch(struct mmc_softc *sc, uint8_t mode, uint8_t grp, uint8_t value,
uint8_t *res)
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
{
int err;
struct mmc_command cmd;
struct mmc_data data;
Consistently init all mmc request, command, and data structures to zero before using them.
2013-08-17 00:19:27 +00:00
memset(&cmd, 0, sizeof(cmd));
memset(&data, 0, sizeof(data));
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
memset(res, 0, 64);
o Define some symbols for a few items that are bare constants in the code. o Use NULL in preference to 0 for a few pointers. o default to bus timing normal, like we default to bus_width_1.
2009-02-03 04:28:45 +00:00
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
cmd.opcode = SD_SWITCH_FUNC;
cmd.flags = MMC_RSP_R1 | MMC_CMD_ADTC;
o Define some symbols for a few items that are bare constants in the code. o Use NULL in preference to 0 for a few pointers. o default to bus timing normal, like we default to bus_width_1.
2009-02-03 04:28:45 +00:00
cmd.arg = mode << 31; /* 0 - check, 1 - set */
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
cmd.arg |= 0x00FFFFFF;
Fix bit offset in mmc_sd_switch().
2008-10-08 18:13:14 +00:00
cmd.arg &= ~(0xF << (grp * 4));
cmd.arg |= value << (grp * 4);
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
cmd.data = &data;
data.data = res;
data.len = 64;
data.flags = MMC_DATA_READ;
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
err = mmc_wait_for_cmd(sc->dev, sc->dev, &cmd, CMD_RETRIES);
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
return (err);
}
static int
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
mmc_set_card_bus_width(struct mmc_softc *sc, struct mmc_ivars *ivar,
enum mmc_bus_timing timing)
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
{
Fix minor style nit for file consistency.
2009-01-21 17:34:14 +00:00
struct mmc_command cmd;
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
int err;
Fix minor style nit for file consistency.
2009-01-21 17:34:14 +00:00
uint8_t value;
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
if (mmcbr_get_mode(sc->dev) == mode_sd) {
Consistently init all mmc request, command, and data structures to zero before using them.
2013-08-17 00:19:27 +00:00
memset(&cmd, 0, sizeof(cmd));
o Fixes: - When switching to 4-bit operation, send a SET_CLR_CARD_DETECT command to disconnect the card-detect pull-up resistor from the DAT3 line before sending the SET_BUS_WIDTH command. - Add the missing "reserved" zero entry to the mantissa table used to decode various CSD fields. This was causing SD cards to report that they could run at 30 MHz instead of the maximum 25 MHz mandated in the spec. o Enhancements: - At the MMC layer, format various info from the CID into a string that uniquely identifies the card instance (manufacturer number, serial number, product name and revision, etc). Export it as an instance variable. - At the MMCSD layer, display the formatted card ID string, and also report the clock speed of the hardware (not the card's max speed), and the number of bits and number of blocks per transfer. It comes out like this now: mmcsd0: 968MB <SD SD01G 8.0 SN 276886905 MFG 08/2008 by 3 SD> at mmc0 22.5MHz/4bit/128-block o Use DEVMETHOD_END. o Use NULL instead of 0 for pointers. PR: 156496 Submitted by: Ian Lepore MFC after: 1 week
2012-04-21 01:51:16 +00:00
cmd.opcode = ACMD_SET_CLR_CARD_DETECT;
cmd.flags = MMC_RSP_R1 | MMC_CMD_AC;
cmd.arg = SD_CLR_CARD_DETECT;
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
err = mmc_wait_for_app_cmd(sc->dev, sc->dev, ivar->rca, &cmd,
CMD_RETRIES);
o Fixes: - When switching to 4-bit operation, send a SET_CLR_CARD_DETECT command to disconnect the card-detect pull-up resistor from the DAT3 line before sending the SET_BUS_WIDTH command. - Add the missing "reserved" zero entry to the mantissa table used to decode various CSD fields. This was causing SD cards to report that they could run at 30 MHz instead of the maximum 25 MHz mandated in the spec. o Enhancements: - At the MMC layer, format various info from the CID into a string that uniquely identifies the card instance (manufacturer number, serial number, product name and revision, etc). Export it as an instance variable. - At the MMCSD layer, display the formatted card ID string, and also report the clock speed of the hardware (not the card's max speed), and the number of bits and number of blocks per transfer. It comes out like this now: mmcsd0: 968MB <SD SD01G 8.0 SN 276886905 MFG 08/2008 by 3 SD> at mmc0 22.5MHz/4bit/128-block o Use DEVMETHOD_END. o Use NULL instead of 0 for pointers. PR: 156496 Submitted by: Ian Lepore MFC after: 1 week
2012-04-21 01:51:16 +00:00
if (err != 0)
return (err);
Consistently init all mmc request, command, and data structures to zero before using them.
2013-08-17 00:19:27 +00:00
memset(&cmd, 0, sizeof(cmd));
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
cmd.opcode = ACMD_SET_BUS_WIDTH;
cmd.flags = MMC_RSP_R1 | MMC_CMD_AC;
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
switch (ivar->bus_width) {
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
case bus_width_1:
cmd.arg = SD_BUS_WIDTH_1;
break;
case bus_width_4:
cmd.arg = SD_BUS_WIDTH_4;
break;
default:
return (MMC_ERR_INVALID);
}
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
err = mmc_wait_for_app_cmd(sc->dev, sc->dev, ivar->rca, &cmd,
CMD_RETRIES);
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
} else {
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
switch (ivar->bus_width) {
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
case bus_width_1:
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
if (timing == bus_timing_mmc_hs400 ||
timing == bus_timing_mmc_hs400es)
return (MMC_ERR_INVALID);
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
value = EXT_CSD_BUS_WIDTH_1;
break;
case bus_width_4:
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
switch (timing) {
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
case bus_timing_mmc_ddr52:
value = EXT_CSD_BUS_WIDTH_4_DDR;
break;
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
case bus_timing_mmc_hs400:
case bus_timing_mmc_hs400es:
return (MMC_ERR_INVALID);
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
default:
value = EXT_CSD_BUS_WIDTH_4;
break;
}
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
break;
case bus_width_8:
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
value = 0;
switch (timing) {
case bus_timing_mmc_hs400es:
value = EXT_CSD_BUS_WIDTH_ES;
/* FALLTHROUGH */
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
case bus_timing_mmc_ddr52:
case bus_timing_mmc_hs400:
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
value |= EXT_CSD_BUS_WIDTH_8_DDR;
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
break;
default:
value = EXT_CSD_BUS_WIDTH_8;
break;
}
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
break;
default:
return (MMC_ERR_INVALID);
}
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
err = mmc_switch(sc->dev, sc->dev, ivar->rca,
EXT_CSD_CMD_SET_NORMAL, EXT_CSD_BUS_WIDTH, value,
ivar->cmd6_time, true);
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
}
return (err);
}
static int
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
mmc_set_power_class(struct mmc_softc *sc, struct mmc_ivars *ivar)
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
{
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
device_t dev;
const uint8_t *ext_csd;
uint32_t clock;
uint8_t value;
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
dev = sc->dev;
if (mmcbr_get_mode(dev) != mode_mmc || ivar->csd.spec_vers < 4)
return (MMC_ERR_NONE);
value = 0;
ext_csd = ivar->raw_ext_csd;
clock = mmcbr_get_clock(dev);
switch (1 << mmcbr_get_vdd(dev)) {
case MMC_OCR_LOW_VOLTAGE:
if (clock <= MMC_TYPE_HS_26_MAX)
value = ext_csd[EXT_CSD_PWR_CL_26_195];
else if (clock <= MMC_TYPE_HS_52_MAX) {
if (mmcbr_get_timing(dev) >= bus_timing_mmc_ddr52 &&
ivar->bus_width >= bus_width_4)
value = ext_csd[EXT_CSD_PWR_CL_52_195_DDR];
else
value = ext_csd[EXT_CSD_PWR_CL_52_195];
} else if (clock <= MMC_TYPE_HS200_HS400ES_MAX)
value = ext_csd[EXT_CSD_PWR_CL_200_195];
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
break;
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
case MMC_OCR_270_280:
case MMC_OCR_280_290:
case MMC_OCR_290_300:
case MMC_OCR_300_310:
case MMC_OCR_310_320:
case MMC_OCR_320_330:
case MMC_OCR_330_340:
case MMC_OCR_340_350:
case MMC_OCR_350_360:
if (clock <= MMC_TYPE_HS_26_MAX)
value = ext_csd[EXT_CSD_PWR_CL_26_360];
else if (clock <= MMC_TYPE_HS_52_MAX) {
if (mmcbr_get_timing(dev) == bus_timing_mmc_ddr52 &&
ivar->bus_width >= bus_width_4)
value = ext_csd[EXT_CSD_PWR_CL_52_360_DDR];
else
value = ext_csd[EXT_CSD_PWR_CL_52_360];
} else if (clock <= MMC_TYPE_HS200_HS400ES_MAX) {
if (ivar->bus_width == bus_width_8)
value = ext_csd[EXT_CSD_PWR_CL_200_360_DDR];
else
value = ext_csd[EXT_CSD_PWR_CL_200_360];
}
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
break;
default:
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
device_printf(dev, "No power class support for VDD 0x%x\n",
1 << mmcbr_get_vdd(dev));
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
return (MMC_ERR_INVALID);
}
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
if (ivar->bus_width == bus_width_8)
value = (value & EXT_CSD_POWER_CLASS_8BIT_MASK) >>
EXT_CSD_POWER_CLASS_8BIT_SHIFT;
else
value = (value & EXT_CSD_POWER_CLASS_4BIT_MASK) >>
EXT_CSD_POWER_CLASS_4BIT_SHIFT;
if (value == 0)
return (MMC_ERR_NONE);
return (mmc_switch(dev, dev, ivar->rca, EXT_CSD_CMD_SET_NORMAL,
EXT_CSD_POWER_CLASS, value, ivar->cmd6_time, true));
}
static int
mmc_set_timing(struct mmc_softc *sc, struct mmc_ivars *ivar,
enum mmc_bus_timing timing)
{
u_char switch_res[64];
uint8_t value;
int err;
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
if (mmcbr_get_mode(sc->dev) == mode_sd) {
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
switch (timing) {
case bus_timing_normal:
value = SD_SWITCH_NORMAL_MODE;
break;
case bus_timing_hs:
value = SD_SWITCH_HS_MODE;
break;
default:
return (MMC_ERR_INVALID);
}
o Define some symbols for a few items that are bare constants in the code. o Use NULL in preference to 0 for a few pointers. o default to bus timing normal, like we default to bus_width_1.
2009-02-03 04:28:45 +00:00
err = mmc_sd_switch(sc, SD_SWITCH_MODE_SET, SD_SWITCH_GROUP1,
value, switch_res);
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
if (err != MMC_ERR_NONE)
return (err);
if ((switch_res[16] & 0xf) != value)
return (MMC_ERR_FAILED);
mmcbr_set_timing(sc->dev, timing);
mmcbr_update_ios(sc->dev);
} else {
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
switch (timing) {
case bus_timing_normal:
value = EXT_CSD_HS_TIMING_BC;
break;
case bus_timing_hs:
case bus_timing_mmc_ddr52:
value = EXT_CSD_HS_TIMING_HS;
break;
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
case bus_timing_mmc_hs200:
value = EXT_CSD_HS_TIMING_HS200;
break;
case bus_timing_mmc_hs400:
case bus_timing_mmc_hs400es:
value = EXT_CSD_HS_TIMING_HS400;
break;
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
default:
return (MMC_ERR_INVALID);
}
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
err = mmc_switch(sc->dev, sc->dev, ivar->rca,
EXT_CSD_CMD_SET_NORMAL, EXT_CSD_HS_TIMING, value,
ivar->cmd6_time, false);
if (err != MMC_ERR_NONE)
return (err);
mmcbr_set_timing(sc->dev, timing);
mmcbr_update_ios(sc->dev);
err = mmc_switch_status(sc->dev, sc->dev, ivar->rca,
ivar->cmd6_time);
}
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
return (err);
}
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
static int
mmc_set_vccq(struct mmc_softc *sc, struct mmc_ivars *ivar,
enum mmc_bus_timing timing)
{
if (isset(&ivar->vccq_120, timing))
mmcbr_set_vccq(sc->dev, vccq_120);
else if (isset(&ivar->vccq_180, timing))
mmcbr_set_vccq(sc->dev, vccq_180);
else
mmcbr_set_vccq(sc->dev, vccq_330);
if (mmcbr_switch_vccq(sc->dev) != 0)
return (MMC_ERR_INVALID);
else
return (MMC_ERR_NONE);
}
o Another round fixes for mmc(4), mmcsd(4) and sdhci(4) regarding comments, marking unused parameters as such, style(9), whitespace, etc. o In the mmc(4) bridges and sdhci(4) (bus) front-ends: - Remove redundant assignments of the default bus_generic_print_child device method (I've whipped these out of the tree as part of r227843 once, but they keep coming back ...), - use DEVMETHOD_END, - use NULL instead of 0 for pointers. o Trim/adjust includes.
2017-03-06 23:47:59 +00:00
static const uint8_t p8[8] = {
0x55, 0xAA, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00
};
static const uint8_t p8ok[8] = {
0xAA, 0x55, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00
};
static const uint8_t p4[4] = {
0x5A, 0x00, 0x00, 0x00
};
static const uint8_t p4ok[4] = {
0xA5, 0x00, 0x00, 0x00
};
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
static int
mmc_test_bus_width(struct mmc_softc *sc)
{
struct mmc_command cmd;
struct mmc_data data;
uint8_t buf[8];
o Another round fixes for mmc(4), mmcsd(4) and sdhci(4) regarding comments, marking unused parameters as such, style(9), whitespace, etc. o In the mmc(4) bridges and sdhci(4) (bus) front-ends: - Remove redundant assignments of the default bus_generic_print_child device method (I've whipped these out of the tree as part of r227843 once, but they keep coming back ...), - use DEVMETHOD_END, - use NULL instead of 0 for pointers. o Trim/adjust includes.
2017-03-06 23:47:59 +00:00
int err;
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
if (mmcbr_get_caps(sc->dev) & MMC_CAP_8_BIT_DATA) {
mmcbr_set_bus_width(sc->dev, bus_width_8);
mmcbr_update_ios(sc->dev);
Log mmc and sd command failures. Reporting of routine expected errors, such as timeouts while probing a bus or testing for a feature, is squelched. Also, error reporting is limited to 5 events per second, because when an sdcard goes bad on a low-end embedded board, flooding the console at high speed isn't helpful. Original logging code contributed by Michal Meloun, but then I fancied it up with squelching and ppsratecheck.
2014-12-20 04:24:40 +00:00
sc->squelched++; /* Errors are expected, squelch reporting. */
Consistently init all mmc request, command, and data structures to zero before using them.
2013-08-17 00:19:27 +00:00
memset(&cmd, 0, sizeof(cmd));
memset(&data, 0, sizeof(data));
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
cmd.opcode = MMC_BUSTEST_W;
cmd.arg = 0;
cmd.flags = MMC_RSP_R1 | MMC_CMD_ADTC;
cmd.data = &data;
o Another round fixes for mmc(4), mmcsd(4) and sdhci(4) regarding comments, marking unused parameters as such, style(9), whitespace, etc. o In the mmc(4) bridges and sdhci(4) (bus) front-ends: - Remove redundant assignments of the default bus_generic_print_child device method (I've whipped these out of the tree as part of r227843 once, but they keep coming back ...), - use DEVMETHOD_END, - use NULL instead of 0 for pointers. o Trim/adjust includes.
2017-03-06 23:47:59 +00:00
data.data = __DECONST(void *, p8);
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
data.len = 8;
data.flags = MMC_DATA_WRITE;
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
mmc_wait_for_cmd(sc->dev, sc->dev, &cmd, 0);
sdhci/mmc: Minor whitespace cleanups No functional change. Submitted by: Johannes Lundberg <yohanesu75 at gmail.com>
2016-12-20 03:38:14 +00:00
Consistently init all mmc request, command, and data structures to zero before using them.
2013-08-17 00:19:27 +00:00
memset(&cmd, 0, sizeof(cmd));
memset(&data, 0, sizeof(data));
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
cmd.opcode = MMC_BUSTEST_R;
cmd.arg = 0;
cmd.flags = MMC_RSP_R1 | MMC_CMD_ADTC;
cmd.data = &data;
data.data = buf;
data.len = 8;
data.flags = MMC_DATA_READ;
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
err = mmc_wait_for_cmd(sc->dev, sc->dev, &cmd, 0);
Log mmc and sd command failures. Reporting of routine expected errors, such as timeouts while probing a bus or testing for a feature, is squelched. Also, error reporting is limited to 5 events per second, because when an sdcard goes bad on a low-end embedded board, flooding the console at high speed isn't helpful. Original logging code contributed by Michal Meloun, but then I fancied it up with squelching and ppsratecheck.
2014-12-20 04:24:40 +00:00
sc->squelched--;
sdhci/mmc: Minor whitespace cleanups No functional change. Submitted by: Johannes Lundberg <yohanesu75 at gmail.com>
2016-12-20 03:38:14 +00:00
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
mmcbr_set_bus_width(sc->dev, bus_width_1);
mmcbr_update_ios(sc->dev);
if (err == MMC_ERR_NONE && memcmp(buf, p8ok, 8) == 0)
return (bus_width_8);
}
if (mmcbr_get_caps(sc->dev) & MMC_CAP_4_BIT_DATA) {
mmcbr_set_bus_width(sc->dev, bus_width_4);
mmcbr_update_ios(sc->dev);
Log mmc and sd command failures. Reporting of routine expected errors, such as timeouts while probing a bus or testing for a feature, is squelched. Also, error reporting is limited to 5 events per second, because when an sdcard goes bad on a low-end embedded board, flooding the console at high speed isn't helpful. Original logging code contributed by Michal Meloun, but then I fancied it up with squelching and ppsratecheck.
2014-12-20 04:24:40 +00:00
sc->squelched++; /* Errors are expected, squelch reporting. */
Consistently init all mmc request, command, and data structures to zero before using them.
2013-08-17 00:19:27 +00:00
memset(&cmd, 0, sizeof(cmd));
memset(&data, 0, sizeof(data));
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
cmd.opcode = MMC_BUSTEST_W;
cmd.arg = 0;
cmd.flags = MMC_RSP_R1 | MMC_CMD_ADTC;
cmd.data = &data;
o Another round fixes for mmc(4), mmcsd(4) and sdhci(4) regarding comments, marking unused parameters as such, style(9), whitespace, etc. o In the mmc(4) bridges and sdhci(4) (bus) front-ends: - Remove redundant assignments of the default bus_generic_print_child device method (I've whipped these out of the tree as part of r227843 once, but they keep coming back ...), - use DEVMETHOD_END, - use NULL instead of 0 for pointers. o Trim/adjust includes.
2017-03-06 23:47:59 +00:00
data.data = __DECONST(void *, p4);
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
data.len = 4;
data.flags = MMC_DATA_WRITE;
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
mmc_wait_for_cmd(sc->dev, sc->dev, &cmd, 0);
sdhci/mmc: Minor whitespace cleanups No functional change. Submitted by: Johannes Lundberg <yohanesu75 at gmail.com>
2016-12-20 03:38:14 +00:00
Consistently init all mmc request, command, and data structures to zero before using them.
2013-08-17 00:19:27 +00:00
memset(&cmd, 0, sizeof(cmd));
memset(&data, 0, sizeof(data));
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
cmd.opcode = MMC_BUSTEST_R;
cmd.arg = 0;
cmd.flags = MMC_RSP_R1 | MMC_CMD_ADTC;
cmd.data = &data;
data.data = buf;
data.len = 4;
data.flags = MMC_DATA_READ;
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
err = mmc_wait_for_cmd(sc->dev, sc->dev, &cmd, 0);
Log mmc and sd command failures. Reporting of routine expected errors, such as timeouts while probing a bus or testing for a feature, is squelched. Also, error reporting is limited to 5 events per second, because when an sdcard goes bad on a low-end embedded board, flooding the console at high speed isn't helpful. Original logging code contributed by Michal Meloun, but then I fancied it up with squelching and ppsratecheck.
2014-12-20 04:24:40 +00:00
sc->squelched--;
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
mmcbr_set_bus_width(sc->dev, bus_width_1);
mmcbr_update_ios(sc->dev);
if (err == MMC_ERR_NONE && memcmp(buf, p4ok, 4) == 0)
return (bus_width_4);
}
return (bus_width_1);
}
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
static uint32_t
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
mmc_get_bits(uint32_t *bits, int bit_len, int start, int size)
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
{
o Use seprate routines to decode cid and csd for sd and mmc cards. All they have in common right now is a memset. This saves a parameter to these routines, as well as a level of indentation. o Make mmc_get_bits a little clearer... It really only works on 128-bit registers right now.
2008-10-09 19:47:28 +00:00
const int i = (bit_len / 32) - (start / 32) - 1;
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
const int shift = start & 31;
uint32_t retval = bits[i] >> shift;
Again, fixes regarding style(4), to comments, includes and unused parameters.
2017-03-17 22:57:37 +00:00
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
if (size + shift > 32)
retval |= bits[i - 1] << (32 - shift);
Fix 32bit bit fields handling. This fixes card serial number fetching. It was just a cosmetic issue, because that number is only reported in logs. Reported by: Michael Butler on current@
2011-01-17 19:31:34 +00:00
return (retval & ((1llu << size) - 1));
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
}
static void
o Use seprate routines to decode cid and csd for sd and mmc cards. All they have in common right now is a memset. This saves a parameter to these routines, as well as a level of indentation. o Make mmc_get_bits a little clearer... It really only works on 128-bit registers right now.
2008-10-09 19:47:28 +00:00
mmc_decode_cid_sd(uint32_t *raw_cid, struct mmc_cid *cid)
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
{
int i;
o Use seprate routines to decode cid and csd for sd and mmc cards. All they have in common right now is a memset. This saves a parameter to these routines, as well as a level of indentation. o Make mmc_get_bits a little clearer... It really only works on 128-bit registers right now.
2008-10-09 19:47:28 +00:00
/* There's no version info, so we take it on faith */
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
memset(cid, 0, sizeof(*cid));
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
cid->mid = mmc_get_bits(raw_cid, 128, 120, 8);
cid->oid = mmc_get_bits(raw_cid, 128, 104, 16);
o Use seprate routines to decode cid and csd for sd and mmc cards. All they have in common right now is a memset. This saves a parameter to these routines, as well as a level of indentation. o Make mmc_get_bits a little clearer... It really only works on 128-bit registers right now.
2008-10-09 19:47:28 +00:00
for (i = 0; i < 5; i++)
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
cid->pnm[i] = mmc_get_bits(raw_cid, 128, 96 - i * 8, 8);
Improve mmc driver verbose logging. Make requests logging controllable by hw.mmc.debug sysctl.
2009-01-28 22:09:00 +00:00
cid->pnm[5] = 0;
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
cid->prv = mmc_get_bits(raw_cid, 128, 56, 8);
cid->psn = mmc_get_bits(raw_cid, 128, 24, 32);
Improve mmc driver verbose logging. Make requests logging controllable by hw.mmc.debug sysctl.
2009-01-28 22:09:00 +00:00
cid->mdt_year = mmc_get_bits(raw_cid, 128, 12, 8) + 2000;
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
cid->mdt_month = mmc_get_bits(raw_cid, 128, 8, 4);
o Use seprate routines to decode cid and csd for sd and mmc cards. All they have in common right now is a memset. This saves a parameter to these routines, as well as a level of indentation. o Make mmc_get_bits a little clearer... It really only works on 128-bit registers right now.
2008-10-09 19:47:28 +00:00
}
static void
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
mmc_decode_cid_mmc(uint32_t *raw_cid, struct mmc_cid *cid, bool is_4_41p)
o Use seprate routines to decode cid and csd for sd and mmc cards. All they have in common right now is a memset. This saves a parameter to these routines, as well as a level of indentation. o Make mmc_get_bits a little clearer... It really only works on 128-bit registers right now.
2008-10-09 19:47:28 +00:00
{
int i;
/* There's no version info, so we take it on faith */
memset(cid, 0, sizeof(*cid));
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
cid->mid = mmc_get_bits(raw_cid, 128, 120, 8);
cid->oid = mmc_get_bits(raw_cid, 128, 104, 8);
o Use seprate routines to decode cid and csd for sd and mmc cards. All they have in common right now is a memset. This saves a parameter to these routines, as well as a level of indentation. o Make mmc_get_bits a little clearer... It really only works on 128-bit registers right now.
2008-10-09 19:47:28 +00:00
for (i = 0; i < 6; i++)
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
cid->pnm[i] = mmc_get_bits(raw_cid, 128, 96 - i * 8, 8);
Improve mmc driver verbose logging. Make requests logging controllable by hw.mmc.debug sysctl.
2009-01-28 22:09:00 +00:00
cid->pnm[6] = 0;
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
cid->prv = mmc_get_bits(raw_cid, 128, 48, 8);
cid->psn = mmc_get_bits(raw_cid, 128, 16, 32);
cid->mdt_month = mmc_get_bits(raw_cid, 128, 12, 4);
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
cid->mdt_year = mmc_get_bits(raw_cid, 128, 8, 4);
if (is_4_41p)
cid->mdt_year += 2013;
else
cid->mdt_year += 1997;
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
}
o Fixes: - When switching to 4-bit operation, send a SET_CLR_CARD_DETECT command to disconnect the card-detect pull-up resistor from the DAT3 line before sending the SET_BUS_WIDTH command. - Add the missing "reserved" zero entry to the mantissa table used to decode various CSD fields. This was causing SD cards to report that they could run at 30 MHz instead of the maximum 25 MHz mandated in the spec. o Enhancements: - At the MMC layer, format various info from the CID into a string that uniquely identifies the card instance (manufacturer number, serial number, product name and revision, etc). Export it as an instance variable. - At the MMCSD layer, display the formatted card ID string, and also report the clock speed of the hardware (not the card's max speed), and the number of bits and number of blocks per transfer. It comes out like this now: mmcsd0: 968MB <SD SD01G 8.0 SN 276886905 MFG 08/2008 by 3 SD> at mmc0 22.5MHz/4bit/128-block o Use DEVMETHOD_END. o Use NULL instead of 0 for pointers. PR: 156496 Submitted by: Ian Lepore MFC after: 1 week
2012-04-21 01:51:16 +00:00
static void
mmc_format_card_id_string(struct mmc_ivars *ivar)
{
char oidstr[8];
uint8_t c1;
uint8_t c2;
/*
* Format a card ID string for use by the mmcsd driver, it's what
* appears between the <> in the following:
- If available, use TRIM instead of ERASE for implementing BIO_DELETE. This also involves adding a quirk table as TRIM is broken for some Kingston eMMC devices, though. Compared to ERASE (declared "legacy" in the eMMC specification v5.1), TRIM has the advantage of operating on write sectors rather than on erase sectors, which typically are of a much larger size. Thus, employing TRIM, we don't need to fiddle with coalescing BIO_DELETE requests that are also of (write) sector units into erase sectors, which might not even add up in all cases. - For some SanDisk iNAND devices, the CMD38 argument, e. g. ERASE, TRIM etc., has to be specified via EXT_CSD[113], which now is also handled via a quirk. - My initial understanding was that for eMMC partitions, the granularity should be used as erase sector size, e. g. 128 KB for boot partitions. However, rereading the relevant parts of the eMMC specification v5.1, this isn't actually correct. So drop the code which used partition granularities for delmaxsize and stripesize. For the most part, this change is a NOP, though, because a) for ERASE, mmcsd_delete() used the erase sector size unconditionally for all partitions anyway and b) g_disk_limit() doesn't actually take the stripesize into account. - Take some more advantage of mmcsd_errmsg() in mmcsd(4) for making error codes human readable.
2017-08-07 23:33:05 +00:00
* mmcsd0: 968MB <SD SD01G 8.0 SN 2686905 MFG 08/2008 by 3 TN> at mmc0
o Fixes: - When switching to 4-bit operation, send a SET_CLR_CARD_DETECT command to disconnect the card-detect pull-up resistor from the DAT3 line before sending the SET_BUS_WIDTH command. - Add the missing "reserved" zero entry to the mantissa table used to decode various CSD fields. This was causing SD cards to report that they could run at 30 MHz instead of the maximum 25 MHz mandated in the spec. o Enhancements: - At the MMC layer, format various info from the CID into a string that uniquely identifies the card instance (manufacturer number, serial number, product name and revision, etc). Export it as an instance variable. - At the MMCSD layer, display the formatted card ID string, and also report the clock speed of the hardware (not the card's max speed), and the number of bits and number of blocks per transfer. It comes out like this now: mmcsd0: 968MB <SD SD01G 8.0 SN 276886905 MFG 08/2008 by 3 SD> at mmc0 22.5MHz/4bit/128-block o Use DEVMETHOD_END. o Use NULL instead of 0 for pointers. PR: 156496 Submitted by: Ian Lepore MFC after: 1 week
2012-04-21 01:51:16 +00:00
* 22.5MHz/4bit/128-block
*
Export an mmc or sd card's serial number from the mmc layer as an ivar. In the mmcsd layer use this value to populate disk->d_ident. Also set disk->d_descr to the full set of card identification info (includes vendor, model, manufacturing date, etc).
2014-07-31 16:54:54 +00:00
* Also format just the card serial number, which the mmcsd driver will
* use as the disk->d_ident string.
*
o Fixes: - When switching to 4-bit operation, send a SET_CLR_CARD_DETECT command to disconnect the card-detect pull-up resistor from the DAT3 line before sending the SET_BUS_WIDTH command. - Add the missing "reserved" zero entry to the mantissa table used to decode various CSD fields. This was causing SD cards to report that they could run at 30 MHz instead of the maximum 25 MHz mandated in the spec. o Enhancements: - At the MMC layer, format various info from the CID into a string that uniquely identifies the card instance (manufacturer number, serial number, product name and revision, etc). Export it as an instance variable. - At the MMCSD layer, display the formatted card ID string, and also report the clock speed of the hardware (not the card's max speed), and the number of bits and number of blocks per transfer. It comes out like this now: mmcsd0: 968MB <SD SD01G 8.0 SN 276886905 MFG 08/2008 by 3 SD> at mmc0 22.5MHz/4bit/128-block o Use DEVMETHOD_END. o Use NULL instead of 0 for pointers. PR: 156496 Submitted by: Ian Lepore MFC after: 1 week
2012-04-21 01:51:16 +00:00
* The card_id_string in mmc_ivars is currently allocated as 64 bytes,
* and our max formatted length is currently 55 bytes if every field
* contains the largest value.
*
* Sometimes the oid is two printable ascii chars; when it's not,
* format it as 0xnnnn instead.
*/
c1 = (ivar->cid.oid >> 8) & 0x0ff;
c2 = ivar->cid.oid & 0x0ff;
if (c1 > 0x1f && c1 < 0x7f && c2 > 0x1f && c2 < 0x7f)
snprintf(oidstr, sizeof(oidstr), "%c%c", c1, c2);
else
snprintf(oidstr, sizeof(oidstr), "0x%04x", ivar->cid.oid);
Export an mmc or sd card's serial number from the mmc layer as an ivar. In the mmcsd layer use this value to populate disk->d_ident. Also set disk->d_descr to the full set of card identification info (includes vendor, model, manufacturing date, etc).
2014-07-31 16:54:54 +00:00
snprintf(ivar->card_sn_string, sizeof(ivar->card_sn_string),
"%08X", ivar->cid.psn);
o Fixes: - When switching to 4-bit operation, send a SET_CLR_CARD_DETECT command to disconnect the card-detect pull-up resistor from the DAT3 line before sending the SET_BUS_WIDTH command. - Add the missing "reserved" zero entry to the mantissa table used to decode various CSD fields. This was causing SD cards to report that they could run at 30 MHz instead of the maximum 25 MHz mandated in the spec. o Enhancements: - At the MMC layer, format various info from the CID into a string that uniquely identifies the card instance (manufacturer number, serial number, product name and revision, etc). Export it as an instance variable. - At the MMCSD layer, display the formatted card ID string, and also report the clock speed of the hardware (not the card's max speed), and the number of bits and number of blocks per transfer. It comes out like this now: mmcsd0: 968MB <SD SD01G 8.0 SN 276886905 MFG 08/2008 by 3 SD> at mmc0 22.5MHz/4bit/128-block o Use DEVMETHOD_END. o Use NULL instead of 0 for pointers. PR: 156496 Submitted by: Ian Lepore MFC after: 1 week
2012-04-21 01:51:16 +00:00
snprintf(ivar->card_id_string, sizeof(ivar->card_id_string),
Export an mmc or sd card's serial number from the mmc layer as an ivar. In the mmcsd layer use this value to populate disk->d_ident. Also set disk->d_descr to the full set of card identification info (includes vendor, model, manufacturing date, etc).
2014-07-31 16:54:54 +00:00
"%s%s %s %d.%d SN %08X MFG %02d/%04d by %d %s",
o Fixes: - When switching to 4-bit operation, send a SET_CLR_CARD_DETECT command to disconnect the card-detect pull-up resistor from the DAT3 line before sending the SET_BUS_WIDTH command. - Add the missing "reserved" zero entry to the mantissa table used to decode various CSD fields. This was causing SD cards to report that they could run at 30 MHz instead of the maximum 25 MHz mandated in the spec. o Enhancements: - At the MMC layer, format various info from the CID into a string that uniquely identifies the card instance (manufacturer number, serial number, product name and revision, etc). Export it as an instance variable. - At the MMCSD layer, display the formatted card ID string, and also report the clock speed of the hardware (not the card's max speed), and the number of bits and number of blocks per transfer. It comes out like this now: mmcsd0: 968MB <SD SD01G 8.0 SN 276886905 MFG 08/2008 by 3 SD> at mmc0 22.5MHz/4bit/128-block o Use DEVMETHOD_END. o Use NULL instead of 0 for pointers. PR: 156496 Submitted by: Ian Lepore MFC after: 1 week
2012-04-21 01:51:16 +00:00
ivar->mode == mode_sd ? "SD" : "MMC", ivar->high_cap ? "HC" : "",
ivar->cid.pnm, ivar->cid.prv >> 4, ivar->cid.prv & 0x0f,
ivar->cid.psn, ivar->cid.mdt_month, ivar->cid.mdt_year,
ivar->cid.mid, oidstr);
}
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
static const int exp[8] = {
1, 10, 100, 1000, 10000, 100000, 1000000, 10000000
};
o Fixes: - When switching to 4-bit operation, send a SET_CLR_CARD_DETECT command to disconnect the card-detect pull-up resistor from the DAT3 line before sending the SET_BUS_WIDTH command. - Add the missing "reserved" zero entry to the mantissa table used to decode various CSD fields. This was causing SD cards to report that they could run at 30 MHz instead of the maximum 25 MHz mandated in the spec. o Enhancements: - At the MMC layer, format various info from the CID into a string that uniquely identifies the card instance (manufacturer number, serial number, product name and revision, etc). Export it as an instance variable. - At the MMCSD layer, display the formatted card ID string, and also report the clock speed of the hardware (not the card's max speed), and the number of bits and number of blocks per transfer. It comes out like this now: mmcsd0: 968MB <SD SD01G 8.0 SN 276886905 MFG 08/2008 by 3 SD> at mmc0 22.5MHz/4bit/128-block o Use DEVMETHOD_END. o Use NULL instead of 0 for pointers. PR: 156496 Submitted by: Ian Lepore MFC after: 1 week
2012-04-21 01:51:16 +00:00
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
static const int mant[16] = {
o Fixes: - When switching to 4-bit operation, send a SET_CLR_CARD_DETECT command to disconnect the card-detect pull-up resistor from the DAT3 line before sending the SET_BUS_WIDTH command. - Add the missing "reserved" zero entry to the mantissa table used to decode various CSD fields. This was causing SD cards to report that they could run at 30 MHz instead of the maximum 25 MHz mandated in the spec. o Enhancements: - At the MMC layer, format various info from the CID into a string that uniquely identifies the card instance (manufacturer number, serial number, product name and revision, etc). Export it as an instance variable. - At the MMCSD layer, display the formatted card ID string, and also report the clock speed of the hardware (not the card's max speed), and the number of bits and number of blocks per transfer. It comes out like this now: mmcsd0: 968MB <SD SD01G 8.0 SN 276886905 MFG 08/2008 by 3 SD> at mmc0 22.5MHz/4bit/128-block o Use DEVMETHOD_END. o Use NULL instead of 0 for pointers. PR: 156496 Submitted by: Ian Lepore MFC after: 1 week
2012-04-21 01:51:16 +00:00
0, 10, 12, 13, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 70, 80
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
};
o Fixes: - When switching to 4-bit operation, send a SET_CLR_CARD_DETECT command to disconnect the card-detect pull-up resistor from the DAT3 line before sending the SET_BUS_WIDTH command. - Add the missing "reserved" zero entry to the mantissa table used to decode various CSD fields. This was causing SD cards to report that they could run at 30 MHz instead of the maximum 25 MHz mandated in the spec. o Enhancements: - At the MMC layer, format various info from the CID into a string that uniquely identifies the card instance (manufacturer number, serial number, product name and revision, etc). Export it as an instance variable. - At the MMCSD layer, display the formatted card ID string, and also report the clock speed of the hardware (not the card's max speed), and the number of bits and number of blocks per transfer. It comes out like this now: mmcsd0: 968MB <SD SD01G 8.0 SN 276886905 MFG 08/2008 by 3 SD> at mmc0 22.5MHz/4bit/128-block o Use DEVMETHOD_END. o Use NULL instead of 0 for pointers. PR: 156496 Submitted by: Ian Lepore MFC after: 1 week
2012-04-21 01:51:16 +00:00
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
static const int cur_min[8] = {
500, 1000, 5000, 10000, 25000, 35000, 60000, 100000
};
o Fixes: - When switching to 4-bit operation, send a SET_CLR_CARD_DETECT command to disconnect the card-detect pull-up resistor from the DAT3 line before sending the SET_BUS_WIDTH command. - Add the missing "reserved" zero entry to the mantissa table used to decode various CSD fields. This was causing SD cards to report that they could run at 30 MHz instead of the maximum 25 MHz mandated in the spec. o Enhancements: - At the MMC layer, format various info from the CID into a string that uniquely identifies the card instance (manufacturer number, serial number, product name and revision, etc). Export it as an instance variable. - At the MMCSD layer, display the formatted card ID string, and also report the clock speed of the hardware (not the card's max speed), and the number of bits and number of blocks per transfer. It comes out like this now: mmcsd0: 968MB <SD SD01G 8.0 SN 276886905 MFG 08/2008 by 3 SD> at mmc0 22.5MHz/4bit/128-block o Use DEVMETHOD_END. o Use NULL instead of 0 for pointers. PR: 156496 Submitted by: Ian Lepore MFC after: 1 week
2012-04-21 01:51:16 +00:00
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
static const int cur_max[8] = {
1000, 5000, 10000, 25000, 35000, 45000, 800000, 200000
};
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
static int
o Use seprate routines to decode cid and csd for sd and mmc cards. All they have in common right now is a memset. This saves a parameter to these routines, as well as a level of indentation. o Make mmc_get_bits a little clearer... It really only works on 128-bit registers right now.
2008-10-09 19:47:28 +00:00
mmc_decode_csd_sd(uint32_t *raw_csd, struct mmc_csd *csd)
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
{
int v;
int m;
int e;
memset(csd, 0, sizeof(*csd));
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
csd->csd_structure = v = mmc_get_bits(raw_csd, 128, 126, 2);
o Use seprate routines to decode cid and csd for sd and mmc cards. All they have in common right now is a memset. This saves a parameter to these routines, as well as a level of indentation. o Make mmc_get_bits a little clearer... It really only works on 128-bit registers right now.
2008-10-09 19:47:28 +00:00
if (v == 0) {
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
m = mmc_get_bits(raw_csd, 128, 115, 4);
e = mmc_get_bits(raw_csd, 128, 112, 3);
- Fix some typos in mmc_acquire_bus() and mmc_send_csd(). - Fix some math errors in mmc_decode_csd_sd(). - Fix incorrect arguments to mmc_send_app_op_cond() in mmc_go_discovery(). - Add reporting of CSD for debug purposes. - Add detection (and skipping) of password-locked cards. - Add setting of block length on card if necessary. Submitted by: Patrick Kelsey MFC after: 3 days
2012-05-27 16:08:58 +00:00
csd->tacc = (exp[e] * mant[m] + 9) / 10;
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
csd->nsac = mmc_get_bits(raw_csd, 128, 104, 8) * 100;
m = mmc_get_bits(raw_csd, 128, 99, 4);
e = mmc_get_bits(raw_csd, 128, 96, 3);
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
csd->tran_speed = exp[e] * 10000 * mant[m];
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
csd->ccc = mmc_get_bits(raw_csd, 128, 84, 12);
csd->read_bl_len = 1 << mmc_get_bits(raw_csd, 128, 80, 4);
csd->read_bl_partial = mmc_get_bits(raw_csd, 128, 79, 1);
csd->write_blk_misalign = mmc_get_bits(raw_csd, 128, 78, 1);
csd->read_blk_misalign = mmc_get_bits(raw_csd, 128, 77, 1);
csd->dsr_imp = mmc_get_bits(raw_csd, 128, 76, 1);
Fix overly long lines, whitespace and other bugs according to style(9).
2017-01-29 00:05:49 +00:00
csd->vdd_r_curr_min =
cur_min[mmc_get_bits(raw_csd, 128, 59, 3)];
csd->vdd_r_curr_max =
cur_max[mmc_get_bits(raw_csd, 128, 56, 3)];
csd->vdd_w_curr_min =
cur_min[mmc_get_bits(raw_csd, 128, 53, 3)];
csd->vdd_w_curr_max =
cur_max[mmc_get_bits(raw_csd, 128, 50, 3)];
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
m = mmc_get_bits(raw_csd, 128, 62, 12);
e = mmc_get_bits(raw_csd, 128, 47, 3);
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
csd->capacity = ((1 + m) << (e + 2)) * csd->read_bl_len;
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
csd->erase_blk_en = mmc_get_bits(raw_csd, 128, 46, 1);
csd->erase_sector = mmc_get_bits(raw_csd, 128, 39, 7) + 1;
csd->wp_grp_size = mmc_get_bits(raw_csd, 128, 32, 7);
csd->wp_grp_enable = mmc_get_bits(raw_csd, 128, 31, 1);
csd->r2w_factor = 1 << mmc_get_bits(raw_csd, 128, 26, 3);
csd->write_bl_len = 1 << mmc_get_bits(raw_csd, 128, 22, 4);
csd->write_bl_partial = mmc_get_bits(raw_csd, 128, 21, 1);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
return (MMC_ERR_NONE);
o Use seprate routines to decode cid and csd for sd and mmc cards. All they have in common right now is a memset. This saves a parameter to these routines, as well as a level of indentation. o Make mmc_get_bits a little clearer... It really only works on 128-bit registers right now.
2008-10-09 19:47:28 +00:00
} else if (v == 1) {
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
m = mmc_get_bits(raw_csd, 128, 115, 4);
e = mmc_get_bits(raw_csd, 128, 112, 3);
- Fix some typos in mmc_acquire_bus() and mmc_send_csd(). - Fix some math errors in mmc_decode_csd_sd(). - Fix incorrect arguments to mmc_send_app_op_cond() in mmc_go_discovery(). - Add reporting of CSD for debug purposes. - Add detection (and skipping) of password-locked cards. - Add setting of block length on card if necessary. Submitted by: Patrick Kelsey MFC after: 3 days
2012-05-27 16:08:58 +00:00
csd->tacc = (exp[e] * mant[m] + 9) / 10;
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
csd->nsac = mmc_get_bits(raw_csd, 128, 104, 8) * 100;
m = mmc_get_bits(raw_csd, 128, 99, 4);
e = mmc_get_bits(raw_csd, 128, 96, 3);
o Use seprate routines to decode cid and csd for sd and mmc cards. All they have in common right now is a memset. This saves a parameter to these routines, as well as a level of indentation. o Make mmc_get_bits a little clearer... It really only works on 128-bit registers right now.
2008-10-09 19:47:28 +00:00
csd->tran_speed = exp[e] * 10000 * mant[m];
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
csd->ccc = mmc_get_bits(raw_csd, 128, 84, 12);
csd->read_bl_len = 1 << mmc_get_bits(raw_csd, 128, 80, 4);
csd->read_bl_partial = mmc_get_bits(raw_csd, 128, 79, 1);
csd->write_blk_misalign = mmc_get_bits(raw_csd, 128, 78, 1);
csd->read_blk_misalign = mmc_get_bits(raw_csd, 128, 77, 1);
csd->dsr_imp = mmc_get_bits(raw_csd, 128, 76, 1);
Fix overly long lines, whitespace and other bugs according to style(9).
2017-01-29 00:05:49 +00:00
csd->capacity = ((uint64_t)mmc_get_bits(raw_csd, 128, 48, 22) +
1) * 512 * 1024;
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
csd->erase_blk_en = mmc_get_bits(raw_csd, 128, 46, 1);
csd->erase_sector = mmc_get_bits(raw_csd, 128, 39, 7) + 1;
csd->wp_grp_size = mmc_get_bits(raw_csd, 128, 32, 7);
csd->wp_grp_enable = mmc_get_bits(raw_csd, 128, 31, 1);
csd->r2w_factor = 1 << mmc_get_bits(raw_csd, 128, 26, 3);
csd->write_bl_len = 1 << mmc_get_bits(raw_csd, 128, 22, 4);
csd->write_bl_partial = mmc_get_bits(raw_csd, 128, 21, 1);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
return (MMC_ERR_NONE);
}
return (MMC_ERR_INVALID);
o Use seprate routines to decode cid and csd for sd and mmc cards. All they have in common right now is a memset. This saves a parameter to these routines, as well as a level of indentation. o Make mmc_get_bits a little clearer... It really only works on 128-bit registers right now.
2008-10-09 19:47:28 +00:00
}
static void
mmc_decode_csd_mmc(uint32_t *raw_csd, struct mmc_csd *csd)
{
int m;
int e;
memset(csd, 0, sizeof(*csd));
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
csd->csd_structure = mmc_get_bits(raw_csd, 128, 126, 2);
csd->spec_vers = mmc_get_bits(raw_csd, 128, 122, 4);
m = mmc_get_bits(raw_csd, 128, 115, 4);
e = mmc_get_bits(raw_csd, 128, 112, 3);
o Use seprate routines to decode cid and csd for sd and mmc cards. All they have in common right now is a memset. This saves a parameter to these routines, as well as a level of indentation. o Make mmc_get_bits a little clearer... It really only works on 128-bit registers right now.
2008-10-09 19:47:28 +00:00
csd->tacc = exp[e] * mant[m] + 9 / 10;
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
csd->nsac = mmc_get_bits(raw_csd, 128, 104, 8) * 100;
m = mmc_get_bits(raw_csd, 128, 99, 4);
e = mmc_get_bits(raw_csd, 128, 96, 3);
o Use seprate routines to decode cid and csd for sd and mmc cards. All they have in common right now is a memset. This saves a parameter to these routines, as well as a level of indentation. o Make mmc_get_bits a little clearer... It really only works on 128-bit registers right now.
2008-10-09 19:47:28 +00:00
csd->tran_speed = exp[e] * 10000 * mant[m];
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
csd->ccc = mmc_get_bits(raw_csd, 128, 84, 12);
csd->read_bl_len = 1 << mmc_get_bits(raw_csd, 128, 80, 4);
csd->read_bl_partial = mmc_get_bits(raw_csd, 128, 79, 1);
csd->write_blk_misalign = mmc_get_bits(raw_csd, 128, 78, 1);
csd->read_blk_misalign = mmc_get_bits(raw_csd, 128, 77, 1);
csd->dsr_imp = mmc_get_bits(raw_csd, 128, 76, 1);
csd->vdd_r_curr_min = cur_min[mmc_get_bits(raw_csd, 128, 59, 3)];
csd->vdd_r_curr_max = cur_max[mmc_get_bits(raw_csd, 128, 56, 3)];
csd->vdd_w_curr_min = cur_min[mmc_get_bits(raw_csd, 128, 53, 3)];
csd->vdd_w_curr_max = cur_max[mmc_get_bits(raw_csd, 128, 50, 3)];
m = mmc_get_bits(raw_csd, 128, 62, 12);
e = mmc_get_bits(raw_csd, 128, 47, 3);
o Use seprate routines to decode cid and csd for sd and mmc cards. All they have in common right now is a memset. This saves a parameter to these routines, as well as a level of indentation. o Make mmc_get_bits a little clearer... It really only works on 128-bit registers right now.
2008-10-09 19:47:28 +00:00
csd->capacity = ((1 + m) << (e + 2)) * csd->read_bl_len;
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
csd->erase_blk_en = 0;
csd->erase_sector = (mmc_get_bits(raw_csd, 128, 42, 5) + 1) *
(mmc_get_bits(raw_csd, 128, 37, 5) + 1);
csd->wp_grp_size = mmc_get_bits(raw_csd, 128, 32, 5);
csd->wp_grp_enable = mmc_get_bits(raw_csd, 128, 31, 1);
csd->r2w_factor = 1 << mmc_get_bits(raw_csd, 128, 26, 3);
csd->write_bl_len = 1 << mmc_get_bits(raw_csd, 128, 22, 4);
csd->write_bl_partial = mmc_get_bits(raw_csd, 128, 21, 1);
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
}
static void
mmc_app_decode_scr(uint32_t *raw_scr, struct mmc_scr *scr)
{
unsigned int scr_struct;
memset(scr, 0, sizeof(*scr));
o Use seprate routines to decode cid and csd for sd and mmc cards. All they have in common right now is a memset. This saves a parameter to these routines, as well as a level of indentation. o Make mmc_get_bits a little clearer... It really only works on 128-bit registers right now.
2008-10-09 19:47:28 +00:00
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
scr_struct = mmc_get_bits(raw_scr, 64, 60, 4);
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
if (scr_struct != 0) {
printf("Unrecognised SCR structure version %d\n",
scr_struct);
return;
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
}
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
scr->sda_vsn = mmc_get_bits(raw_scr, 64, 56, 4);
scr->bus_widths = mmc_get_bits(raw_scr, 64, 48, 4);
}
static void
mmc_app_decode_sd_status(uint32_t *raw_sd_status,
struct mmc_sd_status *sd_status)
{
memset(sd_status, 0, sizeof(*sd_status));
sd_status->bus_width = mmc_get_bits(raw_sd_status, 512, 510, 2);
sd_status->secured_mode = mmc_get_bits(raw_sd_status, 512, 509, 1);
sd_status->card_type = mmc_get_bits(raw_sd_status, 512, 480, 16);
sd_status->prot_area = mmc_get_bits(raw_sd_status, 512, 448, 12);
sd_status->speed_class = mmc_get_bits(raw_sd_status, 512, 440, 8);
sd_status->perf_move = mmc_get_bits(raw_sd_status, 512, 432, 8);
sd_status->au_size = mmc_get_bits(raw_sd_status, 512, 428, 4);
sd_status->erase_size = mmc_get_bits(raw_sd_status, 512, 408, 16);
sd_status->erase_timeout = mmc_get_bits(raw_sd_status, 512, 402, 6);
sd_status->erase_offset = mmc_get_bits(raw_sd_status, 512, 400, 2);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
}
static int
mmc_all_send_cid(struct mmc_softc *sc, uint32_t *rawcid)
{
struct mmc_command cmd;
int err;
Consistently init all mmc request, command, and data structures to zero before using them.
2013-08-17 00:19:27 +00:00
memset(&cmd, 0, sizeof(cmd));
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
cmd.opcode = MMC_ALL_SEND_CID;
cmd.arg = 0;
cmd.flags = MMC_RSP_R2 | MMC_CMD_BCR;
Explicitly set data to NULL rather than relying mmc_wait_for_cmd() to do it for us. There may be commands we have to wait for that need to set data. Submitted by: mav@
2008-09-29 18:33:08 +00:00
cmd.data = NULL;
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
err = mmc_wait_for_cmd(sc->dev, sc->dev, &cmd, CMD_RETRIES);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
memcpy(rawcid, cmd.resp, 4 * sizeof(uint32_t));
return (err);
}
static int
- Fix some typos in mmc_acquire_bus() and mmc_send_csd(). - Fix some math errors in mmc_decode_csd_sd(). - Fix incorrect arguments to mmc_send_app_op_cond() in mmc_go_discovery(). - Add reporting of CSD for debug purposes. - Add detection (and skipping) of password-locked cards. - Add setting of block length on card if necessary. Submitted by: Patrick Kelsey MFC after: 3 days
2012-05-27 16:08:58 +00:00
mmc_send_csd(struct mmc_softc *sc, uint16_t rca, uint32_t *rawcsd)
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
{
struct mmc_command cmd;
int err;
Consistently init all mmc request, command, and data structures to zero before using them.
2013-08-17 00:19:27 +00:00
memset(&cmd, 0, sizeof(cmd));
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
cmd.opcode = MMC_SEND_CSD;
cmd.arg = rca << 16;
cmd.flags = MMC_RSP_R2 | MMC_CMD_BCR;
Explicitly set data to NULL rather than relying mmc_wait_for_cmd() to do it for us. There may be commands we have to wait for that need to set data. Submitted by: mav@
2008-09-29 18:33:08 +00:00
cmd.data = NULL;
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
err = mmc_wait_for_cmd(sc->dev, sc->dev, &cmd, CMD_RETRIES);
- Fix some typos in mmc_acquire_bus() and mmc_send_csd(). - Fix some math errors in mmc_decode_csd_sd(). - Fix incorrect arguments to mmc_send_app_op_cond() in mmc_go_discovery(). - Add reporting of CSD for debug purposes. - Add detection (and skipping) of password-locked cards. - Add setting of block length on card if necessary. Submitted by: Patrick Kelsey MFC after: 3 days
2012-05-27 16:08:58 +00:00
memcpy(rawcsd, cmd.resp, 4 * sizeof(uint32_t));
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
return (err);
}
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
static int
mmc_app_send_scr(struct mmc_softc *sc, uint16_t rca, uint32_t *rawscr)
{
int err;
struct mmc_command cmd;
struct mmc_data data;
Consistently init all mmc request, command, and data structures to zero before using them.
2013-08-17 00:19:27 +00:00
memset(&cmd, 0, sizeof(cmd));
memset(&data, 0, sizeof(data));
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
memset(rawscr, 0, 8);
cmd.opcode = ACMD_SEND_SCR;
cmd.flags = MMC_RSP_R1 | MMC_CMD_ADTC;
cmd.arg = 0;
cmd.data = &data;
data.data = rawscr;
data.len = 8;
data.flags = MMC_DATA_READ;
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
err = mmc_wait_for_app_cmd(sc->dev, sc->dev, rca, &cmd, CMD_RETRIES);
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
rawscr[0] = be32toh(rawscr[0]);
rawscr[1] = be32toh(rawscr[1]);
return (err);
}
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
static int
mmc_app_sd_status(struct mmc_softc *sc, uint16_t rca, uint32_t *rawsdstatus)
{
struct mmc_command cmd;
struct mmc_data data;
Fix some more overly long lines, whitespace and other bugs according to style(9) as well as spelling in comments.
2017-02-04 19:35:38 +00:00
int err, i;
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
Consistently init all mmc request, command, and data structures to zero before using them.
2013-08-17 00:19:27 +00:00
memset(&cmd, 0, sizeof(cmd));
memset(&data, 0, sizeof(data));
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
memset(rawsdstatus, 0, 64);
cmd.opcode = ACMD_SD_STATUS;
cmd.flags = MMC_RSP_R1 | MMC_CMD_ADTC;
cmd.arg = 0;
cmd.data = &data;
data.data = rawsdstatus;
data.len = 64;
data.flags = MMC_DATA_READ;
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
err = mmc_wait_for_app_cmd(sc->dev, sc->dev, rca, &cmd, CMD_RETRIES);
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
for (i = 0; i < 16; i++)
rawsdstatus[i] = be32toh(rawsdstatus[i]);
return (err);
}
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
static int
mmc_set_relative_addr(struct mmc_softc *sc, uint16_t resp)
{
struct mmc_command cmd;
int err;
Consistently init all mmc request, command, and data structures to zero before using them.
2013-08-17 00:19:27 +00:00
memset(&cmd, 0, sizeof(cmd));
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
cmd.opcode = MMC_SET_RELATIVE_ADDR;
cmd.arg = resp << 16;
cmd.flags = MMC_RSP_R6 | MMC_CMD_BCR;
cmd.data = NULL;
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
err = mmc_wait_for_cmd(sc->dev, sc->dev, &cmd, CMD_RETRIES);
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
return (err);
}
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
static int
mmc_send_relative_addr(struct mmc_softc *sc, uint32_t *resp)
{
struct mmc_command cmd;
int err;
Consistently init all mmc request, command, and data structures to zero before using them.
2013-08-17 00:19:27 +00:00
memset(&cmd, 0, sizeof(cmd));
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
cmd.opcode = SD_SEND_RELATIVE_ADDR;
cmd.arg = 0;
cmd.flags = MMC_RSP_R6 | MMC_CMD_BCR;
Explicitly set data to NULL rather than relying mmc_wait_for_cmd() to do it for us. There may be commands we have to wait for that need to set data. Submitted by: mav@
2008-09-29 18:33:08 +00:00
cmd.data = NULL;
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
err = mmc_wait_for_cmd(sc->dev, sc->dev, &cmd, CMD_RETRIES);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
*resp = cmd.resp[0];
return (err);
}
- Fix some typos in mmc_acquire_bus() and mmc_send_csd(). - Fix some math errors in mmc_decode_csd_sd(). - Fix incorrect arguments to mmc_send_app_op_cond() in mmc_go_discovery(). - Add reporting of CSD for debug purposes. - Add detection (and skipping) of password-locked cards. - Add setting of block length on card if necessary. Submitted by: Patrick Kelsey MFC after: 3 days
2012-05-27 16:08:58 +00:00
static int
mmc_set_blocklen(struct mmc_softc *sc, uint32_t len)
{
struct mmc_command cmd;
int err;
Consistently init all mmc request, command, and data structures to zero before using them.
2013-08-17 00:19:27 +00:00
memset(&cmd, 0, sizeof(cmd));
- Fix some typos in mmc_acquire_bus() and mmc_send_csd(). - Fix some math errors in mmc_decode_csd_sd(). - Fix incorrect arguments to mmc_send_app_op_cond() in mmc_go_discovery(). - Add reporting of CSD for debug purposes. - Add detection (and skipping) of password-locked cards. - Add setting of block length on card if necessary. Submitted by: Patrick Kelsey MFC after: 3 days
2012-05-27 16:08:58 +00:00
cmd.opcode = MMC_SET_BLOCKLEN;
cmd.arg = len;
cmd.flags = MMC_RSP_R1 | MMC_CMD_AC;
cmd.data = NULL;
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
err = mmc_wait_for_cmd(sc->dev, sc->dev, &cmd, CMD_RETRIES);
- Fix some typos in mmc_acquire_bus() and mmc_send_csd(). - Fix some math errors in mmc_decode_csd_sd(). - Fix incorrect arguments to mmc_send_app_op_cond() in mmc_go_discovery(). - Add reporting of CSD for debug purposes. - Add detection (and skipping) of password-locked cards. - Add setting of block length on card if necessary. Submitted by: Patrick Kelsey MFC after: 3 days
2012-05-27 16:08:58 +00:00
return (err);
}
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
static uint32_t
mmc_timing_to_dtr(struct mmc_ivars *ivar, enum mmc_bus_timing timing)
{
switch (timing) {
case bus_timing_normal:
return (ivar->tran_speed);
case bus_timing_hs:
return (ivar->hs_tran_speed);
case bus_timing_uhs_sdr12:
return (SD_SDR12_MAX);
case bus_timing_uhs_sdr25:
return (SD_SDR25_MAX);
case bus_timing_uhs_ddr50:
return (SD_DDR50_MAX);
case bus_timing_uhs_sdr50:
return (SD_SDR50_MAX);
case bus_timing_uhs_sdr104:
return (SD_SDR104_MAX);
case bus_timing_mmc_ddr52:
return (MMC_TYPE_DDR52_MAX);
case bus_timing_mmc_hs200:
case bus_timing_mmc_hs400:
case bus_timing_mmc_hs400es:
return (MMC_TYPE_HS200_HS400ES_MAX);
}
return (0);
}
static const char *
mmc_timing_to_string(enum mmc_bus_timing timing)
{
switch (timing) {
case bus_timing_normal:
return ("normal speed");
case bus_timing_hs:
return ("high speed");
case bus_timing_uhs_sdr12:
case bus_timing_uhs_sdr25:
case bus_timing_uhs_sdr50:
case bus_timing_uhs_sdr104:
return ("single data rate");
case bus_timing_uhs_ddr50:
case bus_timing_mmc_ddr52:
return ("dual data rate");
case bus_timing_mmc_hs200:
return ("HS200");
case bus_timing_mmc_hs400:
return ("HS400");
case bus_timing_mmc_hs400es:
return ("HS400 with enhanced strobe");
}
return ("");
}
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
static bool
mmc_host_timing(device_t dev, enum mmc_bus_timing timing)
{
int host_caps;
host_caps = mmcbr_get_caps(dev);
#define HOST_TIMING_CAP(host_caps, cap) ({ \
bool retval; \
if (((host_caps) & (cap)) == (cap)) \
retval = true; \
else \
retval = false; \
retval; \
})
switch (timing) {
case bus_timing_normal:
return (true);
case bus_timing_hs:
return (HOST_TIMING_CAP(host_caps, MMC_CAP_HSPEED));
case bus_timing_uhs_sdr12:
return (HOST_TIMING_CAP(host_caps, MMC_CAP_UHS_SDR12));
case bus_timing_uhs_sdr25:
return (HOST_TIMING_CAP(host_caps, MMC_CAP_UHS_SDR25));
case bus_timing_uhs_ddr50:
return (HOST_TIMING_CAP(host_caps, MMC_CAP_UHS_DDR50));
case bus_timing_uhs_sdr50:
return (HOST_TIMING_CAP(host_caps, MMC_CAP_UHS_SDR50));
case bus_timing_uhs_sdr104:
return (HOST_TIMING_CAP(host_caps, MMC_CAP_UHS_SDR104));
case bus_timing_mmc_ddr52:
return (HOST_TIMING_CAP(host_caps, MMC_CAP_MMC_DDR52));
case bus_timing_mmc_hs200:
return (HOST_TIMING_CAP(host_caps, MMC_CAP_MMC_HS200));
case bus_timing_mmc_hs400:
return (HOST_TIMING_CAP(host_caps, MMC_CAP_MMC_HS400));
case bus_timing_mmc_hs400es:
return (HOST_TIMING_CAP(host_caps, MMC_CAP_MMC_HS400 |
MMC_CAP_MMC_ENH_STROBE));
}
#undef HOST_TIMING_CAP
return (false);
}
Improve mmc driver verbose logging. Make requests logging controllable by hw.mmc.debug sysctl.
2009-01-28 22:09:00 +00:00
static void
mmc_log_card(device_t dev, struct mmc_ivars *ivar, int newcard)
{
Do pass removing some write-only variables from the kernel. This reduces noise when kernel is compiled by newer GCC versions, such as one used by external toolchain ports. Reviewed by: kib, andrew(sys/arm and sys/arm64), emaste(partial), erj(partial) Reviewed by: jhb (sys/dev/pci/* sys/kern/vfs_aio.c and sys/kern/kern_synch.c) Differential Revision: https://reviews.freebsd.org/D10385
2017-12-25 04:48:39 +00:00
enum mmc_bus_timing timing;
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
Print the card relative address in hex, because that's what all the other debugging output does (when it appears in command arguments, for example).
2013-08-16 20:22:57 +00:00
device_printf(dev, "Card at relative address 0x%04x%s:\n",
Improve mmc driver verbose logging. Make requests logging controllable by hw.mmc.debug sysctl.
2009-01-28 22:09:00 +00:00
ivar->rca, newcard ? " added" : "");
o Fixes: - When switching to 4-bit operation, send a SET_CLR_CARD_DETECT command to disconnect the card-detect pull-up resistor from the DAT3 line before sending the SET_BUS_WIDTH command. - Add the missing "reserved" zero entry to the mantissa table used to decode various CSD fields. This was causing SD cards to report that they could run at 30 MHz instead of the maximum 25 MHz mandated in the spec. o Enhancements: - At the MMC layer, format various info from the CID into a string that uniquely identifies the card instance (manufacturer number, serial number, product name and revision, etc). Export it as an instance variable. - At the MMCSD layer, display the formatted card ID string, and also report the clock speed of the hardware (not the card's max speed), and the number of bits and number of blocks per transfer. It comes out like this now: mmcsd0: 968MB <SD SD01G 8.0 SN 276886905 MFG 08/2008 by 3 SD> at mmc0 22.5MHz/4bit/128-block o Use DEVMETHOD_END. o Use NULL instead of 0 for pointers. PR: 156496 Submitted by: Ian Lepore MFC after: 1 week
2012-04-21 01:51:16 +00:00
device_printf(dev, " card: %s\n", ivar->card_id_string);
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
for (timing = bus_timing_max; timing > bus_timing_normal; timing--) {
Do pass removing some write-only variables from the kernel. This reduces noise when kernel is compiled by newer GCC versions, such as one used by external toolchain ports. Reviewed by: kib, andrew(sys/arm and sys/arm64), emaste(partial), erj(partial) Reviewed by: jhb (sys/dev/pci/* sys/kern/vfs_aio.c and sys/kern/kern_synch.c) Differential Revision: https://reviews.freebsd.org/D10385
2017-12-25 04:48:39 +00:00
if (isset(&ivar->timings, timing))
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
break;
}
- If available, use TRIM instead of ERASE for implementing BIO_DELETE. This also involves adding a quirk table as TRIM is broken for some Kingston eMMC devices, though. Compared to ERASE (declared "legacy" in the eMMC specification v5.1), TRIM has the advantage of operating on write sectors rather than on erase sectors, which typically are of a much larger size. Thus, employing TRIM, we don't need to fiddle with coalescing BIO_DELETE requests that are also of (write) sector units into erase sectors, which might not even add up in all cases. - For some SanDisk iNAND devices, the CMD38 argument, e. g. ERASE, TRIM etc., has to be specified via EXT_CSD[113], which now is also handled via a quirk. - My initial understanding was that for eMMC partitions, the granularity should be used as erase sector size, e. g. 128 KB for boot partitions. However, rereading the relevant parts of the eMMC specification v5.1, this isn't actually correct. So drop the code which used partition granularities for delmaxsize and stripesize. For the most part, this change is a NOP, though, because a) for ERASE, mmcsd_delete() used the erase sector size unconditionally for all partitions anyway and b) g_disk_limit() doesn't actually take the stripesize into account. - Take some more advantage of mmcsd_errmsg() in mmcsd(4) for making error codes human readable.
2017-08-07 23:33:05 +00:00
device_printf(dev, " quirks: %b\n", ivar->quirks, MMC_QUIRKS_FMT);
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
device_printf(dev, " bus: %ubit, %uMHz (%s timing)\n",
Improve mmc driver verbose logging. Make requests logging controllable by hw.mmc.debug sysctl.
2009-01-28 22:09:00 +00:00
(ivar->bus_width == bus_width_1 ? 1 :
(ivar->bus_width == bus_width_4 ? 4 : 8)),
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
mmc_timing_to_dtr(ivar, timing) / 1000000,
mmc_timing_to_string(timing));
Improve mmc driver verbose logging. Make requests logging controllable by hw.mmc.debug sysctl.
2009-01-28 22:09:00 +00:00
device_printf(dev, " memory: %u blocks, erase sector %u blocks%s\n",
ivar->sec_count, ivar->erase_sector,
ivar->read_only ? ", read-only" : "");
}
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
static void
mmc_discover_cards(struct mmc_softc *sc)
{
o Another round fixes for mmc(4), mmcsd(4) and sdhci(4) regarding comments, marking unused parameters as such, style(9), whitespace, etc. o In the mmc(4) bridges and sdhci(4) (bus) front-ends: - Remove redundant assignments of the default bus_generic_print_child device method (I've whipped these out of the tree as part of r227843 once, but they keep coming back ...), - use DEVMETHOD_END, - use NULL instead of 0 for pointers. o Trim/adjust includes.
2017-03-06 23:47:59 +00:00
u_char switch_res[64];
uint32_t raw_cid[4];
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
struct mmc_ivars *ivar = NULL;
- If available, use TRIM instead of ERASE for implementing BIO_DELETE. This also involves adding a quirk table as TRIM is broken for some Kingston eMMC devices, though. Compared to ERASE (declared "legacy" in the eMMC specification v5.1), TRIM has the advantage of operating on write sectors rather than on erase sectors, which typically are of a much larger size. Thus, employing TRIM, we don't need to fiddle with coalescing BIO_DELETE requests that are also of (write) sector units into erase sectors, which might not even add up in all cases. - For some SanDisk iNAND devices, the CMD38 argument, e. g. ERASE, TRIM etc., has to be specified via EXT_CSD[113], which now is also handled via a quirk. - My initial understanding was that for eMMC partitions, the granularity should be used as erase sector size, e. g. 128 KB for boot partitions. However, rereading the relevant parts of the eMMC specification v5.1, this isn't actually correct. So drop the code which used partition granularities for delmaxsize and stripesize. For the most part, this change is a NOP, though, because a) for ERASE, mmcsd_delete() used the erase sector size unconditionally for all partitions anyway and b) g_disk_limit() doesn't actually take the stripesize into account. - Take some more advantage of mmcsd_errmsg() in mmcsd(4) for making error codes human readable.
2017-08-07 23:33:05 +00:00
const struct mmc_quirk *quirk;
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
device_t child;
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
int err, host_caps, i, newcard;
o Another round fixes for mmc(4), mmcsd(4) and sdhci(4) regarding comments, marking unused parameters as such, style(9), whitespace, etc. o In the mmc(4) bridges and sdhci(4) (bus) front-ends: - Remove redundant assignments of the default bus_generic_print_child device method (I've whipped these out of the tree as part of r227843 once, but they keep coming back ...), - use DEVMETHOD_END, - use NULL instead of 0 for pointers. o Trim/adjust includes.
2017-03-06 23:47:59 +00:00
uint32_t resp, sec_count, status;
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
uint16_t rca = 2;
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
host_caps = mmcbr_get_caps(sc->dev);
Improve mmc driver verbose logging. Make requests logging controllable by hw.mmc.debug sysctl.
2009-01-28 22:09:00 +00:00
if (bootverbose || mmc_debug)
device_printf(sc->dev, "Probing cards\n");
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
while (1) {
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
child = NULL;
Log mmc and sd command failures. Reporting of routine expected errors, such as timeouts while probing a bus or testing for a feature, is squelched. Also, error reporting is limited to 5 events per second, because when an sdcard goes bad on a low-end embedded board, flooding the console at high speed isn't helpful. Original logging code contributed by Michal Meloun, but then I fancied it up with squelching and ppsratecheck.
2014-12-20 04:24:40 +00:00
sc->squelched++; /* Errors are expected, squelch reporting. */
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
err = mmc_all_send_cid(sc, raw_cid);
Log mmc and sd command failures. Reporting of routine expected errors, such as timeouts while probing a bus or testing for a feature, is squelched. Also, error reporting is limited to 5 events per second, because when an sdcard goes bad on a low-end embedded board, flooding the console at high speed isn't helpful. Original logging code contributed by Michal Meloun, but then I fancied it up with squelching and ppsratecheck.
2014-12-20 04:24:40 +00:00
sc->squelched--;
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
if (err == MMC_ERR_TIMEOUT)
break;
if (err != MMC_ERR_NONE) {
Minor style fixes from mav@ (with similar problems fixed where I noticed them): #define<tab> zero memory when we allocate it Put device name in error message. Submitted by: mav@
2008-09-29 18:17:23 +00:00
device_printf(sc->dev, "Error reading CID %d\n", err);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
break;
}
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
newcard = 1;
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
for (i = 0; i < sc->child_count; i++) {
ivar = device_get_ivars(sc->child_list[i]);
Fix overly long lines, whitespace and other bugs according to style(9).
2017-01-29 00:05:49 +00:00
if (memcmp(ivar->raw_cid, raw_cid, sizeof(raw_cid)) ==
0) {
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
newcard = 0;
break;
}
}
Improve mmc driver verbose logging. Make requests logging controllable by hw.mmc.debug sysctl.
2009-01-28 22:09:00 +00:00
if (bootverbose || mmc_debug) {
Fix overly long lines, whitespace and other bugs according to style(9).
2017-01-29 00:05:49 +00:00
device_printf(sc->dev,
"%sard detected (CID %08x%08x%08x%08x)\n",
Improve mmc driver verbose logging. Make requests logging controllable by hw.mmc.debug sysctl.
2009-01-28 22:09:00 +00:00
newcard ? "New c" : "C",
raw_cid[0], raw_cid[1], raw_cid[2], raw_cid[3]);
}
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
if (newcard) {
ivar = malloc(sizeof(struct mmc_ivars), M_DEVBUF,
M_WAITOK | M_ZERO);
memcpy(ivar->raw_cid, raw_cid, sizeof(raw_cid));
}
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
if (mmcbr_get_ro(sc->dev))
ivar->read_only = 1;
ivar->bus_width = bus_width_1;
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
setbit(&ivar->timings, bus_timing_normal);
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
ivar->mode = mmcbr_get_mode(sc->dev);
if (ivar->mode == mode_sd) {
o Use seprate routines to decode cid and csd for sd and mmc cards. All they have in common right now is a memset. This saves a parameter to these routines, as well as a level of indentation. o Make mmc_get_bits a little clearer... It really only works on 128-bit registers right now.
2008-10-09 19:47:28 +00:00
mmc_decode_cid_sd(ivar->raw_cid, &ivar->cid);
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
err = mmc_send_relative_addr(sc, &resp);
if (err != MMC_ERR_NONE) {
device_printf(sc->dev,
"Error getting RCA %d\n", err);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
goto free_ivar;
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
}
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
ivar->rca = resp >> 16;
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
/* Get card CSD. */
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
err = mmc_send_csd(sc, ivar->rca, ivar->raw_csd);
if (err != MMC_ERR_NONE) {
device_printf(sc->dev,
"Error getting CSD %d\n", err);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
goto free_ivar;
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
}
- Fix some typos in mmc_acquire_bus() and mmc_send_csd(). - Fix some math errors in mmc_decode_csd_sd(). - Fix incorrect arguments to mmc_send_app_op_cond() in mmc_go_discovery(). - Add reporting of CSD for debug purposes. - Add detection (and skipping) of password-locked cards. - Add setting of block length on card if necessary. Submitted by: Patrick Kelsey MFC after: 3 days
2012-05-27 16:08:58 +00:00
if (bootverbose || mmc_debug)
device_printf(sc->dev,
"%sard detected (CSD %08x%08x%08x%08x)\n",
newcard ? "New c" : "C", ivar->raw_csd[0],
ivar->raw_csd[1], ivar->raw_csd[2],
ivar->raw_csd[3]);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
err = mmc_decode_csd_sd(ivar->raw_csd, &ivar->csd);
if (err != MMC_ERR_NONE) {
device_printf(sc->dev, "Error decoding CSD\n");
goto free_ivar;
}
Add high capacity MMC cards support.
2008-10-09 20:09:56 +00:00
ivar->sec_count = ivar->csd.capacity / MMC_SECTOR_SIZE;
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
if (ivar->csd.csd_structure > 0)
ivar->high_cap = 1;
ivar->tran_speed = ivar->csd.tran_speed;
sdhci/mmc: Minor whitespace cleanups No functional change. Submitted by: Johannes Lundberg <yohanesu75 at gmail.com>
2016-12-20 03:38:14 +00:00
ivar->erase_sector = ivar->csd.erase_sector *
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
ivar->csd.write_bl_len / MMC_SECTOR_SIZE;
sdhci/mmc: Minor whitespace cleanups No functional change. Submitted by: Johannes Lundberg <yohanesu75 at gmail.com>
2016-12-20 03:38:14 +00:00
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
err = mmc_send_status(sc->dev, sc->dev, ivar->rca,
&status);
- Fix some typos in mmc_acquire_bus() and mmc_send_csd(). - Fix some math errors in mmc_decode_csd_sd(). - Fix incorrect arguments to mmc_send_app_op_cond() in mmc_go_discovery(). - Add reporting of CSD for debug purposes. - Add detection (and skipping) of password-locked cards. - Add setting of block length on card if necessary. Submitted by: Patrick Kelsey MFC after: 3 days
2012-05-27 16:08:58 +00:00
if (err != MMC_ERR_NONE) {
device_printf(sc->dev,
"Error reading card status %d\n", err);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
goto free_ivar;
- Fix some typos in mmc_acquire_bus() and mmc_send_csd(). - Fix some math errors in mmc_decode_csd_sd(). - Fix incorrect arguments to mmc_send_app_op_cond() in mmc_go_discovery(). - Add reporting of CSD for debug purposes. - Add detection (and skipping) of password-locked cards. - Add setting of block length on card if necessary. Submitted by: Patrick Kelsey MFC after: 3 days
2012-05-27 16:08:58 +00:00
}
if ((status & R1_CARD_IS_LOCKED) != 0) {
device_printf(sc->dev,
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
"Card is password protected, skipping\n");
goto free_ivar;
- Fix some typos in mmc_acquire_bus() and mmc_send_csd(). - Fix some math errors in mmc_decode_csd_sd(). - Fix incorrect arguments to mmc_send_app_op_cond() in mmc_go_discovery(). - Add reporting of CSD for debug purposes. - Add detection (and skipping) of password-locked cards. - Add setting of block length on card if necessary. Submitted by: Patrick Kelsey MFC after: 3 days
2012-05-27 16:08:58 +00:00
}
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
/* Get card SCR. Card must be selected to fetch it. */
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
err = mmc_select_card(sc, ivar->rca);
if (err != MMC_ERR_NONE) {
device_printf(sc->dev,
"Error selecting card %d\n", err);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
goto free_ivar;
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
}
err = mmc_app_send_scr(sc, ivar->rca, ivar->raw_scr);
if (err != MMC_ERR_NONE) {
device_printf(sc->dev,
"Error reading SCR %d\n", err);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
goto free_ivar;
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
}
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
mmc_app_decode_scr(ivar->raw_scr, &ivar->scr);
o Define some symbols for a few items that are bare constants in the code. o Use NULL in preference to 0 for a few pointers. o default to bus timing normal, like we default to bus_width_1.
2009-02-03 04:28:45 +00:00
/* Get card switch capabilities (command class 10). */
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
if ((ivar->scr.sda_vsn >= 1) &&
Fix overly long lines, whitespace and other bugs according to style(9).
2017-01-29 00:05:49 +00:00
(ivar->csd.ccc & (1 << 10))) {
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
err = mmc_sd_switch(sc, SD_SWITCH_MODE_CHECK,
o Define some symbols for a few items that are bare constants in the code. o Use NULL in preference to 0 for a few pointers. o default to bus timing normal, like we default to bus_width_1.
2009-02-03 04:28:45 +00:00
SD_SWITCH_GROUP1, SD_SWITCH_NOCHANGE,
switch_res);
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
if (err == MMC_ERR_NONE &&
switch_res[13] & (1 << SD_SWITCH_HS_MODE)) {
setbit(&ivar->timings, bus_timing_hs);
ivar->hs_tran_speed = SD_HS_MAX;
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
}
}
Re-select the SD card before getting the SD status. On a couple Atmel boards, this prevents some error messages during enumeration and also gives us the correct erase block size. They appear to be harmless elsewhere. # Note: we treat too many commands as 'can't fail' if they don't work # after a couple of retries. We need to fix that, but not today...
2015-05-19 21:16:53 +00:00
/*
Deselect the sd card before re-selecting it when working around a problem with some cards that causes them to become deselected after probing for switch capabilities. The old workaround fixes the behavior with some cards, but causes problems with the cards the behave correctly and don't become deselected. Forcing a deselect then reselect appears to work correctly with all cards in initial testing.
2015-07-18 16:56:51 +00:00
* We deselect then reselect the card here. Some cards
* become unselected and timeout with the above two
* commands, although the state tables / diagrams in the
* standard suggest they go back to the transfer state.
* Other cards don't become deselected, and if we
Fix some more overly long lines, whitespace and other bugs according to style(9) as well as spelling in comments.
2017-02-04 19:35:38 +00:00
* attempt to blindly re-select them, we get timeout
Deselect the sd card before re-selecting it when working around a problem with some cards that causes them to become deselected after probing for switch capabilities. The old workaround fixes the behavior with some cards, but causes problems with the cards the behave correctly and don't become deselected. Forcing a deselect then reselect appears to work correctly with all cards in initial testing.
2015-07-18 16:56:51 +00:00
* errors from some controllers. So we deselect then
* reselect to handle all situations. The only thing we
* use from the sd_status is the erase sector size, but
* it is still nice to get that right.
Re-select the SD card before getting the SD status. On a couple Atmel boards, this prevents some error messages during enumeration and also gives us the correct erase block size. They appear to be harmless elsewhere. # Note: we treat too many commands as 'can't fail' if they don't work # after a couple of retries. We need to fix that, but not today...
2015-05-19 21:16:53 +00:00
*/
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
(void)mmc_select_card(sc, 0);
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
(void)mmc_select_card(sc, ivar->rca);
(void)mmc_app_sd_status(sc, ivar->rca,
ivar->raw_sd_status);
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
mmc_app_decode_sd_status(ivar->raw_sd_status,
&ivar->sd_status);
if (ivar->sd_status.au_size != 0) {
ivar->erase_sector =
16 << ivar->sd_status.au_size;
}
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
/* Find maximum supported bus width. */
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
if ((host_caps & MMC_CAP_4_BIT_DATA) &&
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
(ivar->scr.bus_widths & SD_SCR_BUS_WIDTH_4))
ivar->bus_width = bus_width_4;
- Fix some typos in mmc_acquire_bus() and mmc_send_csd(). - Fix some math errors in mmc_decode_csd_sd(). - Fix incorrect arguments to mmc_send_app_op_cond() in mmc_go_discovery(). - Add reporting of CSD for debug purposes. - Add detection (and skipping) of password-locked cards. - Add setting of block length on card if necessary. Submitted by: Patrick Kelsey MFC after: 3 days
2012-05-27 16:08:58 +00:00
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
goto child_common;
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
}
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
ivar->rca = rca++;
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
err = mmc_set_relative_addr(sc, ivar->rca);
if (err != MMC_ERR_NONE) {
device_printf(sc->dev, "Error setting RCA %d\n", err);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
goto free_ivar;
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
}
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
/* Get card CSD. */
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
err = mmc_send_csd(sc, ivar->rca, ivar->raw_csd);
if (err != MMC_ERR_NONE) {
device_printf(sc->dev, "Error getting CSD %d\n", err);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
goto free_ivar;
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
}
- Fix some typos in mmc_acquire_bus() and mmc_send_csd(). - Fix some math errors in mmc_decode_csd_sd(). - Fix incorrect arguments to mmc_send_app_op_cond() in mmc_go_discovery(). - Add reporting of CSD for debug purposes. - Add detection (and skipping) of password-locked cards. - Add setting of block length on card if necessary. Submitted by: Patrick Kelsey MFC after: 3 days
2012-05-27 16:08:58 +00:00
if (bootverbose || mmc_debug)
device_printf(sc->dev,
"%sard detected (CSD %08x%08x%08x%08x)\n",
newcard ? "New c" : "C", ivar->raw_csd[0],
ivar->raw_csd[1], ivar->raw_csd[2],
ivar->raw_csd[3]);
o Use seprate routines to decode cid and csd for sd and mmc cards. All they have in common right now is a memset. This saves a parameter to these routines, as well as a level of indentation. o Make mmc_get_bits a little clearer... It really only works on 128-bit registers right now.
2008-10-09 19:47:28 +00:00
mmc_decode_csd_mmc(ivar->raw_csd, &ivar->csd);
Add high capacity MMC cards support.
2008-10-09 20:09:56 +00:00
ivar->sec_count = ivar->csd.capacity / MMC_SECTOR_SIZE;
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
ivar->tran_speed = ivar->csd.tran_speed;
sdhci/mmc: Minor whitespace cleanups No functional change. Submitted by: Johannes Lundberg <yohanesu75 at gmail.com>
2016-12-20 03:38:14 +00:00
ivar->erase_sector = ivar->csd.erase_sector *
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
ivar->csd.write_bl_len / MMC_SECTOR_SIZE;
- Fix some typos in mmc_acquire_bus() and mmc_send_csd(). - Fix some math errors in mmc_decode_csd_sd(). - Fix incorrect arguments to mmc_send_app_op_cond() in mmc_go_discovery(). - Add reporting of CSD for debug purposes. - Add detection (and skipping) of password-locked cards. - Add setting of block length on card if necessary. Submitted by: Patrick Kelsey MFC after: 3 days
2012-05-27 16:08:58 +00:00
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
err = mmc_send_status(sc->dev, sc->dev, ivar->rca, &status);
- Fix some typos in mmc_acquire_bus() and mmc_send_csd(). - Fix some math errors in mmc_decode_csd_sd(). - Fix incorrect arguments to mmc_send_app_op_cond() in mmc_go_discovery(). - Add reporting of CSD for debug purposes. - Add detection (and skipping) of password-locked cards. - Add setting of block length on card if necessary. Submitted by: Patrick Kelsey MFC after: 3 days
2012-05-27 16:08:58 +00:00
if (err != MMC_ERR_NONE) {
device_printf(sc->dev,
"Error reading card status %d\n", err);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
goto free_ivar;
- Fix some typos in mmc_acquire_bus() and mmc_send_csd(). - Fix some math errors in mmc_decode_csd_sd(). - Fix incorrect arguments to mmc_send_app_op_cond() in mmc_go_discovery(). - Add reporting of CSD for debug purposes. - Add detection (and skipping) of password-locked cards. - Add setting of block length on card if necessary. Submitted by: Patrick Kelsey MFC after: 3 days
2012-05-27 16:08:58 +00:00
}
if ((status & R1_CARD_IS_LOCKED) != 0) {
device_printf(sc->dev,
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
"Card is password protected, skipping\n");
goto free_ivar;
- Fix some typos in mmc_acquire_bus() and mmc_send_csd(). - Fix some math errors in mmc_decode_csd_sd(). - Fix incorrect arguments to mmc_send_app_op_cond() in mmc_go_discovery(). - Add reporting of CSD for debug purposes. - Add detection (and skipping) of password-locked cards. - Add setting of block length on card if necessary. Submitted by: Patrick Kelsey MFC after: 3 days
2012-05-27 16:08:58 +00:00
}
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
err = mmc_select_card(sc, ivar->rca);
if (err != MMC_ERR_NONE) {
device_printf(sc->dev, "Error selecting card %d\n",
err);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
goto free_ivar;
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
}
Always select the card before we do the 4.x specific stuff and deselect it after setting the block size. This is a similar bug that was fixed elsewhere, but not here. This makes sure that we leave the card deselected at the end of the loop, and we don't send any commands to the card without it selected. Reviewed by: ian@
2014-12-23 05:50:53 +00:00
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
/* Only MMC >= 4.x devices support EXT_CSD. */
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
if (ivar->csd.spec_vers >= 4) {
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
err = mmc_send_ext_csd(sc->dev, sc->dev,
ivar->raw_ext_csd);
if (err != MMC_ERR_NONE) {
device_printf(sc->dev,
"Error reading EXT_CSD %d\n", err);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
goto free_ivar;
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
}
Add high capacity MMC cards support.
2008-10-09 20:09:56 +00:00
/* Handle extended capacity from EXT_CSD */
sec_count = ivar->raw_ext_csd[EXT_CSD_SEC_CNT] +
(ivar->raw_ext_csd[EXT_CSD_SEC_CNT + 1] << 8) +
(ivar->raw_ext_csd[EXT_CSD_SEC_CNT + 2] << 16) +
(ivar->raw_ext_csd[EXT_CSD_SEC_CNT + 3] << 24);
if (sec_count != 0) {
ivar->sec_count = sec_count;
ivar->high_cap = 1;
}
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
/* Find maximum supported bus width. */
ivar->bus_width = mmc_test_bus_width(sc);
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
/* Get device speeds beyond normal mode. */
if ((ivar->raw_ext_csd[EXT_CSD_CARD_TYPE] &
EXT_CSD_CARD_TYPE_HS_52) != 0) {
setbit(&ivar->timings, bus_timing_hs);
ivar->hs_tran_speed = MMC_TYPE_HS_52_MAX;
} else if ((ivar->raw_ext_csd[EXT_CSD_CARD_TYPE] &
EXT_CSD_CARD_TYPE_HS_26) != 0) {
setbit(&ivar->timings, bus_timing_hs);
ivar->hs_tran_speed = MMC_TYPE_HS_26_MAX;
}
if ((ivar->raw_ext_csd[EXT_CSD_CARD_TYPE] &
EXT_CSD_CARD_TYPE_DDR_52_1_2V) != 0 &&
(host_caps & MMC_CAP_SIGNALING_120) != 0) {
setbit(&ivar->timings, bus_timing_mmc_ddr52);
setbit(&ivar->vccq_120, bus_timing_mmc_ddr52);
}
if ((ivar->raw_ext_csd[EXT_CSD_CARD_TYPE] &
EXT_CSD_CARD_TYPE_DDR_52_1_8V) != 0 &&
(host_caps & MMC_CAP_SIGNALING_180) != 0) {
setbit(&ivar->timings, bus_timing_mmc_ddr52);
setbit(&ivar->vccq_180, bus_timing_mmc_ddr52);
}
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
if ((ivar->raw_ext_csd[EXT_CSD_CARD_TYPE] &
EXT_CSD_CARD_TYPE_HS200_1_2V) != 0 &&
(host_caps & MMC_CAP_SIGNALING_120) != 0) {
setbit(&ivar->timings, bus_timing_mmc_hs200);
setbit(&ivar->vccq_120, bus_timing_mmc_hs200);
}
if ((ivar->raw_ext_csd[EXT_CSD_CARD_TYPE] &
EXT_CSD_CARD_TYPE_HS200_1_8V) != 0 &&
(host_caps & MMC_CAP_SIGNALING_180) != 0) {
setbit(&ivar->timings, bus_timing_mmc_hs200);
setbit(&ivar->vccq_180, bus_timing_mmc_hs200);
}
if ((ivar->raw_ext_csd[EXT_CSD_CARD_TYPE] &
EXT_CSD_CARD_TYPE_HS400_1_2V) != 0 &&
(host_caps & MMC_CAP_SIGNALING_120) != 0 &&
ivar->bus_width == bus_width_8) {
setbit(&ivar->timings, bus_timing_mmc_hs400);
setbit(&ivar->vccq_120, bus_timing_mmc_hs400);
}
if ((ivar->raw_ext_csd[EXT_CSD_CARD_TYPE] &
EXT_CSD_CARD_TYPE_HS400_1_8V) != 0 &&
(host_caps & MMC_CAP_SIGNALING_180) != 0 &&
ivar->bus_width == bus_width_8) {
setbit(&ivar->timings, bus_timing_mmc_hs400);
setbit(&ivar->vccq_180, bus_timing_mmc_hs400);
}
if ((ivar->raw_ext_csd[EXT_CSD_CARD_TYPE] &
EXT_CSD_CARD_TYPE_HS400_1_2V) != 0 &&
(ivar->raw_ext_csd[EXT_CSD_STROBE_SUPPORT] &
EXT_CSD_STROBE_SUPPORT_EN) != 0 &&
(host_caps & MMC_CAP_SIGNALING_120) != 0 &&
ivar->bus_width == bus_width_8) {
setbit(&ivar->timings, bus_timing_mmc_hs400es);
setbit(&ivar->vccq_120, bus_timing_mmc_hs400es);
}
if ((ivar->raw_ext_csd[EXT_CSD_CARD_TYPE] &
EXT_CSD_CARD_TYPE_HS400_1_8V) != 0 &&
(ivar->raw_ext_csd[EXT_CSD_STROBE_SUPPORT] &
EXT_CSD_STROBE_SUPPORT_EN) != 0 &&
(host_caps & MMC_CAP_SIGNALING_180) != 0 &&
ivar->bus_width == bus_width_8) {
setbit(&ivar->timings, bus_timing_mmc_hs400es);
setbit(&ivar->vccq_180, bus_timing_mmc_hs400es);
}
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
/*
* Determine generic switch timeout (provided in
* units of 10 ms), defaulting to 500 ms.
*/
ivar->cmd6_time = 500 * 1000;
Use the correct revision specifier (EXT_CSD revision rather than system specification version) for deciding whether the EXT_CSD register includes the EXT_CSD_GEN_CMD6_TIME field. Submitted by: Masanobu SAITOH
2018-01-13 17:36:11 +00:00
if (ivar->raw_ext_csd[EXT_CSD_REV] >= 6)
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
ivar->cmd6_time = 10 *
ivar->raw_ext_csd[EXT_CSD_GEN_CMD6_TIME];
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
/* Handle HC erase sector size. */
if (ivar->raw_ext_csd[EXT_CSD_ERASE_GRP_SIZE] != 0) {
ivar->erase_sector = 1024 *
ivar->raw_ext_csd[EXT_CSD_ERASE_GRP_SIZE];
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
err = mmc_switch(sc->dev, sc->dev, ivar->rca,
EXT_CSD_CMD_SET_NORMAL,
EXT_CSD_ERASE_GRP_DEF,
EXT_CSD_ERASE_GRP_DEF_EN,
ivar->cmd6_time, true);
if (err != MMC_ERR_NONE) {
device_printf(sc->dev,
"Error setting erase group %d\n",
err);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
goto free_ivar;
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
}
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
}
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
}
- Fix some typos in mmc_acquire_bus() and mmc_send_csd(). - Fix some math errors in mmc_decode_csd_sd(). - Fix incorrect arguments to mmc_send_app_op_cond() in mmc_go_discovery(). - Add reporting of CSD for debug purposes. - Add detection (and skipping) of password-locked cards. - Add setting of block length on card if necessary. Submitted by: Patrick Kelsey MFC after: 3 days
2012-05-27 16:08:58 +00:00
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
mmc_decode_cid_mmc(ivar->raw_cid, &ivar->cid,
ivar->raw_ext_csd[EXT_CSD_REV] >= 5);
child_common:
- If available, use TRIM instead of ERASE for implementing BIO_DELETE. This also involves adding a quirk table as TRIM is broken for some Kingston eMMC devices, though. Compared to ERASE (declared "legacy" in the eMMC specification v5.1), TRIM has the advantage of operating on write sectors rather than on erase sectors, which typically are of a much larger size. Thus, employing TRIM, we don't need to fiddle with coalescing BIO_DELETE requests that are also of (write) sector units into erase sectors, which might not even add up in all cases. - For some SanDisk iNAND devices, the CMD38 argument, e. g. ERASE, TRIM etc., has to be specified via EXT_CSD[113], which now is also handled via a quirk. - My initial understanding was that for eMMC partitions, the granularity should be used as erase sector size, e. g. 128 KB for boot partitions. However, rereading the relevant parts of the eMMC specification v5.1, this isn't actually correct. So drop the code which used partition granularities for delmaxsize and stripesize. For the most part, this change is a NOP, though, because a) for ERASE, mmcsd_delete() used the erase sector size unconditionally for all partitions anyway and b) g_disk_limit() doesn't actually take the stripesize into account. - Take some more advantage of mmcsd_errmsg() in mmcsd(4) for making error codes human readable.
2017-08-07 23:33:05 +00:00
for (quirk = &mmc_quirks[0]; quirk->mid != 0x0; quirk++) {
if ((quirk->mid == MMC_QUIRK_MID_ANY ||
quirk->mid == ivar->cid.mid) &&
(quirk->oid == MMC_QUIRK_OID_ANY ||
quirk->oid == ivar->cid.oid) &&
strncmp(quirk->pnm, ivar->cid.pnm,
sizeof(ivar->cid.pnm)) == 0) {
ivar->quirks = quirk->quirks;
break;
}
}
- Fix some typos in mmc_acquire_bus() and mmc_send_csd(). - Fix some math errors in mmc_decode_csd_sd(). - Fix incorrect arguments to mmc_send_app_op_cond() in mmc_go_discovery(). - Add reporting of CSD for debug purposes. - Add detection (and skipping) of password-locked cards. - Add setting of block length on card if necessary. Submitted by: Patrick Kelsey MFC after: 3 days
2012-05-27 16:08:58 +00:00
/*
* Some cards that report maximum I/O block sizes greater
* than 512 require the block length to be set to 512, even
* though that is supposed to be the default. Example:
*
* Transcend 2GB SDSC card, CID:
* mid=0x1b oid=0x534d pnm="00000" prv=1.0 mdt=00.2000
*/
if (ivar->csd.read_bl_len != MMC_SECTOR_SIZE ||
ivar->csd.write_bl_len != MMC_SECTOR_SIZE)
mmc_set_blocklen(sc, MMC_SECTOR_SIZE);
o Fixes: - When switching to 4-bit operation, send a SET_CLR_CARD_DETECT command to disconnect the card-detect pull-up resistor from the DAT3 line before sending the SET_BUS_WIDTH command. - Add the missing "reserved" zero entry to the mantissa table used to decode various CSD fields. This was causing SD cards to report that they could run at 30 MHz instead of the maximum 25 MHz mandated in the spec. o Enhancements: - At the MMC layer, format various info from the CID into a string that uniquely identifies the card instance (manufacturer number, serial number, product name and revision, etc). Export it as an instance variable. - At the MMCSD layer, display the formatted card ID string, and also report the clock speed of the hardware (not the card's max speed), and the number of bits and number of blocks per transfer. It comes out like this now: mmcsd0: 968MB <SD SD01G 8.0 SN 276886905 MFG 08/2008 by 3 SD> at mmc0 22.5MHz/4bit/128-block o Use DEVMETHOD_END. o Use NULL instead of 0 for pointers. PR: 156496 Submitted by: Ian Lepore MFC after: 1 week
2012-04-21 01:51:16 +00:00
mmc_format_card_id_string(ivar);
- Fix some typos in mmc_acquire_bus() and mmc_send_csd(). - Fix some math errors in mmc_decode_csd_sd(). - Fix incorrect arguments to mmc_send_app_op_cond() in mmc_go_discovery(). - Add reporting of CSD for debug purposes. - Add detection (and skipping) of password-locked cards. - Add setting of block length on card if necessary. Submitted by: Patrick Kelsey MFC after: 3 days
2012-05-27 16:08:58 +00:00
Improve mmc driver verbose logging. Make requests logging controllable by hw.mmc.debug sysctl.
2009-01-28 22:09:00 +00:00
if (bootverbose || mmc_debug)
mmc_log_card(sc->dev, ivar, newcard);
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
if (newcard) {
/* Add device. */
child = device_add_child(sc->dev, NULL, -1);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
if (child != NULL) {
device_set_ivars(child, ivar);
sc->child_list = realloc(sc->child_list,
Correctly use the size of a pointer rather than that of a pointer to a pointer. Reported by: Coverity CID: 1378432
2017-07-26 21:59:37 +00:00
sizeof(device_t) * sc->child_count + 1,
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
M_DEVBUF, M_WAITOK);
sc->child_list[sc->child_count++] = child;
} else
device_printf(sc->dev, "Error adding child\n");
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
}
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
free_ivar:
if (newcard && child == NULL)
free(ivar, M_DEVBUF);
(void)mmc_select_card(sc, 0);
/*
* Not returning here when one MMC device could no be added
* potentially would mean looping forever when that device
* is broken (in which case it also may impact the remainder
* of the bus anyway, though).
*/
if ((newcard && child == NULL) ||
mmcbr_get_mode(sc->dev) == mode_sd)
return;
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
}
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
}
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
static void
mmc_update_child_list(struct mmc_softc *sc)
{
device_t child;
int i, j;
if (sc->child_count == 0) {
free(sc->child_list, M_DEVBUF);
return;
}
for (i = j = 0; i < sc->child_count; i++) {
for (;;) {
child = sc->child_list[j++];
if (child != NULL)
break;
}
if (i != j)
sc->child_list[i] = child;
}
Correctly use the size of a pointer rather than that of a pointer to a pointer. Reported by: Coverity CID: 1378432
2017-07-26 21:59:37 +00:00
sc->child_list = realloc(sc->child_list, sizeof(device_t) *
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
sc->child_count, M_DEVBUF, M_WAITOK);
}
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
static void
mmc_rescan_cards(struct mmc_softc *sc)
{
Fix some more overly long lines, whitespace and other bugs according to style(9) as well as spelling in comments.
2017-02-04 19:35:38 +00:00
struct mmc_ivars *ivar;
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
int err, i, j;
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
for (i = j = 0; i < sc->child_count; i++) {
ivar = device_get_ivars(sc->child_list[i]);
Again, fixes regarding style(4), to comments, includes and unused parameters.
2017-03-17 22:57:37 +00:00
if (mmc_select_card(sc, ivar->rca) != MMC_ERR_NONE) {
Improve mmc driver verbose logging. Make requests logging controllable by hw.mmc.debug sysctl.
2009-01-28 22:09:00 +00:00
if (bootverbose || mmc_debug)
Fix overly long lines, whitespace and other bugs according to style(9).
2017-01-29 00:05:49 +00:00
device_printf(sc->dev,
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
"Card at relative address %d lost\n",
Improve mmc driver verbose logging. Make requests logging controllable by hw.mmc.debug sysctl.
2009-01-28 22:09:00 +00:00
ivar->rca);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
err = device_delete_child(sc->dev, sc->child_list[i]);
if (err != 0) {
j++;
continue;
}
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
free(ivar, M_DEVBUF);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
} else
j++;
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
}
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
if (sc->child_count == j)
goto out;
sc->child_count = j;
mmc_update_child_list(sc);
out:
(void)mmc_select_card(sc, 0);
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
}
static int
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
mmc_delete_cards(struct mmc_softc *sc, bool final)
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
{
struct mmc_ivars *ivar;
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
int err, i, j;
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
err = 0;
for (i = j = 0; i < sc->child_count; i++) {
ivar = device_get_ivars(sc->child_list[i]);
Improve mmc driver verbose logging. Make requests logging controllable by hw.mmc.debug sysctl.
2009-01-28 22:09:00 +00:00
if (bootverbose || mmc_debug)
Fix overly long lines, whitespace and other bugs according to style(9).
2017-01-29 00:05:49 +00:00
device_printf(sc->dev,
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
"Card at relative address %d deleted\n",
Improve mmc driver verbose logging. Make requests logging controllable by hw.mmc.debug sysctl.
2009-01-28 22:09:00 +00:00
ivar->rca);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
err = device_delete_child(sc->dev, sc->child_list[i]);
if (err != 0) {
j++;
if (final == false)
continue;
else
break;
}
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
free(ivar, M_DEVBUF);
}
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
sc->child_count = j;
mmc_update_child_list(sc);
return (err);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
}
static void
mmc_go_discovery(struct mmc_softc *sc)
{
uint32_t ocr;
device_t dev;
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
int err;
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
dev = sc->dev;
if (mmcbr_get_power_mode(dev) != power_on) {
Convert all C++ comments to C style. There's lots of folks that don't like them and they don't add enough to be worth standing out.
2008-09-29 01:32:21 +00:00
/*
* First, try SD modes
*/
Log mmc and sd command failures. Reporting of routine expected errors, such as timeouts while probing a bus or testing for a feature, is squelched. Also, error reporting is limited to 5 events per second, because when an sdcard goes bad on a low-end embedded board, flooding the console at high speed isn't helpful. Original logging code contributed by Michal Meloun, but then I fancied it up with squelching and ppsratecheck.
2014-12-20 04:24:40 +00:00
sc->squelched++; /* Errors are expected, squelch reporting. */
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
mmcbr_set_mode(dev, mode_sd);
mmc_power_up(sc);
mmcbr_set_bus_mode(dev, pushpull);
Improve mmc driver verbose logging. Make requests logging controllable by hw.mmc.debug sysctl.
2009-01-28 22:09:00 +00:00
if (bootverbose || mmc_debug)
device_printf(sc->dev, "Probing bus\n");
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
mmc_idle_cards(sc);
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
err = mmc_send_if_cond(sc, 1);
Improve mmc driver verbose logging. Make requests logging controllable by hw.mmc.debug sysctl.
2009-01-28 22:09:00 +00:00
if ((bootverbose || mmc_debug) && err == 0)
Fix overly long lines, whitespace and other bugs according to style(9).
2017-01-29 00:05:49 +00:00
device_printf(sc->dev,
"SD 2.0 interface conditions: OK\n");
- Fix some typos in mmc_acquire_bus() and mmc_send_csd(). - Fix some math errors in mmc_decode_csd_sd(). - Fix incorrect arguments to mmc_send_app_op_cond() in mmc_go_discovery(). - Add reporting of CSD for debug purposes. - Add detection (and skipping) of password-locked cards. - Add setting of block length on card if necessary. Submitted by: Patrick Kelsey MFC after: 3 days
2012-05-27 16:08:58 +00:00
if (mmc_send_app_op_cond(sc, 0, &ocr) != MMC_ERR_NONE) {
Improve mmc driver verbose logging. Make requests logging controllable by hw.mmc.debug sysctl.
2009-01-28 22:09:00 +00:00
if (bootverbose || mmc_debug)
device_printf(sc->dev, "SD probe: failed\n");
Convert all C++ comments to C style. There's lots of folks that don't like them and they don't add enough to be worth standing out.
2008-09-29 01:32:21 +00:00
/*
* Failed, try MMC
*/
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
mmcbr_set_mode(dev, mode_mmc);
Improve mmc driver verbose logging. Make requests logging controllable by hw.mmc.debug sysctl.
2009-01-28 22:09:00 +00:00
if (mmc_send_op_cond(sc, 0, &ocr) != MMC_ERR_NONE) {
if (bootverbose || mmc_debug)
Fix overly long lines, whitespace and other bugs according to style(9).
2017-01-29 00:05:49 +00:00
device_printf(sc->dev,
"MMC probe: failed\n");
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
ocr = 0; /* Failed both, powerdown. */
Improve mmc driver verbose logging. Make requests logging controllable by hw.mmc.debug sysctl.
2009-01-28 22:09:00 +00:00
} else if (bootverbose || mmc_debug)
device_printf(sc->dev,
"MMC probe: OK (OCR: 0x%08x)\n", ocr);
} else if (bootverbose || mmc_debug)
Fix overly long lines, whitespace and other bugs according to style(9).
2017-01-29 00:05:49 +00:00
device_printf(sc->dev, "SD probe: OK (OCR: 0x%08x)\n",
ocr);
Log mmc and sd command failures. Reporting of routine expected errors, such as timeouts while probing a bus or testing for a feature, is squelched. Also, error reporting is limited to 5 events per second, because when an sdcard goes bad on a low-end embedded board, flooding the console at high speed isn't helpful. Original logging code contributed by Michal Meloun, but then I fancied it up with squelching and ppsratecheck.
2014-12-20 04:24:40 +00:00
sc->squelched--;
Improve mmc driver verbose logging. Make requests logging controllable by hw.mmc.debug sysctl.
2009-01-28 22:09:00 +00:00
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
mmcbr_set_ocr(dev, mmc_select_vdd(sc, ocr));
if (mmcbr_get_ocr(dev) != 0)
mmc_idle_cards(sc);
} else {
mmcbr_set_bus_mode(dev, opendrain);
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
mmcbr_set_clock(dev, SD_MMC_CARD_ID_FREQUENCY);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
mmcbr_update_ios(dev);
Convert all C++ comments to C style. There's lots of folks that don't like them and they don't add enough to be worth standing out.
2008-09-29 01:32:21 +00:00
/* XXX recompute vdd based on new cards? */
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
}
/*
* Make sure that we have a mutually agreeable voltage to at least
* one card on the bus.
*/
Improve mmc driver verbose logging. Make requests logging controllable by hw.mmc.debug sysctl.
2009-01-28 22:09:00 +00:00
if (bootverbose || mmc_debug)
Fix overly long lines, whitespace and other bugs according to style(9).
2017-01-29 00:05:49 +00:00
device_printf(sc->dev, "Current OCR: 0x%08x\n",
mmcbr_get_ocr(dev));
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
if (mmcbr_get_ocr(dev) == 0) {
If no compatible cards were found after probing the bus, say so.
2014-02-15 20:36:54 +00:00
device_printf(sc->dev, "No compatible cards found on bus\n");
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
(void)mmc_delete_cards(sc, false);
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
mmc_power_down(sc);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
return;
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
}
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
/*
* Reselect the cards after we've idled them above.
*/
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
if (mmcbr_get_mode(dev) == mode_sd) {
err = mmc_send_if_cond(sc, 1);
mmc_send_app_op_cond(sc,
style(9): spaces around operators.
2008-10-12 07:30:05 +00:00
(err ? 0 : MMC_OCR_CCS) | mmcbr_get_ocr(dev), NULL);
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
} else
MMC: Signal HC capability to (e)MMC. Summary: For new eMMC chips, we must signal controller HC capability in OP_COND command. Reviewers: imp, ian Reviewed By: ian Differential Revision: https://reviews.freebsd.org/D1920
2015-02-27 15:14:25 +00:00
mmc_send_op_cond(sc, MMC_OCR_CCS | mmcbr_get_ocr(dev), NULL);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
mmc_discover_cards(sc);
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
mmc_rescan_cards(sc);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
mmcbr_set_bus_mode(dev, pushpull);
mmcbr_update_ios(dev);
Give mmcsd driver a bit more information about card. It allows to reorganize log message in a way a bit more common for disk devices. Also it will allow mmcsd driver to use MMC/SD specific commands when needed.
2008-10-11 13:05:13 +00:00
mmc_calculate_clock(sc);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
}
static int
mmc_calculate_clock(struct mmc_softc *sc)
{
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
device_t dev;
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
struct mmc_ivars *ivar;
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
int i;
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
uint32_t dtr, max_dtr;
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
uint16_t rca;
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
enum mmc_bus_timing max_timing, timing;
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
bool changed, hs400;
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
dev = sc->dev;
max_dtr = mmcbr_get_f_max(dev);
max_timing = bus_timing_max;
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
do {
changed = false;
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
for (i = 0; i < sc->child_count; i++) {
ivar = device_get_ivars(sc->child_list[i]);
if (isclr(&ivar->timings, max_timing) ||
!mmc_host_timing(dev, max_timing)) {
for (timing = max_timing - 1; timing >=
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
bus_timing_normal; timing--) {
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
if (isset(&ivar->timings, timing) &&
mmc_host_timing(dev, timing)) {
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
max_timing = timing;
break;
}
}
changed = true;
}
dtr = mmc_timing_to_dtr(ivar, max_timing);
if (dtr < max_dtr) {
max_dtr = dtr;
changed = true;
}
}
} while (changed == true);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
if (bootverbose || mmc_debug) {
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
device_printf(dev,
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
"setting transfer rate to %d.%03dMHz (%s timing)\n",
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
max_dtr / 1000000, (max_dtr / 1000) % 1000,
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
mmc_timing_to_string(max_timing));
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
}
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
/*
* HS400 must be tuned in HS200 mode, so in case of HS400 we begin
* with HS200 following the sequence as described in "6.6.2.2 HS200
* timing mode selection" of the eMMC specification v5.1, too, and
* switch to max_timing later. HS400ES requires no tuning and, thus,
* can be switch to directly, but requires the same detour via high
* speed mode as does HS400 (see mmc_switch_to_hs400()).
*/
hs400 = max_timing == bus_timing_mmc_hs400;
timing = hs400 == true ? bus_timing_mmc_hs200 : max_timing;
for (i = 0; i < sc->child_count; i++) {
ivar = device_get_ivars(sc->child_list[i]);
o Add support for eMMC DDR bus speed mode at 52 MHz to sdhci(4) and mmc(4). For the most part, this consists of support for: - Switching the signal voltage (VCCQ) to 1.8 V or (if supported by the host controller) to 1.2 V, - setting the UHS mode as appropriate in the SDHCI_HOST_CONTROL2 register, - setting the power class in the eMMC device according to the core supply voltage (VCC), - using different bits for enabling a bus width of 4 and 8 bits in the the eMMC device at DDR or higher timings respectively, - arbitrating timings faster than high speed if there actually are additional devices on the same MMC bus. Given that support for DDR52 is not denoted by SDHCI capability registers, availability of that timing is indicated by a new quirk SDHCI_QUIRK_MMC_DDR52 and only enabled for Intel SDHCI controllers so far. Generally, what it takes for a sdhci(4) front-end to enable support for DDR52 is to hook up the bridge method mmcbr_switch_vccq (which especially for 1.2 V signaling support is chip/board specific) and the sdhci_set_uhs_timing sdhci(4) method. As a side-effect, this change also fixes communication with some eMMC devices at SDR high speed mode with 52 MHz due to the signaling voltage and UHS bits in the SDHCI controller no longer being left in an inappropriate state. Compared to 52 MHz at SDR high speed which typically yields ~45 MB/s with the eMMC chips tested, throughput goes up to ~80 MB/s at DDR52. Additionally, this change already adds infrastructure and quite some code for modes up to HS400ES and SDR104 respectively (I did not want to add to much stuff at a time, though). Essentially, what is still missing in order to be able to activate support for these latter is is support for and handling of (re-)tuning. o In sdhci(4), add two tunables hw.sdhci.quirk_clear as well as hw.sdhci.quirk_set, which (when hooked up in the front-end) allow to set/clear sdhci(4) quirks for debugging and testing purposes. However, especially for SDHCI controllers on the PCI bus which have no specific support code so far and, thus, are picked up as generic SDHCI controllers, hw.sdhci.quirk_set allows for setting the necessary quirks (if required). o In mmc(4), check and handle the return values of some more function calls instead of assuming that everything went right. In case failures actually are not problematic, indicate that by casting the return value to void. Reviewed by: jmcneill
2017-03-19 23:27:17 +00:00
if ((ivar->timings & ~(1 << bus_timing_normal)) == 0)
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
continue;
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
rca = ivar->rca;
if (mmc_select_card(sc, rca) != MMC_ERR_NONE) {
device_printf(dev, "Card at relative address %d "
"failed to select\n", rca);
continue;
}
if (timing == bus_timing_mmc_hs200 || /* includes HS400 */
timing == bus_timing_mmc_hs400es) {
if (mmc_set_vccq(sc, ivar, timing) != MMC_ERR_NONE) {
device_printf(dev, "Failed to set VCCQ for "
"card at relative address %d\n", rca);
continue;
}
}
if (timing == bus_timing_mmc_hs200) { /* includes HS400 */
/* Set bus width (required for initial tuning). */
if (mmc_set_card_bus_width(sc, ivar, timing) !=
MMC_ERR_NONE) {
device_printf(dev, "Card at relative address "
"%d failed to set bus width\n", rca);
continue;
}
mmcbr_set_bus_width(dev, ivar->bus_width);
mmcbr_update_ios(dev);
} else if (timing == bus_timing_mmc_hs400es) {
if (mmc_switch_to_hs400(sc, ivar, max_dtr, timing) !=
MMC_ERR_NONE) {
device_printf(dev, "Card at relative address "
"%d failed to set %s timing\n", rca,
mmc_timing_to_string(timing));
continue;
}
goto power_class;
}
if (mmc_set_timing(sc, ivar, timing) != MMC_ERR_NONE) {
device_printf(dev, "Card at relative address %d "
"failed to set %s timing\n", rca,
mmc_timing_to_string(timing));
continue;
}
if (timing == bus_timing_mmc_ddr52) {
/*
* Set EXT_CSD_BUS_WIDTH_n_DDR in EXT_CSD_BUS_WIDTH
* (must be done after switching to EXT_CSD_HS_TIMING).
*/
if (mmc_set_card_bus_width(sc, ivar, timing) !=
MMC_ERR_NONE) {
device_printf(dev, "Card at relative address "
"%d failed to set bus width\n", rca);
continue;
}
mmcbr_set_bus_width(dev, ivar->bus_width);
mmcbr_update_ios(dev);
if (mmc_set_vccq(sc, ivar, timing) != MMC_ERR_NONE) {
device_printf(dev, "Failed to set VCCQ for "
"card at relative address %d\n", rca);
continue;
}
}
/* Set clock (must be done before initial tuning). */
mmcbr_set_clock(dev, max_dtr);
mmcbr_update_ios(dev);
if (mmcbr_tune(dev, hs400) != 0) {
device_printf(dev, "Card at relative address %d "
"failed to execute initial tuning\n", rca);
continue;
}
if (hs400 == true && mmc_switch_to_hs400(sc, ivar, max_dtr,
max_timing) != MMC_ERR_NONE) {
device_printf(dev, "Card at relative address %d "
"failed to set %s timing\n", rca,
mmc_timing_to_string(max_timing));
continue;
}
power_class:
if (mmc_set_power_class(sc, ivar) != MMC_ERR_NONE) {
device_printf(dev, "Card at relative address %d "
"failed to set power class\n", rca);
}
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
}
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
(void)mmc_select_card(sc, 0);
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
return (max_dtr);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
}
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
/*
* Switch from HS200 to HS400 (either initially or for re-tuning) or directly
* to HS400ES. This follows the sequences described in "6.6.2.3 HS400 timing
* mode selection" of the eMMC specification v5.1.
*/
static int
mmc_switch_to_hs400(struct mmc_softc *sc, struct mmc_ivars *ivar,
uint32_t clock, enum mmc_bus_timing max_timing)
{
device_t dev;
int err;
uint16_t rca;
dev = sc->dev;
rca = ivar->rca;
/*
* Both clock and timing must be set as appropriate for high speed
* before eventually switching to HS400/HS400ES; mmc_set_timing()
* will issue mmcbr_update_ios().
*/
mmcbr_set_clock(dev, ivar->hs_tran_speed);
err = mmc_set_timing(sc, ivar, bus_timing_hs);
if (err != MMC_ERR_NONE)
return (err);
/*
* Set EXT_CSD_BUS_WIDTH_8_DDR in EXT_CSD_BUS_WIDTH (and additionally
* EXT_CSD_BUS_WIDTH_ES for HS400ES).
*/
err = mmc_set_card_bus_width(sc, ivar, max_timing);
if (err != MMC_ERR_NONE)
return (err);
mmcbr_set_bus_width(dev, ivar->bus_width);
mmcbr_update_ios(dev);
/* Finally, switch to HS400/HS400ES mode. */
err = mmc_set_timing(sc, ivar, max_timing);
if (err != MMC_ERR_NONE)
return (err);
mmcbr_set_clock(dev, clock);
mmcbr_update_ios(dev);
return (MMC_ERR_NONE);
}
/*
* Switch from HS400 to HS200 (for re-tuning).
*/
static int
mmc_switch_to_hs200(struct mmc_softc *sc, struct mmc_ivars *ivar,
uint32_t clock)
{
device_t dev;
int err;
uint16_t rca;
dev = sc->dev;
rca = ivar->rca;
/*
* Both clock and timing must initially be set as appropriate for
* DDR52 before eventually switching to HS200; mmc_set_timing()
* will issue mmcbr_update_ios().
*/
mmcbr_set_clock(dev, ivar->hs_tran_speed);
err = mmc_set_timing(sc, ivar, bus_timing_mmc_ddr52);
if (err != MMC_ERR_NONE)
return (err);
/*
* Next, switch to high speed. Thus, clear EXT_CSD_BUS_WIDTH_n_DDR
* in EXT_CSD_BUS_WIDTH and update bus width and timing in ios.
*/
err = mmc_set_card_bus_width(sc, ivar, bus_timing_hs);
if (err != MMC_ERR_NONE)
return (err);
mmcbr_set_bus_width(dev, ivar->bus_width);
mmcbr_set_timing(sc->dev, bus_timing_hs);
mmcbr_update_ios(dev);
/* Finally, switch to HS200 mode. */
err = mmc_set_timing(sc, ivar, bus_timing_mmc_hs200);
if (err != MMC_ERR_NONE)
return (err);
mmcbr_set_clock(dev, clock);
mmcbr_update_ios(dev);
return (MMC_ERR_NONE);
}
static int
mmc_retune(device_t busdev, device_t dev, bool reset)
{
struct mmc_softc *sc;
struct mmc_ivars *ivar;
int err;
uint32_t clock;
enum mmc_bus_timing timing;
if (device_get_parent(dev) != busdev)
return (MMC_ERR_INVALID);
sc = device_get_softc(busdev);
if (sc->retune_needed != 1 && sc->retune_paused != 0)
return (MMC_ERR_INVALID);
timing = mmcbr_get_timing(busdev);
if (timing == bus_timing_mmc_hs400) {
/*
* Controllers use the data strobe line to latch data from
* the devices in HS400 mode so periodic re-tuning isn't
* expected to be required, i. e. only if a CRC or tuning
* error is signaled to the bridge. In these latter cases
* we are asked to reset the tuning circuit and need to do
* the switch timing dance.
*/
if (reset == false)
return (0);
ivar = device_get_ivars(dev);
clock = mmcbr_get_clock(busdev);
if (mmc_switch_to_hs200(sc, ivar, clock) != MMC_ERR_NONE)
return (MMC_ERR_BADCRC);
}
err = mmcbr_retune(busdev, reset);
if (err != 0 && timing == bus_timing_mmc_hs400)
return (MMC_ERR_BADCRC);
switch (err) {
case 0:
break;
case EIO:
return (MMC_ERR_FAILED);
default:
return (MMC_ERR_INVALID);
}
if (timing == bus_timing_mmc_hs400) {
if (mmc_switch_to_hs400(sc, ivar, clock, timing) !=
MMC_ERR_NONE)
return (MMC_ERR_BADCRC);
}
return (MMC_ERR_NONE);
}
static void
mmc_retune_pause(device_t busdev, device_t dev, bool retune)
{
struct mmc_softc *sc;
sc = device_get_softc(busdev);
KASSERT(device_get_parent(dev) == busdev,
("%s: %s is not a child of %s", __func__, device_get_nameunit(dev),
device_get_nameunit(busdev)));
KASSERT(sc->owner != NULL,
("%s: Request from %s without bus being acquired.", __func__,
device_get_nameunit(dev)));
if (retune == true && sc->retune_paused == 0)
sc->retune_needed = 1;
sc->retune_paused++;
}
static void
mmc_retune_unpause(device_t busdev, device_t dev)
{
struct mmc_softc *sc;
sc = device_get_softc(busdev);
KASSERT(device_get_parent(dev) == busdev,
("%s: %s is not a child of %s", __func__, device_get_nameunit(dev),
device_get_nameunit(busdev)));
KASSERT(sc->owner != NULL,
("%s: Request from %s without bus being acquired.", __func__,
device_get_nameunit(dev)));
KASSERT(sc->retune_paused != 0,
("%s: Re-tune pause count already at 0", __func__));
sc->retune_paused--;
}
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
static void
mmc_scan(struct mmc_softc *sc)
{
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
device_t dev = sc->dev;
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
int err;
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
err = mmc_acquire_bus(dev, dev);
if (err != 0) {
device_printf(dev, "Failed to acquire bus for scanning\n");
return;
}
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
mmc_go_discovery(sc);
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
err = mmc_release_bus(dev, dev);
if (err != 0) {
device_printf(dev, "Failed to release bus after scanning\n");
return;
}
(void)bus_generic_attach(dev);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
}
static int
read_ivar takes a uintptr_t * not a u_char *.
2009-03-12 06:36:44 +00:00
mmc_read_ivar(device_t bus, device_t child, int which, uintptr_t *result)
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
{
struct mmc_ivars *ivar = device_get_ivars(child);
switch (which) {
default:
return (EINVAL);
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
case MMC_IVAR_SPEC_VERS:
*result = ivar->csd.spec_vers;
break;
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
case MMC_IVAR_DSR_IMP:
Fix read_ivar implementation for MMC and SD. 1. Both mmc_read_ivar() and sdhci_read_ivar() use the expression '*(int *)result = val' to assign to result which is uintptr_t *. This does not work on big-endian 64 bit systems. 2. The media_size ivar is declared as 'off_t' which does not fit into uintptr_t in 32bit systems, change this to long. Submitted by: kanthms at netlogicmicro com (initial version)
2011-05-30 06:23:51 +00:00
*result = ivar->csd.dsr_imp;
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
break;
case MMC_IVAR_MEDIA_SIZE:
Fix read_ivar implementation for MMC and SD. 1. Both mmc_read_ivar() and sdhci_read_ivar() use the expression '*(int *)result = val' to assign to result which is uintptr_t *. This does not work on big-endian 64 bit systems. 2. The media_size ivar is declared as 'off_t' which does not fit into uintptr_t in 32bit systems, change this to long. Submitted by: kanthms at netlogicmicro com (initial version)
2011-05-30 06:23:51 +00:00
*result = ivar->sec_count;
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
break;
case MMC_IVAR_RCA:
Fix read_ivar implementation for MMC and SD. 1. Both mmc_read_ivar() and sdhci_read_ivar() use the expression '*(int *)result = val' to assign to result which is uintptr_t *. This does not work on big-endian 64 bit systems. 2. The media_size ivar is declared as 'off_t' which does not fit into uintptr_t in 32bit systems, change this to long. Submitted by: kanthms at netlogicmicro com (initial version)
2011-05-30 06:23:51 +00:00
*result = ivar->rca;
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
break;
case MMC_IVAR_SECTOR_SIZE:
Fix read_ivar implementation for MMC and SD. 1. Both mmc_read_ivar() and sdhci_read_ivar() use the expression '*(int *)result = val' to assign to result which is uintptr_t *. This does not work on big-endian 64 bit systems. 2. The media_size ivar is declared as 'off_t' which does not fit into uintptr_t in 32bit systems, change this to long. Submitted by: kanthms at netlogicmicro com (initial version)
2011-05-30 06:23:51 +00:00
*result = MMC_SECTOR_SIZE;
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
break;
case MMC_IVAR_TRAN_SPEED:
Fix read_ivar implementation for MMC and SD. 1. Both mmc_read_ivar() and sdhci_read_ivar() use the expression '*(int *)result = val' to assign to result which is uintptr_t *. This does not work on big-endian 64 bit systems. 2. The media_size ivar is declared as 'off_t' which does not fit into uintptr_t in 32bit systems, change this to long. Submitted by: kanthms at netlogicmicro com (initial version)
2011-05-30 06:23:51 +00:00
*result = mmcbr_get_clock(bus);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
break;
Propigate read-only status of cards. Right now it is read only at device attach time. We may need to read this more often in the future, but for now simplicity of implementation wins. Submitted by: mav@
2008-09-28 22:40:11 +00:00
case MMC_IVAR_READ_ONLY:
Fix read_ivar implementation for MMC and SD. 1. Both mmc_read_ivar() and sdhci_read_ivar() use the expression '*(int *)result = val' to assign to result which is uintptr_t *. This does not work on big-endian 64 bit systems. 2. The media_size ivar is declared as 'off_t' which does not fit into uintptr_t in 32bit systems, change this to long. Submitted by: kanthms at netlogicmicro com (initial version)
2011-05-30 06:23:51 +00:00
*result = ivar->read_only;
Propigate read-only status of cards. Right now it is read only at device attach time. We may need to read this more often in the future, but for now simplicity of implementation wins. Submitted by: mav@
2008-09-28 22:40:11 +00:00
break;
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
case MMC_IVAR_HIGH_CAP:
Fix read_ivar implementation for MMC and SD. 1. Both mmc_read_ivar() and sdhci_read_ivar() use the expression '*(int *)result = val' to assign to result which is uintptr_t *. This does not work on big-endian 64 bit systems. 2. The media_size ivar is declared as 'off_t' which does not fit into uintptr_t in 32bit systems, change this to long. Submitted by: kanthms at netlogicmicro com (initial version)
2011-05-30 06:23:51 +00:00
*result = ivar->high_cap;
Set of mmc layer improvements: - add MMC support. - add SDHC support. - add 4 and 8 bit bus width support. - add High Speed bus timing support.
2008-10-08 17:35:41 +00:00
break;
Give mmcsd driver a bit more information about card. It allows to reorganize log message in a way a bit more common for disk devices. Also it will allow mmcsd driver to use MMC/SD specific commands when needed.
2008-10-11 13:05:13 +00:00
case MMC_IVAR_CARD_TYPE:
Fix read_ivar implementation for MMC and SD. 1. Both mmc_read_ivar() and sdhci_read_ivar() use the expression '*(int *)result = val' to assign to result which is uintptr_t *. This does not work on big-endian 64 bit systems. 2. The media_size ivar is declared as 'off_t' which does not fit into uintptr_t in 32bit systems, change this to long. Submitted by: kanthms at netlogicmicro com (initial version)
2011-05-30 06:23:51 +00:00
*result = ivar->mode;
Give mmcsd driver a bit more information about card. It allows to reorganize log message in a way a bit more common for disk devices. Also it will allow mmcsd driver to use MMC/SD specific commands when needed.
2008-10-11 13:05:13 +00:00
break;
case MMC_IVAR_BUS_WIDTH:
Fix read_ivar implementation for MMC and SD. 1. Both mmc_read_ivar() and sdhci_read_ivar() use the expression '*(int *)result = val' to assign to result which is uintptr_t *. This does not work on big-endian 64 bit systems. 2. The media_size ivar is declared as 'off_t' which does not fit into uintptr_t in 32bit systems, change this to long. Submitted by: kanthms at netlogicmicro com (initial version)
2011-05-30 06:23:51 +00:00
*result = ivar->bus_width;
Give mmcsd driver a bit more information about card. It allows to reorganize log message in a way a bit more common for disk devices. Also it will allow mmcsd driver to use MMC/SD specific commands when needed.
2008-10-11 13:05:13 +00:00
break;
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
case MMC_IVAR_ERASE_SECTOR:
Fix read_ivar implementation for MMC and SD. 1. Both mmc_read_ivar() and sdhci_read_ivar() use the expression '*(int *)result = val' to assign to result which is uintptr_t *. This does not work on big-endian 64 bit systems. 2. The media_size ivar is declared as 'off_t' which does not fit into uintptr_t in 32bit systems, change this to long. Submitted by: kanthms at netlogicmicro com (initial version)
2011-05-30 06:23:51 +00:00
*result = ivar->erase_sector;
Implement BIO_DELETE command with MMC and SD erase commands. Erase operation gives card's logic information about unused areas to help it implement wear-leveling with lower overhead comparing to usual writing. Erase is much faster then write and does not depends on data bus speed. Also as result of hitting in-card write logic optimizations I have measured up to 50% performance boost on writing undersized blocks into preerased areas. At the same time there are strict limitations on size and allignment of erase operations. We can erase only blocks aligned to the erase sector size and with size multiple of it. Different cards has different erase sector size which usually varies from 64KB to 4MB. SD cards actually allow to erase smaller blocks, but it is much more expensive as it is implemented via read-erase-write sequence and so not sutable for the BIO_DELETE purposes. Reviewed by: imp@
2008-10-18 16:17:04 +00:00
break;
Allow card reader bridge driver to report maximum supported transfer size. sdhci supports up to 65535 blocks transfers, at91_mci - one block. Enable multiblock operations disabled before to follow at91_mci driver limitations. Reviewed by: imp@
2008-10-29 20:01:26 +00:00
case MMC_IVAR_MAX_DATA:
Fix read_ivar implementation for MMC and SD. 1. Both mmc_read_ivar() and sdhci_read_ivar() use the expression '*(int *)result = val' to assign to result which is uintptr_t *. This does not work on big-endian 64 bit systems. 2. The media_size ivar is declared as 'off_t' which does not fit into uintptr_t in 32bit systems, change this to long. Submitted by: kanthms at netlogicmicro com (initial version)
2011-05-30 06:23:51 +00:00
*result = mmcbr_get_max_data(bus);
Allow card reader bridge driver to report maximum supported transfer size. sdhci supports up to 65535 blocks transfers, at91_mci - one block. Enable multiblock operations disabled before to follow at91_mci driver limitations. Reviewed by: imp@
2008-10-29 20:01:26 +00:00
break;
- If available, use TRIM instead of ERASE for implementing BIO_DELETE. This also involves adding a quirk table as TRIM is broken for some Kingston eMMC devices, though. Compared to ERASE (declared "legacy" in the eMMC specification v5.1), TRIM has the advantage of operating on write sectors rather than on erase sectors, which typically are of a much larger size. Thus, employing TRIM, we don't need to fiddle with coalescing BIO_DELETE requests that are also of (write) sector units into erase sectors, which might not even add up in all cases. - For some SanDisk iNAND devices, the CMD38 argument, e. g. ERASE, TRIM etc., has to be specified via EXT_CSD[113], which now is also handled via a quirk. - My initial understanding was that for eMMC partitions, the granularity should be used as erase sector size, e. g. 128 KB for boot partitions. However, rereading the relevant parts of the eMMC specification v5.1, this isn't actually correct. So drop the code which used partition granularities for delmaxsize and stripesize. For the most part, this change is a NOP, though, because a) for ERASE, mmcsd_delete() used the erase sector size unconditionally for all partitions anyway and b) g_disk_limit() doesn't actually take the stripesize into account. - Take some more advantage of mmcsd_errmsg() in mmcsd(4) for making error codes human readable.
2017-08-07 23:33:05 +00:00
case MMC_IVAR_CMD6_TIMEOUT:
*result = ivar->cmd6_time;
break;
case MMC_IVAR_QUIRKS:
*result = ivar->quirks;
break;
o Fixes: - When switching to 4-bit operation, send a SET_CLR_CARD_DETECT command to disconnect the card-detect pull-up resistor from the DAT3 line before sending the SET_BUS_WIDTH command. - Add the missing "reserved" zero entry to the mantissa table used to decode various CSD fields. This was causing SD cards to report that they could run at 30 MHz instead of the maximum 25 MHz mandated in the spec. o Enhancements: - At the MMC layer, format various info from the CID into a string that uniquely identifies the card instance (manufacturer number, serial number, product name and revision, etc). Export it as an instance variable. - At the MMCSD layer, display the formatted card ID string, and also report the clock speed of the hardware (not the card's max speed), and the number of bits and number of blocks per transfer. It comes out like this now: mmcsd0: 968MB <SD SD01G 8.0 SN 276886905 MFG 08/2008 by 3 SD> at mmc0 22.5MHz/4bit/128-block o Use DEVMETHOD_END. o Use NULL instead of 0 for pointers. PR: 156496 Submitted by: Ian Lepore MFC after: 1 week
2012-04-21 01:51:16 +00:00
case MMC_IVAR_CARD_ID_STRING:
*(char **)result = ivar->card_id_string;
break;
Export an mmc or sd card's serial number from the mmc layer as an ivar. In the mmcsd layer use this value to populate disk->d_ident. Also set disk->d_descr to the full set of card identification info (includes vendor, model, manufacturing date, etc).
2014-07-31 16:54:54 +00:00
case MMC_IVAR_CARD_SN_STRING:
*(char **)result = ivar->card_sn_string;
break;
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
}
return (0);
}
static int
mmc_write_ivar(device_t bus, device_t child, int which, uintptr_t value)
{
Fix overly long lines, whitespace and other bugs according to style(9).
2017-01-29 00:05:49 +00:00
Convert all C++ comments to C style. There's lots of folks that don't like them and they don't add enough to be worth standing out.
2008-09-29 01:32:21 +00:00
/*
* None are writable ATM
*/
return (EINVAL);
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
}
static void
mmc_delayed_attach(void *xsc)
{
struct mmc_softc *sc = xsc;
sdhci/mmc: Minor whitespace cleanups No functional change. Submitted by: Johannes Lundberg <yohanesu75 at gmail.com>
2016-12-20 03:38:14 +00:00
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
mmc_scan(sc);
config_intrhook_disestablish(&sc->config_intrhook);
}
Report relative card address to NewBus as location string.
2010-05-23 09:44:48 +00:00
static int
mmc_child_location_str(device_t dev, device_t child, char *buf,
size_t buflen)
{
snprintf(buf, buflen, "rca=0x%04x", mmc_get_rca(child));
return (0);
}
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
static device_method_t mmc_methods[] = {
/* device_if */
DEVMETHOD(device_probe, mmc_probe),
DEVMETHOD(device_attach, mmc_attach),
DEVMETHOD(device_detach, mmc_detach),
Implement suspend/resume for mmc and mmcsd drivers. Now it is possible to suspend/resume with inserted and active card. To reinitialize card on resume and to detect card change while suspended, implement bus rescan routines. It can also be used by controllers without card presence detection signals or with multiple cards per slot support. While there, cleanup msleep() usage. We have no any rights to exit without "request done" signal from driver as it could lead to modify after free.
2008-12-06 21:41:27 +00:00
DEVMETHOD(device_suspend, mmc_suspend),
DEVMETHOD(device_resume, mmc_resume),
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
/* Bus interface */
DEVMETHOD(bus_read_ivar, mmc_read_ivar),
DEVMETHOD(bus_write_ivar, mmc_write_ivar),
Report relative card address to NewBus as location string.
2010-05-23 09:44:48 +00:00
DEVMETHOD(bus_child_location_str, mmc_child_location_str),
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
/* MMC Bus interface */
o Add support for eMMC HS200 and HS400 bus speed modes at 200 MHz to sdhci(4), mmc(4) and mmcsd(4). For the most part, this consists of: - Correcting and extending the infrastructure for negotiating and enabling post-DDR52 modes already added as part of r315598. In fact, HS400ES now should work as well but hasn't been activated due to lack of corresponding hardware. - Adding support executing standard SDHCI initial tuning as well as re-tuning as required for eMMC HS200/HS400 and the fast UHS-I SD card modes. Currently, corresponding methods are only hooked up to the ACPI and PCI front-ends of sdhci(4), though. Moreover, sdhci(4) won't offer any modes requiring (re-)tuning to the MMC/SD layer in order to not break operations with other sdhci(4) front- ends. Likewise, sdhci(4) now no longer offers modes requiring the set_uhs_timing method introduced in r315598 to be implemented/ hooked up (previously, this method was used with DDR52 only, which in turn is only available with Intel controllers so far, i. e. no such limitation was necessary before). Similarly for 1.2/1.8 V VCCQ support and the switch_vccq method. - Addition of locking to the IOCTL half of mmcsd(4) to prevent races with detachment and suspension, especially since it's required to immediately switch away from RPMB partitions again after an access to these (so re-tuning can take place anew, given that the current eMMC specification v5.1 doesn't allow tuning commands to be issued with a RPMB partition selected). Therefore, the existing part_mtx lock in the mmcsd(4) softc is additionally renamed to disk_mtx in order to denote that it only refers to the disk(9) half, likewise for corresponding macros. On the system where the addition of DDR52 support increased the read throughput to ~80 MB/s (from ~45 MB/s at high speed), HS200 yields ~154 MB/s and HS400 ~187 MB/s, i. e. performance now has more than quadrupled compared to pre-r315598. Also, with the advent of (re-)tuning support, most infrastructure necessary for SD card UHS-I modes up to SDR104 now is also in place. Note, though, that the standard SDHCI way of (re-)tuning is special in several ways, which also is why sending the actual tuning requests to the device is part of sdhci(4). SDHCI implementations not following the specification, MMC and non-SDHCI SD card controllers likely will use a generic implementation in the MMC/SD layer for executing tuning, which hasn't been written so far, though. However, in fact this isn't a feature-only change; there are boards based on Intel Bay Trail where DDR52 is problematic and the suggested workaround is to use HS200 mode instead. So far exact details are unknown, however, i. e. whether that's due to a defect in these SoCs or on the boards. Moreover, due to the above changes requiring to be aware of possible MMC siblings in the fast path of mmc(4), corresponding information now is cached in mmc_softc. As a side-effect, mmc_calculate_clock(), mmc_delete_cards(), mmc_discover_cards() and mmc_rescan_cards() now all are guaranteed to operate on the same set of devices as there no longer is any use of device_get_children(9), which can fail in low memory situations. Likewise, mmc_calculate_clock() now longer will trigger a panic due to the latter. o Fix a bug in the failure reporting of mmcsd_delete(); in case of an error when the starting block of a previously stored erase request is used (in order to be able to erase a full erase sector worth of data), the starting block of the newly supplied bio_pblkno has to be returned for indicating no progress. Otherwise, upper layers might be told that a negative number of BIOs have been completed, leading to a panic. o Fix 2 bugs on resume: - Things done in fork1(9) like the acquisition of an SX lock or the sleepable memory allocation are incompatible with a MTX_DEF taken. Thus, mmcsd_resume() must not call kproc_create(9), which in turn uses fork1(9), with the disk_mtx (formerly part_mtx) held. - In mmc_suspend(), the bus is powered down, which in the typical case of a device being selected at the time of suspension, causes the device deselection as part of the bus acquisition by mmc(4) in mmc_scan() to fail as the bus isn't powered up again before later in mmc_go_discovery(). Thus, power down with the bus acquired in mmc_suspend(), which will trigger the deselection up-front. o Fix a memory leak in mmcsd_ioctl() in case copyin(9) fails. [1] o Fix missing variable initialization in mmc_switch_status(). [2] o Fix R1_SWITCH_ERROR detection in mmc_switch_status(). [3] o Handle the case of device_add_child(9) failing, for example due to a memory shortage, gracefully in mmc(4) and sdhci(4), including not leaking memory for the instance variables in case of mmc(4) (which might or might not fix [4] as the latter problem has been discovered independently). o Handle the case of an unknown SD CSD version in mmc_decode_csd_sd() gracefully instead of calling panic(9). o Again, check and handle the return values of some additional function calls in mmc(4) instead of assuming that everything went right or mark non-fatal errors by casting the return value to void. o Correct a typo in the Linux IOCTL compatibility; it should have been MMC_IOC_MULTI_CMD rather than MMC_IOC_CMD_MULTI. o Now that we are reaching ever faster speeds (more improvement in this regard is to be expected when adding ADMA support to sdhci(4)), apply a few micro-optimizations like predicting mmc(4) and sdhci(4) debugging to be off or caching erase sector and maximum data sizes as well support of block addressing in mmsd(4) (instead of doing 2 indirections on every read/write request for determining the maximum data size for example). Reported by: Coverity CID: 1372612 [1], 1372624 [2], 1372594 [3], 1007069 [4]
2017-07-23 16:11:47 +00:00
DEVMETHOD(mmcbus_retune_pause, mmc_retune_pause),
DEVMETHOD(mmcbus_retune_unpause, mmc_retune_unpause),
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
DEVMETHOD(mmcbus_wait_for_request, mmc_wait_for_request),
DEVMETHOD(mmcbus_acquire_bus, mmc_acquire_bus),
DEVMETHOD(mmcbus_release_bus, mmc_release_bus),
o Fixes: - When switching to 4-bit operation, send a SET_CLR_CARD_DETECT command to disconnect the card-detect pull-up resistor from the DAT3 line before sending the SET_BUS_WIDTH command. - Add the missing "reserved" zero entry to the mantissa table used to decode various CSD fields. This was causing SD cards to report that they could run at 30 MHz instead of the maximum 25 MHz mandated in the spec. o Enhancements: - At the MMC layer, format various info from the CID into a string that uniquely identifies the card instance (manufacturer number, serial number, product name and revision, etc). Export it as an instance variable. - At the MMCSD layer, display the formatted card ID string, and also report the clock speed of the hardware (not the card's max speed), and the number of bits and number of blocks per transfer. It comes out like this now: mmcsd0: 968MB <SD SD01G 8.0 SN 276886905 MFG 08/2008 by 3 SD> at mmc0 22.5MHz/4bit/128-block o Use DEVMETHOD_END. o Use NULL instead of 0 for pointers. PR: 156496 Submitted by: Ian Lepore MFC after: 1 week
2012-04-21 01:51:16 +00:00
DEVMETHOD_END
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
};
Move the DRIVER_MODULE() statements that declare mmc(4) to be a child of the various bridge drivers out of dev/mmc.c and into the bridge drivers. Requested by: jhb (almost two years ago; better late than never)
2015-12-14 01:09:25 +00:00
driver_t mmc_driver = {
Preliminary MMC stack. This stack supports SD 1.0 cards only, but should be easily adapted to SD 2.0 (aka SDHC), SDIO, MMC and MMCplus cards. At the present time, there's only one bridge driver for the ARM9 based Atmel AT91RM9200.
2006-10-20 06:39:59 +00:00
"mmc",
mmc_methods,
sizeof(struct mmc_softc),
};
Move the DRIVER_MODULE() statements that declare mmc(4) to be a child of the various bridge drivers out of dev/mmc.c and into the bridge drivers. Requested by: jhb (almost two years ago; better late than never)
2015-12-14 01:09:25 +00:00
devclass_t mmc_devclass;
Fix fallout from r292180 (Dec 2015)... ensure that every driver which has a DRIVER_MODULE() referencing mmc_driver has a MODULE_DEPEND() on mmc. This is because the kernel linker only searches for symbols in dependent modules, so loading sdhci_pci (and other bus-flavors of sdhci) would fail when mmc was not compiled into the kernel (even if you hand-loaded mmc first). (Thanks to jilles@ for providing the vital clue about the kernel linker.)
2016-03-21 00:52:24 +00:00
- Add support for eMMC "partitions". Besides the user data area, i. e. the default partition, eMMC v4.41 and later devices can additionally provide up to: 1 enhanced user data area partition 2 boot partitions 1 RPMB (Replay Protected Memory Block) partition 4 general purpose partitions (optionally with a enhanced or extended attribute) Of these "partitions", only the enhanced user data area one actually slices the user data area partition and, thus, gets handled with the help of geom_flashmap(4). The other types of partitions have address space independent from the default partition and need to be switched to via CMD6 (SWITCH), i. e. constitute a set of additional "disks". The second kind of these "partitions" doesn't fit that well into the design of mmc(4) and mmcsd(4). I've decided to let mmcsd(4) hook all of these "partitions" up as disk(9)'s (except for the RPMB partition as it didn't seem to make much sense to be able to put a file-system there and may require authentication; therefore, RPMB partitions are solely accessible via the newly added IOCTL interface currently; see also below). This approach for one resulted in cleaner code. Second, it retains the notion of mmcsd(4) children corresponding to a single physical device each. With the addition of some layering violations, it also would have been possible for mmc(4) to add separate mmcsd(4) instances with one disk each for all of these "partitions", however. Still, both mmc(4) and mmcsd(4) share some common code now e. g. for issuing CMD6, which has been factored out into mmc_subr.c. Besides simply subdividing eMMC devices, some Intel NUCs having UEFI code in the boot partitions etc., another use case for the partition support is the activation of pseudo-SLC mode, which manufacturers of eMMC chips typically associate with the enhanced user data area and/ or the enhanced attribute of general purpose partitions. CAVEAT EMPTOR: Partitioning eMMC devices is a one-time operation. - Now that properly issuing CMD6 is crucial (so data isn't written to the wrong partition for example), make a step into the direction of correctly handling the timeout for these commands in the MMC layer. Also, do a SEND_STATUS when CMD6 is invoked with an R1B response as recommended by relevant specifications. However, quite some work is left to be done in this regard; all other R1B-type commands done by the MMC layer also should be followed by a SEND_STATUS (CMD13), the erase timeout calculations/handling as documented in specifications are entirely ignored so far, the MMC layer doesn't provide timeouts applicable up to the bridge drivers and at least sdhci(4) currently is hardcoding 1 s as timeout for all command types unconditionally. Let alone already available return codes often not being checked in the MMC layer ... - Add an IOCTL interface to mmcsd(4); this is sufficiently compatible with Linux so that the GNU mmc-utils can be ported to and used with FreeBSD (note that due to the remaining deficiencies outlined above SANITIZE operations issued by/with `mmc` currently most likely will fail). These latter will be added to ports as sysutils/mmc-utils in a bit. Among others, the `mmc` tool of the GNU mmc-utils allows for partitioning eMMC devices (tested working). - For devices following the eMMC specification v4.41 or later, year 0 is 2013 rather than 1997; so correct this for assembling the device ID string properly. - Let mmcsd.ko depend on mmc.ko. Additionally, bump MMC_VERSION as at least for some of the above a matching pair is required. - In the ACPI front-end of sdhci(4) describe the Intel eMMC and SDXC controllers as such in order to match the PCI one. Additionally, in the entry for the 80860F14 SDXC controller remove the eMMC-only SDHCI_QUIRK_INTEL_POWER_UP_RESET. OKed by: imp Submitted by: ian (mmc_switch_status() implementation)
2017-03-16 22:23:04 +00:00
MODULE_VERSION(mmc, MMC_VERSION);
Reference in New Issue Copy Permalink