d/freebsd-skq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
92dc84876a
freebsd-skq/sys/dev/ata/ata-chipset.c

5139 lines
159 KiB
C
Raw Normal View History

First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
/*-
Get rid of the advertising clause in the copyright.
2006-01-05 21:27:19 +00:00
* Copyright (c) 1998 - 2006 S<EFBFBD>ren Schmidt <sos@FreeBSD.org>
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
* All rights reserved.
*
* 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,
* without modification, immediately at the beginning of the file.
* 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.
*/
Use __FBSDID(). Also some minor style cleanups.
2003-08-24 17:55:58 +00:00
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
#include "opt_ata.h"
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/kernel.h>
#include <sys/ata.h>
#include <sys/bus.h>
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
#include <sys/endian.h>
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
#include <sys/malloc.h>
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
#include <sys/lock.h>
#include <sys/mutex.h>
Overhaul of the timeout/reinit framework. This should clear up most of the leftovers from the old version that really doesn't work anymore. Add a reset function for host-end of the ATA channel. This is needed for the SiI3112 in order to whack it back to reality if a device locks up the SATA interface (thereby preventing that we can reset the device). The result is that ATA now recovers from the timeouts that happens with the SiI3112A and more or less all disks based on old PATA electronics with a Marvell PATA->SATA converter. This includes lots of the popular SATA dongles and the WDC Raptor disks..
2004-01-11 22:08:34 +00:00
#include <sys/sema.h>
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
#include <sys/taskqueue.h>
Use UMA instead of plain malloc for getting ATA request storage. This gives +10% performance on simple tests, so definitly worth it. A few percent more could be had by not using M_ZERO'd alloc's, but we then need to clear fields all over the place to be safe, and that was deemed not worth the trouble (and it makes life dangerous).
2004-01-14 21:26:35 +00:00
#include <vm/uma.h>
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
#include <machine/stdarg.h>
#include <machine/resource.h>
#include <machine/bus.h>
#include <sys/rman.h>
Prefer new location of pci include files (which have only been in the tree for two or more years now), except in a few places where there's code to be compatible with older versions of FreeBSD.
2003-08-22 05:54:52 +00:00
#include <dev/pci/pcivar.h>
#include <dev/pci/pcireg.h>
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
#include <dev/ata/ata-all.h>
#include <dev/ata/ata-pci.h>
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
#include <ata_if.h>
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
/* local prototypes */
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
/* ata-chipset.c */
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
static int ata_generic_chipinit(device_t dev);
static void ata_generic_intr(void *data);
static void ata_generic_setmode(device_t dev, int mode);
static void ata_sata_phy_enable(struct ata_channel *ch);
static void ata_sata_phy_event(void *context, int dummy);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static int ata_sata_connect(struct ata_channel *ch);
static void ata_sata_setmode(device_t dev, int mode);
Update AHCI support to be more generic. Add support for AHCI on the VIA VT8251.
2006-06-28 09:59:09 +00:00
static int ata_ahci_chipinit(device_t dev);
Add support for AHCI compliant ATA devices. For now just support the Intel ICH6 as that the HW at hand. Sponsored by: pair.com
2005-05-11 16:10:08 +00:00
static int ata_ahci_allocate(device_t dev);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static int ata_ahci_status(device_t dev);
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
static int ata_ahci_begin_transaction(struct ata_request *request);
static int ata_ahci_end_transaction(struct ata_request *request);
static void ata_ahci_reset(device_t dev);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static void ata_ahci_dmasetprd(void *xsc, bus_dma_segment_t *segs, int nsegs, int error);
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
static void ata_ahci_dmainit(device_t dev);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static int ata_ahci_setup_fis(u_int8_t *fis, struct ata_request *request);
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
static int ata_acard_chipinit(device_t dev);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static int ata_acard_allocate(device_t dev);
static int ata_acard_status(device_t dev);
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
static void ata_acard_850_setmode(device_t dev, int mode);
static void ata_acard_86X_setmode(device_t dev, int mode);
static int ata_ali_chipinit(device_t dev);
static int ata_ali_allocate(device_t dev);
Add support for working around controllers that cannot do DMA in 48bit mode. The workaround use PIO mode above ~137GB to allow using the disk. Add the Acer chips with rev < 0xc4 as first candidate.
2005-08-17 15:00:33 +00:00
static int ata_ali_sata_allocate(device_t dev);
Apply fix for "pr82261 DMA-support on Sparc64 broken" The Acer chip or wiring that SUN uses has problems that this patch tries to work around. Original patch by Marius Strobl, hacked into shape by me..
2005-08-23 08:53:01 +00:00
static void ata_ali_reset(device_t dev);
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
static void ata_ali_setmode(device_t dev, int mode);
static int ata_amd_chipinit(device_t dev);
Add support for the ATI IXP[234]00 series chipsets. HW donated by: sentex
2005-10-12 20:00:26 +00:00
static int ata_ati_chipinit(device_t dev);
static void ata_ati_setmode(device_t dev, int mode);
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
static int ata_cyrix_chipinit(device_t dev);
static void ata_cyrix_setmode(device_t dev, int mode);
static int ata_cypress_chipinit(device_t dev);
static void ata_cypress_setmode(device_t dev, int mode);
static int ata_highpoint_chipinit(device_t dev);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static int ata_highpoint_allocate(device_t dev);
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
static void ata_highpoint_setmode(device_t dev, int mode);
static int ata_highpoint_check_80pin(device_t dev, int mode);
static int ata_intel_chipinit(device_t dev);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static int ata_intel_allocate(device_t dev);
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
static void ata_intel_reset(device_t dev);
static void ata_intel_old_setmode(device_t dev, int mode);
static void ata_intel_new_setmode(device_t dev, int mode);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static int ata_intel_31244_allocate(device_t dev);
static int ata_intel_31244_status(device_t dev);
static int ata_intel_31244_command(struct ata_request *request);
static void ata_intel_31244_reset(device_t dev);
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
static int ata_ite_chipinit(device_t dev);
static void ata_ite_setmode(device_t dev, int mode);
Add support for the JMicron JMB360 SATAII controller. Thanks to JMicron for providing needed info. HW donated by: Ralf Folkerts
2006-01-25 23:07:42 +00:00
static int ata_jmicron_chipinit(device_t dev);
Add support for the JMicron JMB363 dual SATA + single PATA controller. Documentation and HW kindly provided by JMicron.
2006-02-16 17:09:24 +00:00
static int ata_jmicron_allocate(device_t dev);
static void ata_jmicron_reset(device_t dev);
static void ata_jmicron_dmainit(device_t dev);
static void ata_jmicron_setmode(device_t dev, int mode);
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
static int ata_marvell_chipinit(device_t dev);
static int ata_marvell_allocate(device_t dev);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static int ata_marvell_status(device_t dev);
static int ata_marvell_begin_transaction(struct ata_request *request);
static int ata_marvell_end_transaction(struct ata_request *request);
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
static void ata_marvell_reset(device_t dev);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static void ata_marvell_dmasetprd(void *xsc, bus_dma_segment_t *segs, int nsegs, int error);
static void ata_marvell_dmainit(device_t dev);
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
static int ata_national_chipinit(device_t dev);
static void ata_national_setmode(device_t dev, int mode);
static int ata_nvidia_chipinit(device_t dev);
static int ata_nvidia_allocate(device_t dev);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static int ata_nvidia_status(device_t dev);
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
static void ata_nvidia_reset(device_t dev);
static int ata_promise_chipinit(device_t dev);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static int ata_promise_allocate(device_t dev);
static int ata_promise_status(device_t dev);
static int ata_promise_dmastart(device_t dev);
static int ata_promise_dmastop(device_t dev);
static void ata_promise_dmareset(device_t dev);
static void ata_promise_dmainit(device_t dev);
static void ata_promise_setmode(device_t dev, int mode);
static int ata_promise_tx2_allocate(device_t dev);
static int ata_promise_tx2_status(device_t dev);
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
static int ata_promise_mio_allocate(device_t dev);
static void ata_promise_mio_intr(void *data);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static int ata_promise_mio_status(device_t dev);
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
static int ata_promise_mio_command(struct ata_request *request);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static void ata_promise_mio_reset(device_t dev);
static void ata_promise_mio_dmainit(device_t dev);
static void ata_promise_mio_setmode(device_t dev, int mode);
static void ata_promise_sx4_intr(void *data);
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
static int ata_promise_sx4_command(struct ata_request *request);
static int ata_promise_apkt(u_int8_t *bytep, struct ata_request *request);
static void ata_promise_queue_hpkt(struct ata_pci_controller *ctlr, u_int32_t hpkt);
static void ata_promise_next_hpkt(struct ata_pci_controller *ctlr);
static int ata_serverworks_chipinit(device_t dev);
Add preliminary support for the Serverworks HT1000 chip. HW sponsored by: Yahoo!
2006-07-04 20:36:03 +00:00
static int ata_serverworks_allocate(device_t dev);
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
static void ata_serverworks_setmode(device_t dev, int mode);
static int ata_sii_chipinit(device_t dev);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static int ata_cmd_allocate(device_t dev);
static int ata_cmd_status(device_t dev);
static void ata_cmd_setmode(device_t dev, int mode);
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
static int ata_sii_allocate(device_t dev);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static int ata_sii_status(device_t dev);
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
static void ata_sii_reset(device_t dev);
static void ata_sii_setmode(device_t dev, int mode);
static int ata_sis_chipinit(device_t dev);
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
static int ata_sis_allocate(device_t dev);
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
static void ata_sis_reset(device_t dev);
static void ata_sis_setmode(device_t dev, int mode);
static int ata_via_chipinit(device_t dev);
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
static int ata_via_allocate(device_t dev);
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
static void ata_via_reset(device_t dev);
static void ata_via_southbridge_fixup(device_t dev);
static void ata_via_family_setmode(device_t dev, int mode);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static struct ata_chip_id *ata_match_chip(device_t dev, struct ata_chip_id *index);
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
static struct ata_chip_id *ata_find_chip(device_t dev, struct ata_chip_id *index, int slot);
static int ata_setup_interrupt(device_t dev);
static int ata_serialize(device_t dev, int flags);
static void ata_print_cable(device_t dev, u_int8_t *who);
static int ata_atapi(device_t dev);
static int ata_check_80pin(device_t dev, int mode);
static int ata_mode2idx(int mode);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
/*
* generic ATA support functions
*/
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
int
ata_generic_ident(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
device_set_desc(dev, "GENERIC ATA controller");
ctlr->chipinit = ata_generic_chipinit;
return 0;
}
static int
ata_generic_chipinit(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
if (ata_setup_interrupt(dev))
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
return ENXIO;
ctlr->setmode = ata_generic_setmode;
return 0;
}
static void
ata_generic_intr(void *data)
{
struct ata_pci_controller *ctlr = data;
struct ata_channel *ch;
int unit;
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
for (unit = 0; unit < ctlr->channels; unit++) {
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
if ((ch = ctlr->interrupt[unit].argument))
ctlr->interrupt[unit].function(ch);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
}
static void
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_generic_setmode(device_t dev, int mode)
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
{
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
struct ata_device *atadev = device_get_softc(dev);
mode = ata_limit_mode(dev, mode, ATA_UDMA2);
mode = ata_check_80pin(dev, mode);
if (!ata_controlcmd(dev, ATA_SETFEATURES, ATA_SF_SETXFER, 0, mode))
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
atadev->mode = mode;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
/*
* SATA support functions
*/
Add support for the VIA 8237 (both PATA and SATA part). Cleanup the SATA support a bit now we are here anyways.
2003-10-30 13:16:21 +00:00
static void
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ata_sata_phy_enable(struct ata_channel *ch)
Add support for the VIA 8237 (both PATA and SATA part). Cleanup the SATA support a bit now we are here anyways.
2003-10-30 13:16:21 +00:00
{
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
int loop, retry;
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
if ((ATA_IDX_INL(ch, ATA_SCONTROL) & ATA_SC_DET_MASK) == ATA_SC_DET_IDLE) {
ata_sata_connect(ch);
return;
Cleanup the storing and printing of the device transfermode for SATA.
2004-09-01 12:15:44 +00:00
}
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
for (retry = 0; retry < 10; retry++) {
for (loop = 0; loop < 10; loop++) {
ATA_IDX_OUTL(ch, ATA_SCONTROL, ATA_SC_DET_RESET);
ata_udelay(100);
if ((ATA_IDX_INL(ch, ATA_SCONTROL) &
ATA_SC_DET_MASK) == ATA_SC_DET_RESET)
break;
}
ata_udelay(5000);
for (loop = 0; loop < 10; loop++) {
Add support for the JMicron JMB360 SATAII controller. Thanks to JMicron for providing needed info. HW donated by: Ralf Folkerts
2006-01-25 23:07:42 +00:00
ATA_IDX_OUTL(ch, ATA_SCONTROL, ATA_SC_DET_IDLE |
ATA_SC_IPM_DIS_PARTIAL |
ATA_SC_IPM_DIS_SLUMBER);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ata_udelay(100);
if ((ATA_IDX_INL(ch, ATA_SCONTROL) & ATA_SC_DET_MASK) == 0) {
ata_sata_connect(ch);
return;
}
}
}
}
static void
ata_sata_phy_event(void *context, int dummy)
{
struct ata_connect_task *tp = (struct ata_connect_task *)context;
struct ata_channel *ch = device_get_softc(tp->dev);
device_t *children;
int nchildren, i;
mtx_lock(&Giant); /* newbus suckage it needs Giant */
if (tp->action == ATA_C_ATTACH) {
if (bootverbose)
device_printf(tp->dev, "CONNECTED\n");
ata_sata_connect(ch);
ata_identify(tp->dev);
}
if (tp->action == ATA_C_DETACH) {
if (!device_get_children(tp->dev, &children, &nchildren)) {
for (i = 0; i < nchildren; i++)
if (children[i])
device_delete_child(tp->dev, children[i]);
free(children, M_TEMP);
}
mtx_lock(&ch->state_mtx);
ch->state = ATA_IDLE;
mtx_unlock(&ch->state_mtx);
if (bootverbose)
device_printf(tp->dev, "DISCONNECTED\n");
Cleanup the storing and printing of the device transfermode for SATA.
2004-09-01 12:15:44 +00:00
}
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
mtx_unlock(&Giant); /* suckage code dealt with, release Giant */
free(tp, M_ATA);
Add support for the VIA 8237 (both PATA and SATA part). Cleanup the SATA support a bit now we are here anyways.
2003-10-30 13:16:21 +00:00
}
Generalise the SATA PHY handling code so it wont be duplicated for each SATA chip. Promise and Silicon Image are the current candidates for this.
2005-04-08 09:37:47 +00:00
static int
ata_sata_connect(struct ata_channel *ch)
{
u_int32_t status;
int timeout;
/* wait up to 1 second for "connect well" */
for (timeout = 0; timeout < 100 ; timeout++) {
status = ATA_IDX_INL(ch, ATA_SSTATUS);
if ((status & ATA_SS_CONWELL_MASK) == ATA_SS_CONWELL_GEN1 ||
(status & ATA_SS_CONWELL_MASK) == ATA_SS_CONWELL_GEN2)
break;
ata_udelay(10000);
}
if (timeout >= 100) {
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
if (bootverbose)
Generalise the SATA PHY handling code so it wont be duplicated for each SATA chip. Promise and Silicon Image are the current candidates for this.
2005-04-08 09:37:47 +00:00
device_printf(ch->dev, "SATA connect status=%08x\n", status);
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
return 0;
Generalise the SATA PHY handling code so it wont be duplicated for each SATA chip. Promise and Silicon Image are the current candidates for this.
2005-04-08 09:37:47 +00:00
}
/* clear SATA error register */
ATA_IDX_OUTL(ch, ATA_SERROR, ATA_IDX_INL(ch, ATA_SERROR));
/* find out what type device we got poll for spec'd 31 seconds */
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
/* XXX SOS 10 secs for now as I have little patience */
ch->devices = 0;
for (timeout = 0; timeout < 1000; timeout++) {
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
if (ATA_IDX_INB(ch, ATA_STATUS) & ATA_S_BUSY)
Generalise the SATA PHY handling code so it wont be duplicated for each SATA chip. Promise and Silicon Image are the current candidates for this.
2005-04-08 09:37:47 +00:00
DELAY(10000);
else
break;
}
Add support for AHCI compliant ATA devices. For now just support the Intel ICH6 as that the HW at hand. Sponsored by: pair.com
2005-05-11 16:10:08 +00:00
if (bootverbose)
Generalise the SATA PHY handling code so it wont be duplicated for each SATA chip. Promise and Silicon Image are the current candidates for this.
2005-04-08 09:37:47 +00:00
device_printf(ch->dev, "SATA connect ready time=%dms\n", timeout * 10);
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
if (timeout < 1000) {
Change the way ioctls are issue to ATA. The most prominent part is that its now possible to issue ata_requests directly to say acd0, instead of going through the cumbersome /dev/ata device.
2005-05-16 13:07:27 +00:00
if ((ATA_IDX_INB(ch, ATA_CYL_LSB) == ATAPI_MAGIC_LSB) &&
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
(ATA_IDX_INB(ch, ATA_CYL_MSB) == ATAPI_MAGIC_MSB))
ch->devices = ATA_ATAPI_MASTER;
Change the way ioctls are issue to ATA. The most prominent part is that its now possible to issue ata_requests directly to say acd0, instead of going through the cumbersome /dev/ata device.
2005-05-16 13:07:27 +00:00
else
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
ch->devices = ATA_ATA_MASTER;
}
Add support for AHCI compliant ATA devices. For now just support the Intel ICH6 as that the HW at hand. Sponsored by: pair.com
2005-05-11 16:10:08 +00:00
if (bootverbose)
Change the way ioctls are issue to ATA. The most prominent part is that its now possible to issue ata_requests directly to say acd0, instead of going through the cumbersome /dev/ata device.
2005-05-16 13:07:27 +00:00
device_printf(ch->dev, "sata_connect devices=0x%b\n",
ch->devices, "\20\3ATAPI_MASTER\1ATA_MASTER");
Generalise the SATA PHY handling code so it wont be duplicated for each SATA chip. Promise and Silicon Image are the current candidates for this.
2005-04-08 09:37:47 +00:00
return 1;
}
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
static void
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ata_sata_setmode(device_t dev, int mode)
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
{
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
struct ata_device *atadev = device_get_softc(dev);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
/*
* if we detect that the device isn't a real SATA device we limit
* the transfer mode to UDMA5/ATA100.
* this works around the problems some devices has with the
* Marvell 88SX8030 SATA->PATA converters and UDMA6/ATA133.
*/
if (atadev->param.satacapabilities != 0x0000 &&
atadev->param.satacapabilities != 0xffff) {
Whitespace cleanup.
2006-01-18 13:10:17 +00:00
struct ata_channel *ch = device_get_softc(device_get_parent(dev));
Properly report SATA connection speed.
2006-01-18 10:02:23 +00:00
/* on some drives we need to set the transfer mode */
ata_controlcmd(dev, ATA_SETFEATURES, ATA_SF_SETXFER, 0,
ata_limit_mode(dev, mode, ATA_UDMA6));
/* query SATA STATUS for the speed */
Fix panic on boot due to not all SATA chipsets having SSTATUS registeri resources set.
2006-01-21 17:38:10 +00:00
if (ch->r_io[ATA_SSTATUS].res &&
((ATA_IDX_INL(ch, ATA_SSTATUS) & ATA_SS_CONWELL_MASK) ==
ATA_SS_CONWELL_GEN2))
Properly report SATA connection speed.
2006-01-18 10:02:23 +00:00
atadev->mode = ATA_SA300;
else
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
atadev->mode = ATA_SA150;
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
}
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
else {
mode = ata_limit_mode(dev, mode, ATA_UDMA5);
if (!ata_controlcmd(dev, ATA_SETFEATURES, ATA_SF_SETXFER, 0, mode))
atadev->mode = mode;
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
}
}
Add support for AHCI compliant ATA devices. For now just support the Intel ICH6 as that the HW at hand. Sponsored by: pair.com
2005-05-11 16:10:08 +00:00
/*
* AHCI v1.0 compliant SATA chipset support functions
*/
struct ata_ahci_dma_prd {
Change the way ioctls are issue to ATA. The most prominent part is that its now possible to issue ata_requests directly to say acd0, instead of going through the cumbersome /dev/ata device.
2005-05-16 13:07:27 +00:00
u_int64_t dba;
u_int32_t reserved;
u_int32_t dbc; /* 0 based */
#define ATA_AHCI_PRD_MASK 0x003fffff /* max 4MB */
#define ATA_AHCI_PRD_IPC (1<<31)
Add support for AHCI compliant ATA devices. For now just support the Intel ICH6 as that the HW at hand. Sponsored by: pair.com
2005-05-11 16:10:08 +00:00
} __packed;
struct ata_ahci_cmd_tab {
Change the way ioctls are issue to ATA. The most prominent part is that its now possible to issue ata_requests directly to say acd0, instead of going through the cumbersome /dev/ata device.
2005-05-16 13:07:27 +00:00
u_int8_t cfis[64];
u_int8_t acmd[32];
u_int8_t reserved[32];
struct ata_ahci_dma_prd prd_tab[16];
Add support for AHCI compliant ATA devices. For now just support the Intel ICH6 as that the HW at hand. Sponsored by: pair.com
2005-05-11 16:10:08 +00:00
} __packed;
struct ata_ahci_cmd_list {
Change the way ioctls are issue to ATA. The most prominent part is that its now possible to issue ata_requests directly to say acd0, instead of going through the cumbersome /dev/ata device.
2005-05-16 13:07:27 +00:00
u_int16_t cmd_flags;
u_int16_t prd_length; /* PRD entries */
u_int32_t bytecount;
u_int64_t cmd_table_phys; /* 128byte aligned */
Add support for AHCI compliant ATA devices. For now just support the Intel ICH6 as that the HW at hand. Sponsored by: pair.com
2005-05-11 16:10:08 +00:00
} __packed;
Update AHCI support to be more generic. Add support for AHCI on the VIA VT8251.
2006-06-28 09:59:09 +00:00
static int
ata_ahci_chipinit(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
u_int32_t version;
/* reset AHCI controller */
ATA_OUTL(ctlr->r_res2, ATA_AHCI_GHC,
ATA_INL(ctlr->r_res2, ATA_AHCI_GHC) | ATA_AHCI_GHC_HR);
DELAY(1000000);
if (ATA_INL(ctlr->r_res2, ATA_AHCI_GHC) & ATA_AHCI_GHC_HR) {
bus_release_resource(dev, ctlr->r_type2, ctlr->r_rid2, ctlr->r_res2);
device_printf(dev, "AHCI controller reset failure\n");
return ENXIO;
}
/* enable AHCI mode */
ATA_OUTL(ctlr->r_res2, ATA_AHCI_GHC,
ATA_INL(ctlr->r_res2, ATA_AHCI_GHC) | ATA_AHCI_GHC_AE);
/* get the number of HW channels */
ctlr->channels = (ATA_INL(ctlr->r_res2, ATA_AHCI_CAP) & ATA_AHCI_NPMASK)+1;
/* clear interrupts */
ATA_OUTL(ctlr->r_res2, ATA_AHCI_IS, ATA_INL(ctlr->r_res2, ATA_AHCI_IS));
/* enable AHCI interrupts */
ATA_OUTL(ctlr->r_res2, ATA_AHCI_GHC,
ATA_INL(ctlr->r_res2, ATA_AHCI_GHC) | ATA_AHCI_GHC_IE);
ctlr->reset = ata_ahci_reset;
ctlr->dmainit = ata_ahci_dmainit;
ctlr->allocate = ata_ahci_allocate;
ctlr->setmode = ata_sata_setmode;
/* enable PCI interrupt */
pci_write_config(dev, PCIR_COMMAND,
pci_read_config(dev, PCIR_COMMAND, 2) & ~0x0400, 2);
/* announce we support the HW */
version = ATA_INL(ctlr->r_res2, ATA_AHCI_VS);
device_printf(dev,
"AHCI Version %x%x.%x%x controller with %d ports detected\n",
(version >> 24) & 0xff, (version >> 16) & 0xff,
(version >> 8) & 0xff, version & 0xff, ctlr->channels);
return 0;
}
Add support for AHCI compliant ATA devices. For now just support the Intel ICH6 as that the HW at hand. Sponsored by: pair.com
2005-05-11 16:10:08 +00:00
static int
ata_ahci_allocate(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(device_get_parent(dev));
struct ata_channel *ch = device_get_softc(dev);
int offset = (ch->unit << 7);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
/* setup legacy cruft we need */
Add support for AHCI compliant ATA devices. For now just support the Intel ICH6 as that the HW at hand. Sponsored by: pair.com
2005-05-11 16:10:08 +00:00
ch->r_io[ATA_CYL_LSB].res = ctlr->r_res2;
ch->r_io[ATA_CYL_LSB].offset = ATA_AHCI_P_SIG + 1 + offset;
Add support for working around controllers that cannot do DMA in 48bit mode. The workaround use PIO mode above ~137GB to allow using the disk. Add the Acer chips with rev < 0xc4 as first candidate.
2005-08-17 15:00:33 +00:00
ch->r_io[ATA_CYL_MSB].res = ctlr->r_res2;
Add support for AHCI compliant ATA devices. For now just support the Intel ICH6 as that the HW at hand. Sponsored by: pair.com
2005-05-11 16:10:08 +00:00
ch->r_io[ATA_CYL_MSB].offset = ATA_AHCI_P_SIG + 3 + offset;
ch->r_io[ATA_STATUS].res = ctlr->r_res2;
ch->r_io[ATA_STATUS].offset = ATA_AHCI_P_TFD + offset;
ch->r_io[ATA_ALTSTAT].res = ctlr->r_res2;
ch->r_io[ATA_ALTSTAT].offset = ATA_AHCI_P_TFD + offset;
/* set the SATA resources */
ch->r_io[ATA_SSTATUS].res = ctlr->r_res2;
ch->r_io[ATA_SSTATUS].offset = ATA_AHCI_P_SSTS + offset;
ch->r_io[ATA_SERROR].res = ctlr->r_res2;
ch->r_io[ATA_SERROR].offset = ATA_AHCI_P_SERR + offset;
ch->r_io[ATA_SCONTROL].res = ctlr->r_res2;
ch->r_io[ATA_SCONTROL].offset = ATA_AHCI_P_SCTL + offset;
ch->r_io[ATA_SACTIVE].res = ctlr->r_res2;
ch->r_io[ATA_SACTIVE].offset = ATA_AHCI_P_SACT + offset;
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ch->hw.status = ata_ahci_status;
Add support for AHCI compliant ATA devices. For now just support the Intel ICH6 as that the HW at hand. Sponsored by: pair.com
2005-05-11 16:10:08 +00:00
ch->hw.begin_transaction = ata_ahci_begin_transaction;
ch->hw.end_transaction = ata_ahci_end_transaction;
Change the way ioctls are issue to ATA. The most prominent part is that its now possible to issue ata_requests directly to say acd0, instead of going through the cumbersome /dev/ata device.
2005-05-16 13:07:27 +00:00
ch->hw.command = NULL; /* not used here */
Add support for AHCI compliant ATA devices. For now just support the Intel ICH6 as that the HW at hand. Sponsored by: pair.com
2005-05-11 16:10:08 +00:00
/* setup the work areas */
ATA_OUTL(ctlr->r_res2, ATA_AHCI_P_CLB + offset,
ch->dma->work_bus + ATA_AHCI_CL_OFFSET);
ATA_OUTL(ctlr->r_res2, ATA_AHCI_P_CLBU + offset, 0x00000000);
ATA_OUTL(ctlr->r_res2, ATA_AHCI_P_FB + offset,
ch->dma->work_bus + ATA_AHCI_FB_OFFSET);
ATA_OUTL(ctlr->r_res2, ATA_AHCI_P_FBU + offset, 0x00000000);
/* enable wanted port interrupts */
ATA_OUTL(ctlr->r_res2, ATA_AHCI_P_IE + offset,
(ATA_AHCI_P_IX_CPD | ATA_AHCI_P_IX_TFE | ATA_AHCI_P_IX_HBF |
ATA_AHCI_P_IX_HBD | ATA_AHCI_P_IX_IF | ATA_AHCI_P_IX_OF |
ATA_AHCI_P_IX_PRC | ATA_AHCI_P_IX_PC | ATA_AHCI_P_IX_DP |
ATA_AHCI_P_IX_UF | ATA_AHCI_P_IX_SDB | ATA_AHCI_P_IX_DS |
ATA_AHCI_P_IX_PS | ATA_AHCI_P_IX_DHR));
/* start operations on this channel */
ATA_OUTL(ctlr->r_res2, ATA_AHCI_P_CMD + offset,
(ATA_AHCI_P_CMD_ACTIVE | ATA_AHCI_P_CMD_FRE |
ATA_AHCI_P_CMD_POD | ATA_AHCI_P_CMD_SUD | ATA_AHCI_P_CMD_ST));
return 0;
}
static int
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ata_ahci_status(device_t dev)
Add support for AHCI compliant ATA devices. For now just support the Intel ICH6 as that the HW at hand. Sponsored by: pair.com
2005-05-11 16:10:08 +00:00
{
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
struct ata_pci_controller *ctlr = device_get_softc(device_get_parent(dev));
struct ata_channel *ch = device_get_softc(dev);
struct ata_connect_task *tp;
Add support for the JMicron JMB360 SATAII controller. Thanks to JMicron for providing needed info. HW donated by: Ralf Folkerts
2006-01-25 23:07:42 +00:00
u_int32_t action, istatus, sstatus, error, issued;
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
int offset = (ch->unit << 7);
int tag = 0;
Add support for AHCI compliant ATA devices. For now just support the Intel ICH6 as that the HW at hand. Sponsored by: pair.com
2005-05-11 16:10:08 +00:00
Update AHCI support to be more generic. Add support for AHCI on the VIA VT8251.
2006-06-28 09:59:09 +00:00
action = ATA_INL(ctlr->r_res2, ATA_AHCI_IS);
if (action & (1 << ch->unit)) {
Add support for the JMicron JMB360 SATAII controller. Thanks to JMicron for providing needed info. HW donated by: Ralf Folkerts
2006-01-25 23:07:42 +00:00
istatus = ATA_INL(ctlr->r_res2, ATA_AHCI_P_IS + offset);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
issued = ATA_INL(ctlr->r_res2, ATA_AHCI_P_CI + offset);
Add support for the JMicron JMB360 SATAII controller. Thanks to JMicron for providing needed info. HW donated by: Ralf Folkerts
2006-01-25 23:07:42 +00:00
sstatus = ATA_INL(ctlr->r_res2, ATA_AHCI_P_SSTS + offset);
error = ATA_INL(ctlr->r_res2, ATA_AHCI_P_SERR + offset);
/* clear interrupt(s) */
ATA_OUTL(ctlr->r_res2, ATA_AHCI_IS, action);
ATA_OUTL(ctlr->r_res2, ATA_AHCI_P_IS + offset, istatus);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ATA_OUTL(ctlr->r_res2, ATA_AHCI_P_SERR + offset, error);
Add support for AHCI compliant ATA devices. For now just support the Intel ICH6 as that the HW at hand. Sponsored by: pair.com
2005-05-11 16:10:08 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
/* do we have cold connect surprise */
Add support for the JMicron JMB360 SATAII controller. Thanks to JMicron for providing needed info. HW donated by: Ralf Folkerts
2006-01-25 23:07:42 +00:00
if (istatus & ATA_AHCI_P_IX_CPD) {
Update AHCI support to be more generic. Add support for AHCI on the VIA VT8251.
2006-06-28 09:59:09 +00:00
printf("ata_ahci_status status=%08x sstatus=%08x error=%08x\n",
Add support for the JMicron JMB360 SATAII controller. Thanks to JMicron for providing needed info. HW donated by: Ralf Folkerts
2006-01-25 23:07:42 +00:00
istatus, sstatus, error);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
}
Add support for AHCI compliant ATA devices. For now just support the Intel ICH6 as that the HW at hand. Sponsored by: pair.com
2005-05-11 16:10:08 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
/* check for and handle connect events */
Add support for the JMicron JMB360 SATAII controller. Thanks to JMicron for providing needed info. HW donated by: Ralf Folkerts
2006-01-25 23:07:42 +00:00
if ((istatus & ATA_AHCI_P_IX_PC) &&
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
(tp = (struct ata_connect_task *)
malloc(sizeof(struct ata_connect_task),
M_ATA, M_NOWAIT | M_ZERO))) {
Add support for AHCI compliant ATA devices. For now just support the Intel ICH6 as that the HW at hand. Sponsored by: pair.com
2005-05-11 16:10:08 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
if (bootverbose)
device_printf(ch->dev, "CONNECT requested\n");
tp->action = ATA_C_ATTACH;
tp->dev = ch->dev;
TASK_INIT(&tp->task, 0, ata_sata_phy_event, tp);
taskqueue_enqueue(taskqueue_thread, &tp->task);
}
Add support for AHCI compliant ATA devices. For now just support the Intel ICH6 as that the HW at hand. Sponsored by: pair.com
2005-05-11 16:10:08 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
/* check for and handle disconnect events */
Add support for the JMicron JMB360 SATAII controller. Thanks to JMicron for providing needed info. HW donated by: Ralf Folkerts
2006-01-25 23:07:42 +00:00
else if ((istatus & ATA_AHCI_P_IX_PRC) &&
!((sstatus & ATA_SS_CONWELL_MASK) == ATA_SS_CONWELL_GEN1 ||
(sstatus & ATA_SS_CONWELL_MASK) == ATA_SS_CONWELL_GEN2) &&
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
(tp = (struct ata_connect_task *)
malloc(sizeof(struct ata_connect_task),
M_ATA, M_NOWAIT | M_ZERO))) {
if (bootverbose)
device_printf(ch->dev, "DISCONNECT requested\n");
tp->action = ATA_C_DETACH;
tp->dev = ch->dev;
TASK_INIT(&tp->task, 0, ata_sata_phy_event, tp);
taskqueue_enqueue(taskqueue_thread, &tp->task);
}
Whitespace cleanup.
2006-01-18 13:10:17 +00:00
/* do we have any device action ? */
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
if (!(issued & (1 << tag)))
return 1;
}
return 0;
Add support for AHCI compliant ATA devices. For now just support the Intel ICH6 as that the HW at hand. Sponsored by: pair.com
2005-05-11 16:10:08 +00:00
}
/* must be called with ATA channel locked and state_mtx held */
static int
ata_ahci_begin_transaction(struct ata_request *request)
{
struct ata_pci_controller *ctlr=device_get_softc(GRANDPARENT(request->dev));
struct ata_channel *ch = device_get_softc(device_get_parent(request->dev));
struct ata_ahci_cmd_tab *ctp;
struct ata_ahci_cmd_list *clp;
Update AHCI support to be more generic. Add support for AHCI on the VIA VT8251.
2006-06-28 09:59:09 +00:00
int tag = 0, entries = 0;
int fis_size;
Add support for AHCI compliant ATA devices. For now just support the Intel ICH6 as that the HW at hand. Sponsored by: pair.com
2005-05-11 16:10:08 +00:00
/* get a piece of the workspace for this request */
ctp = (struct ata_ahci_cmd_tab *)
(ch->dma->work + ATA_AHCI_CT_OFFSET + (ATA_AHCI_CT_SIZE * tag));
/* setup the FIS for this request */ /* XXX SOS ATAPI missing still */
if (!(fis_size = ata_ahci_setup_fis(&ctp->cfis[0], request))) {
device_printf(request->dev, "setting up SATA FIS failed\n");
request->result = EIO;
return ATA_OP_FINISHED;
}
/* if request moves data setup and load SG list */
if (request->flags & (ATA_R_READ | ATA_R_WRITE)) {
Change the way ioctls are issue to ATA. The most prominent part is that its now possible to issue ata_requests directly to say acd0, instead of going through the cumbersome /dev/ata device.
2005-05-16 13:07:27 +00:00
if (ch->dma->load(ch->dev, request->data, request->bytecount,
request->flags & ATA_R_READ,
Add support for AHCI compliant ATA devices. For now just support the Intel ICH6 as that the HW at hand. Sponsored by: pair.com
2005-05-11 16:10:08 +00:00
ctp->prd_tab, &entries)) {
device_printf(request->dev, "setting up DMA failed\n");
request->result = EIO;
return ATA_OP_FINISHED;
Change the way ioctls are issue to ATA. The most prominent part is that its now possible to issue ata_requests directly to say acd0, instead of going through the cumbersome /dev/ata device.
2005-05-16 13:07:27 +00:00
}
Add support for AHCI compliant ATA devices. For now just support the Intel ICH6 as that the HW at hand. Sponsored by: pair.com
2005-05-11 16:10:08 +00:00
}
/* setup the command list entry */
clp = (struct ata_ahci_cmd_list *)
(ch->dma->work + ATA_AHCI_CL_OFFSET + (ATA_AHCI_CL_SIZE * tag));
clp->prd_length = entries;
clp->cmd_flags = (request->flags & ATA_R_WRITE ? (1<<6) : 0) |
Change the way ioctls are issue to ATA. The most prominent part is that its now possible to issue ata_requests directly to say acd0, instead of going through the cumbersome /dev/ata device.
2005-05-16 13:07:27 +00:00
(request->flags & ATA_R_ATAPI ? (1<<5) : 0) |
(fis_size / sizeof(u_int32_t));
Add support for AHCI compliant ATA devices. For now just support the Intel ICH6 as that the HW at hand. Sponsored by: pair.com
2005-05-11 16:10:08 +00:00
clp->bytecount = 0;
clp->cmd_table_phys = htole64(ch->dma->work_bus + ATA_AHCI_CT_OFFSET +
Change the way ioctls are issue to ATA. The most prominent part is that its now possible to issue ata_requests directly to say acd0, instead of going through the cumbersome /dev/ata device.
2005-05-16 13:07:27 +00:00
(ATA_AHCI_CT_SIZE * tag));
Add support for AHCI compliant ATA devices. For now just support the Intel ICH6 as that the HW at hand. Sponsored by: pair.com
2005-05-11 16:10:08 +00:00
/* clear eventual ACTIVE bit */
ATA_IDX_OUTL(ch, ATA_SACTIVE, ATA_IDX_INL(ch, ATA_SACTIVE) & (1 << tag));
/* issue the command */
ATA_OUTL(ctlr->r_res2, ATA_AHCI_P_CI + (ch->unit << 7), (1 << tag));
/* start the timeout */
callout_reset(&request->callout, request->timeout * hz,
(timeout_t*)ata_timeout, request);
return ATA_OP_CONTINUES;
}
/* must be called with ATA channel locked and state_mtx held */
static int
ata_ahci_end_transaction(struct ata_request *request)
{
struct ata_pci_controller *ctlr=device_get_softc(GRANDPARENT(request->dev));
struct ata_channel *ch = device_get_softc(device_get_parent(request->dev));
struct ata_ahci_cmd_list *clp;
u_int32_t tf_data;
int tag = 0;
/* kill the timeout */
callout_stop(&request->callout);
/* get status */
tf_data = ATA_INL(ctlr->r_res2, ATA_AHCI_P_TFD + (ch->unit << 7));
request->status = tf_data;
/* if error status get details */
if (request->status & ATA_S_ERROR)
request->error = tf_data >> 8;
/* record how much data we actually moved */
clp = (struct ata_ahci_cmd_list *)
Change the way ioctls are issue to ATA. The most prominent part is that its now possible to issue ata_requests directly to say acd0, instead of going through the cumbersome /dev/ata device.
2005-05-16 13:07:27 +00:00
(ch->dma->work + ATA_AHCI_CL_OFFSET + (ATA_AHCI_CL_SIZE * tag));
Add support for AHCI compliant ATA devices. For now just support the Intel ICH6 as that the HW at hand. Sponsored by: pair.com
2005-05-11 16:10:08 +00:00
request->donecount = clp->bytecount;
/* release SG list etc */
ch->dma->unload(ch->dev);
return ATA_OP_FINISHED;
}
static void
ata_ahci_reset(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(device_get_parent(dev));
struct ata_channel *ch = device_get_softc(dev);
u_int32_t cmd;
Update AHCI support to be more generic. Add support for AHCI on the VIA VT8251.
2006-06-28 09:59:09 +00:00
int timeout, offset = (ch->unit << 7);
Add support for AHCI compliant ATA devices. For now just support the Intel ICH6 as that the HW at hand. Sponsored by: pair.com
2005-05-11 16:10:08 +00:00
/* kill off all activity on this channel */
cmd = ATA_INL(ctlr->r_res2, ATA_AHCI_P_CMD + offset);
ATA_OUTL(ctlr->r_res2, ATA_AHCI_P_CMD + offset,
Update AHCI support to be more generic. Add support for AHCI on the VIA VT8251.
2006-06-28 09:59:09 +00:00
cmd & ~(ATA_AHCI_P_CMD_FRE | ATA_AHCI_P_CMD_ST));
/* XXX SOS this is not entirely wrong */
timeout = 0;
do {
DELAY(1000);
if (timeout++ > 500)
device_printf(dev, "stopping AHCI engine failed\n");
break;
}
while (ATA_INL(ctlr->r_res2, ATA_AHCI_P_CMD + offset) & ATA_AHCI_P_CMD_CR);
/* issue Command List Override if supported */
if (ATA_INL(ctlr->r_res2, ATA_AHCI_CAP) & ATA_AHCI_CAP_CLO) {
cmd = ATA_INL(ctlr->r_res2, ATA_AHCI_P_CMD + offset);
cmd |= ATA_AHCI_P_CMD_CLO;
ATA_OUTL(ctlr->r_res2, ATA_AHCI_P_CMD + offset, cmd);
timeout = 0;
do {
DELAY(1000);
if (timeout++ > 500)
device_printf(dev, "executing CLO failed\n");
break;
}
while (ATA_INL(ctlr->r_res2, ATA_AHCI_P_CMD+offset)&ATA_AHCI_P_CMD_CLO);
}
Add support for AHCI compliant ATA devices. For now just support the Intel ICH6 as that the HW at hand. Sponsored by: pair.com
2005-05-11 16:10:08 +00:00
/* spin up device */
ATA_OUTL(ctlr->r_res2, ATA_AHCI_P_CMD + offset, ATA_AHCI_P_CMD_SUD);
ata_sata_phy_enable(ch);
/* clear any interrupts pending on this channel */
ATA_OUTL(ctlr->r_res2, ATA_AHCI_P_IS + offset,
ATA_INL(ctlr->r_res2, ATA_AHCI_P_IS + offset));
/* start operations on this channel */
ATA_OUTL(ctlr->r_res2, ATA_AHCI_P_CMD + offset,
(ATA_AHCI_P_CMD_ACTIVE | ATA_AHCI_P_CMD_FRE |
ATA_AHCI_P_CMD_POD | ATA_AHCI_P_CMD_SUD | ATA_AHCI_P_CMD_ST));
}
static void
ata_ahci_dmasetprd(void *xsc, bus_dma_segment_t *segs, int nsegs, int error)
{
struct ata_dmasetprd_args *args = xsc;
struct ata_ahci_dma_prd *prd = args->dmatab;
int i;
if (!(args->error = error)) {
for (i = 0; i < nsegs; i++) {
prd[i].dba = htole64(segs[i].ds_addr);
prd[i].dbc = htole32((segs[i].ds_len - 1) & ATA_AHCI_PRD_MASK);
}
}
args->nsegs = nsegs;
}
static void
ata_ahci_dmainit(device_t dev)
{
struct ata_channel *ch = device_get_softc(dev);
ata_dmainit(dev);
if (ch->dma) {
/* note start and stop are not used here */
ch->dma->setprd = ata_ahci_dmasetprd;
ch->dma->max_iosize = 8192 * DEV_BSIZE;
}
}
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static int
ata_ahci_setup_fis(u_int8_t *fis, struct ata_request *request)
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
{
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
struct ata_device *atadev = device_get_softc(request->dev);
int idx = 0;
/* XXX SOS add ATAPI commands support later */
ata_modify_if_48bit(request);
fis[idx++] = 0x27; /* host to device */
fis[idx++] = 0x80; /* command FIS (note PM goes here) */
fis[idx++] = request->u.ata.command;
fis[idx++] = request->u.ata.feature;
fis[idx++] = request->u.ata.lba;
fis[idx++] = request->u.ata.lba >> 8;
fis[idx++] = request->u.ata.lba >> 16;
fis[idx] = ATA_D_LBA | atadev->unit;
if (atadev->flags & ATA_D_48BIT_ACTIVE)
idx++;
else
fis[idx++] |= (request->u.ata.lba >> 24 & 0x0f);
fis[idx++] = request->u.ata.lba >> 24;
fis[idx++] = request->u.ata.lba >> 32;
fis[idx++] = request->u.ata.lba >> 40;
fis[idx++] = request->u.ata.feature >> 8;
fis[idx++] = request->u.ata.count;
fis[idx++] = request->u.ata.count >> 8;
fis[idx++] = 0x00;
fis[idx++] = ATA_A_4BIT;
fis[idx++] = 0x00;
fis[idx++] = 0x00;
fis[idx++] = 0x00;
fis[idx++] = 0x00;
return idx;
}
/*
* Acard chipset support functions
*/
int
ata_acard_ident(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
struct ata_chip_id *idx;
static struct ata_chip_id ids[] =
{{ ATA_ATP850R, 0, ATPOLD, 0x00, ATA_UDMA2, "ATP850" },
{ ATA_ATP860A, 0, 0, 0x00, ATA_UDMA4, "ATP860A" },
{ ATA_ATP860R, 0, 0, 0x00, ATA_UDMA4, "ATP860R" },
{ ATA_ATP865A, 0, 0, 0x00, ATA_UDMA6, "ATP865A" },
{ ATA_ATP865R, 0, 0, 0x00, ATA_UDMA6, "ATP865R" },
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
{ 0, 0, 0, 0, 0, 0}};
char buffer[64];
Support the ICH5 SATA part. Fix HPT374 UDMA133 timing. Fix Promise ID. Cosmetics on probe print for Promise & HPT. Approved by: re
2003-05-18 16:45:48 +00:00
if (!(idx = ata_match_chip(dev, ids)))
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
return ENXIO;
Cleanup the strings for printing chipset names, saves quite some space. Add a couble new nVidia chips now we are here.
2005-12-16 08:12:13 +00:00
sprintf(buffer, "Acard %s %s controller",
idx->text, ata_mode2str(idx->max_dma));
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
device_set_desc_copy(dev, buffer);
ctlr->chip = idx;
ctlr->chipinit = ata_acard_chipinit;
return 0;
}
static int
ata_acard_chipinit(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
if (ata_setup_interrupt(dev))
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
return ENXIO;
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ctlr->allocate = ata_acard_allocate;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (ctlr->chip->cfg1 == ATPOLD) {
ctlr->setmode = ata_acard_850_setmode;
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
ctlr->locking = ata_serialize;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
else
ctlr->setmode = ata_acard_86X_setmode;
return 0;
}
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static int
ata_acard_allocate(device_t dev)
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
{
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
struct ata_channel *ch = device_get_softc(dev);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
/* setup the usual register normal pci style */
if (ata_pci_allocate(dev))
return ENXIO;
ch->hw.status = ata_acard_status;
return 0;
}
static int
ata_acard_status(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(device_get_parent(dev));
struct ata_channel *ch = device_get_softc(dev);
if (ctlr->chip->cfg1 == ATPOLD &&
ATA_LOCKING(ch->dev, ATA_LF_WHICH) != ch->unit)
return 0;
if (ch->dma && (ch->dma->flags & ATA_DMA_ACTIVE)) {
int bmstat = ATA_IDX_INB(ch, ATA_BMSTAT_PORT) & ATA_BMSTAT_MASK;
if ((bmstat & (ATA_BMSTAT_ACTIVE | ATA_BMSTAT_INTERRUPT)) !=
ATA_BMSTAT_INTERRUPT)
return 0;
ATA_IDX_OUTB(ch, ATA_BMSTAT_PORT, bmstat & ~ATA_BMSTAT_ERROR);
DELAY(1);
ATA_IDX_OUTB(ch, ATA_BMCMD_PORT,
Finally fix support for the newer MCP51/MCP55 nVidia chipsets. The register layout has changed since the original NV4 - sigh. Hotplug support has been fixed for all nVidia chipsets that supports it (including the MCP51/55). HW donated by: Kingsley College
2006-07-24 10:44:50 +00:00
ATA_IDX_INB(ch, ATA_BMCMD_PORT) & ~ATA_BMCMD_START_STOP);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
DELAY(1);
}
if (ATA_IDX_INB(ch, ATA_ALTSTAT) & ATA_S_BUSY) {
DELAY(100);
if (ATA_IDX_INB(ch, ATA_ALTSTAT) & ATA_S_BUSY)
return 0;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
return 1;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
static void
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_acard_850_setmode(device_t dev, int mode)
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
{
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
device_t gparent = GRANDPARENT(dev);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
struct ata_pci_controller *ctlr = device_get_softc(gparent);
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
struct ata_channel *ch = device_get_softc(device_get_parent(dev));
struct ata_device *atadev = device_get_softc(dev);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
int devno = (ch->unit << 1) + ATA_DEV(atadev->unit);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
int error;
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
mode = ata_limit_mode(dev, mode,
ata_atapi(dev) ? ATA_PIO_MAX : ctlr->chip->max_dma);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
/* XXX SOS missing WDMA0+1 + PIO modes */
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (mode >= ATA_WDMA2) {
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
error = ata_controlcmd(dev, ATA_SETFEATURES, ATA_SF_SETXFER, 0, mode);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (bootverbose)
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
device_printf(dev, "%ssetting %s on %s chip\n",
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
(error) ? "FAILURE " : "",
ata_mode2str(mode), ctlr->chip->text);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (!error) {
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
u_int8_t reg54 = pci_read_config(gparent, 0x54, 1);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
reg54 &= ~(0x03 << (devno << 1));
if (mode >= ATA_UDMA0)
reg54 |= (((mode & ATA_MODE_MASK) + 1) << (devno << 1));
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, 0x54, reg54, 1);
pci_write_config(gparent, 0x4a, 0xa6, 1);
pci_write_config(gparent, 0x40 + (devno << 1), 0x0301, 2);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
atadev->mode = mode;
return;
}
}
/* we could set PIO mode timings, but we assume the BIOS did that */
}
static void
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_acard_86X_setmode(device_t dev, int mode)
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
{
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
device_t gparent = GRANDPARENT(dev);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
struct ata_pci_controller *ctlr = device_get_softc(gparent);
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
struct ata_channel *ch = device_get_softc(device_get_parent(dev));
struct ata_device *atadev = device_get_softc(dev);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
int devno = (ch->unit << 1) + ATA_DEV(atadev->unit);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
int error;
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
mode = ata_limit_mode(dev, mode,
ata_atapi(dev) ? ATA_PIO_MAX : ctlr->chip->max_dma);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
mode = ata_check_80pin(dev, mode);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
/* XXX SOS missing WDMA0+1 + PIO modes */
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (mode >= ATA_WDMA2) {
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
error = ata_controlcmd(dev, ATA_SETFEATURES, ATA_SF_SETXFER, 0, mode);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (bootverbose)
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
device_printf(dev, "%ssetting %s on %s chip\n",
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
(error) ? "FAILURE " : "",
ata_mode2str(mode), ctlr->chip->text);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (!error) {
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
u_int16_t reg44 = pci_read_config(gparent, 0x44, 2);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
reg44 &= ~(0x000f << (devno << 2));
if (mode >= ATA_UDMA0)
reg44 |= (((mode & ATA_MODE_MASK) + 1) << (devno << 2));
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, 0x44, reg44, 2);
pci_write_config(gparent, 0x4a, 0xa6, 1);
pci_write_config(gparent, 0x40 + devno, 0x31, 1);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
atadev->mode = mode;
return;
}
}
/* we could set PIO mode timings, but we assume the BIOS did that */
}
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
/*
* Acer Labs Inc (ALI) chipset support functions
*/
int
ata_ali_ident(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
struct ata_chip_id *idx;
static struct ata_chip_id ids[] =
Cleanup the strings for printing chipset names, saves quite some space. Add a couble new nVidia chips now we are here.
2005-12-16 08:12:13 +00:00
{{ ATA_ALI_5289, 0x00, 2, ALISATA, ATA_SA150, "M5289" },
{ ATA_ALI_5287, 0x00, 4, ALISATA, ATA_SA150, "M5287" },
{ ATA_ALI_5281, 0x00, 2, ALISATA, ATA_SA150, "M5281" },
{ ATA_ALI_5229, 0xc5, 0, ALINEW, ATA_UDMA6, "M5229" },
{ ATA_ALI_5229, 0xc4, 0, ALINEW, ATA_UDMA5, "M5229" },
{ ATA_ALI_5229, 0xc2, 0, ALINEW, ATA_UDMA4, "M5229" },
{ ATA_ALI_5229, 0x20, 0, ALIOLD, ATA_UDMA2, "M5229" },
{ ATA_ALI_5229, 0x00, 0, ALIOLD, ATA_WDMA2, "M5229" },
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
{ 0, 0, 0, 0, 0, 0}};
char buffer[64];
Support the ICH5 SATA part. Fix HPT374 UDMA133 timing. Fix Promise ID. Cosmetics on probe print for Promise & HPT. Approved by: re
2003-05-18 16:45:48 +00:00
if (!(idx = ata_match_chip(dev, ids)))
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
return ENXIO;
Cleanup the strings for printing chipset names, saves quite some space. Add a couble new nVidia chips now we are here.
2005-12-16 08:12:13 +00:00
sprintf(buffer, "AcerLabs %s %s controller",
idx->text, ata_mode2str(idx->max_dma));
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
device_set_desc_copy(dev, buffer);
ctlr->chip = idx;
ctlr->chipinit = ata_ali_chipinit;
return 0;
}
static int
ata_ali_chipinit(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
if (ata_setup_interrupt(dev))
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
return ENXIO;
Add firstshot support for Acer Labs Inc SATA parts Update ALi PATA support to handle ATA133 as well
2005-04-05 14:51:43 +00:00
switch (ctlr->chip->cfg2) {
case ALISATA:
ctlr->channels = ctlr->chip->cfg1;
Add support for working around controllers that cannot do DMA in 48bit mode. The workaround use PIO mode above ~137GB to allow using the disk. Add the Acer chips with rev < 0xc4 as first candidate.
2005-08-17 15:00:33 +00:00
ctlr->allocate = ata_ali_sata_allocate;
Only try to allocate and use the SATA resource if they are enabled by the BIOS. It seems some BIOS's doesn't get this right, and that would result in ATA panic'ing.
2005-04-25 07:50:51 +00:00
ctlr->setmode = ata_sata_setmode;
Add support for the JMicron JMB360 SATAII controller. Thanks to JMicron for providing needed info. HW donated by: Ralf Folkerts
2006-01-25 23:07:42 +00:00
/* enable PCI interrupt */
pci_write_config(dev, PCIR_COMMAND,
pci_read_config(dev, PCIR_COMMAND, 2) & ~0x0400, 2);
Add firstshot support for Acer Labs Inc SATA parts Update ALi PATA support to handle ATA133 as well
2005-04-05 14:51:43 +00:00
break;
case ALINEW:
Apply fix for "pr82261 DMA-support on Sparc64 broken" The Acer chip or wiring that SUN uses has problems that this patch tries to work around. Original patch by Marius Strobl, hacked into shape by me..
2005-08-23 08:53:01 +00:00
/* use device interrupt as byte count end */
pci_write_config(dev, 0x4a, pci_read_config(dev, 0x4a, 1) | 0x20, 1);
Add firstshot support for Acer Labs Inc SATA parts Update ALi PATA support to handle ATA133 as well
2005-04-05 14:51:43 +00:00
/* enable cable detection and UDMA support on newer chips */
Fix support for Acer chips. The UDMA enable regs wasn't set properly.
2003-02-25 21:22:27 +00:00
pci_write_config(dev, 0x4b, pci_read_config(dev, 0x4b, 1) | 0x09, 1);
Add support for working around controllers that cannot do DMA in 48bit mode. The workaround use PIO mode above ~137GB to allow using the disk. Add the Acer chips with rev < 0xc4 as first candidate.
2005-08-17 15:00:33 +00:00
Apply fix for "pr82261 DMA-support on Sparc64 broken" The Acer chip or wiring that SUN uses has problems that this patch tries to work around. Original patch by Marius Strobl, hacked into shape by me..
2005-08-23 08:53:01 +00:00
/* enable ATAPI UDMA mode */
pci_write_config(dev, 0x53, pci_read_config(dev, 0x53, 1) | 0x01, 1);
Add support for working around controllers that cannot do DMA in 48bit mode. The workaround use PIO mode above ~137GB to allow using the disk. Add the Acer chips with rev < 0xc4 as first candidate.
2005-08-17 15:00:33 +00:00
/* only chips with revision > 0xc4 can do 48bit DMA */
if (ctlr->chip->chiprev <= 0xc4)
device_printf(dev,
"using PIO transfers above 137GB as workaround for "
"48bit DMA access bug, expect reduced performance\n");
ctlr->allocate = ata_ali_allocate;
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ctlr->reset = ata_ali_reset;
Add firstshot support for Acer Labs Inc SATA parts Update ALi PATA support to handle ATA133 as well
2005-04-05 14:51:43 +00:00
ctlr->setmode = ata_ali_setmode;
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
break;
Add firstshot support for Acer Labs Inc SATA parts Update ALi PATA support to handle ATA133 as well
2005-04-05 14:51:43 +00:00
case ALIOLD:
/* deactivate the ATAPI FIFO and enable ATAPI UDMA */
Apply fix for "pr82261 DMA-support on Sparc64 broken" The Acer chip or wiring that SUN uses has problems that this patch tries to work around. Original patch by Marius Strobl, hacked into shape by me..
2005-08-23 08:53:01 +00:00
pci_write_config(dev, 0x53, pci_read_config(dev, 0x53, 1) | 0x03, 1);
Add firstshot support for Acer Labs Inc SATA parts Update ALi PATA support to handle ATA133 as well
2005-04-05 14:51:43 +00:00
ctlr->setmode = ata_ali_setmode;
break;
}
return 0;
}
static int
ata_ali_allocate(device_t dev)
Add support for working around controllers that cannot do DMA in 48bit mode. The workaround use PIO mode above ~137GB to allow using the disk. Add the Acer chips with rev < 0xc4 as first candidate.
2005-08-17 15:00:33 +00:00
{
struct ata_pci_controller *ctlr = device_get_softc(device_get_parent(dev));
struct ata_channel *ch = device_get_softc(dev);
/* setup the usual register normal pci style */
Properly react to allocation failures. Found by: imp@
2005-10-10 19:12:43 +00:00
if (ata_pci_allocate(dev))
return ENXIO;
Add support for working around controllers that cannot do DMA in 48bit mode. The workaround use PIO mode above ~137GB to allow using the disk. Add the Acer chips with rev < 0xc4 as first candidate.
2005-08-17 15:00:33 +00:00
Properly react to allocation failures. Found by: imp@
2005-10-10 19:12:43 +00:00
/* older chips can't do 48bit DMA transfers */
Add support for working around controllers that cannot do DMA in 48bit mode. The workaround use PIO mode above ~137GB to allow using the disk. Add the Acer chips with rev < 0xc4 as first candidate.
2005-08-17 15:00:33 +00:00
if (ctlr->chip->chiprev <= 0xc4)
ch->flags |= ATA_NO_48BIT_DMA;
return 0;
}
static int
ata_ali_sata_allocate(device_t dev)
Add firstshot support for Acer Labs Inc SATA parts Update ALi PATA support to handle ATA133 as well
2005-04-05 14:51:43 +00:00
{
device_t parent = device_get_parent(dev);
struct ata_pci_controller *ctlr = device_get_softc(parent);
struct ata_channel *ch = device_get_softc(dev);
Add support for controllers that doesn't have the usual taskfile layout. No functional changes.
2005-04-06 10:22:56 +00:00
struct resource *io = NULL, *ctlio = NULL;
Add firstshot support for Acer Labs Inc SATA parts Update ALi PATA support to handle ATA133 as well
2005-04-05 14:51:43 +00:00
int unit01 = (ch->unit & 1), unit10 = (ch->unit & 2);
int i, rid;
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
Add firstshot support for Acer Labs Inc SATA parts Update ALi PATA support to handle ATA133 as well
2005-04-05 14:51:43 +00:00
rid = PCIR_BAR(0) + (unit01 ? 8 : 0);
io = bus_alloc_resource_any(parent, SYS_RES_IOPORT, &rid, RF_ACTIVE);
if (!io)
return ENXIO;
rid = PCIR_BAR(1) + (unit01 ? 8 : 0);
Add support for controllers that doesn't have the usual taskfile layout. No functional changes.
2005-04-06 10:22:56 +00:00
ctlio = bus_alloc_resource_any(parent, SYS_RES_IOPORT, &rid, RF_ACTIVE);
if (!ctlio) {
Add firstshot support for Acer Labs Inc SATA parts Update ALi PATA support to handle ATA133 as well
2005-04-05 14:51:43 +00:00
bus_release_resource(dev, SYS_RES_IOPORT, ATA_IOADDR_RID, io);
return ENXIO;
}
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
Add support for controllers that doesn't have the usual taskfile layout. No functional changes.
2005-04-06 10:22:56 +00:00
for (i = ATA_DATA; i <= ATA_COMMAND; i ++) {
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ch->r_io[i].res = io;
ch->r_io[i].offset = i + (unit10 ? 8 : 0);
Add firstshot support for Acer Labs Inc SATA parts Update ALi PATA support to handle ATA133 as well
2005-04-05 14:51:43 +00:00
}
Add support for controllers that doesn't have the usual taskfile layout. No functional changes.
2005-04-06 10:22:56 +00:00
ch->r_io[ATA_CONTROL].res = ctlio;
ch->r_io[ATA_CONTROL].offset = 2 + (unit10 ? 4 : 0);
Add firstshot support for Acer Labs Inc SATA parts Update ALi PATA support to handle ATA133 as well
2005-04-05 14:51:43 +00:00
ch->r_io[ATA_IDX_ADDR].res = io;
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_default_registers(dev);
Add firstshot support for Acer Labs Inc SATA parts Update ALi PATA support to handle ATA133 as well
2005-04-05 14:51:43 +00:00
if (ctlr->r_res1) {
for (i = ATA_BMCMD_PORT; i <= ATA_BMDTP_PORT; i++) {
ch->r_io[i].res = ctlr->r_res1;
ch->r_io[i].offset = (i - ATA_BMCMD_PORT)+(ch->unit * ATA_BMIOSIZE);
}
}
Add support for controllers that doesn't have the usual taskfile layout. No functional changes.
2005-04-06 10:22:56 +00:00
ch->flags |= ATA_NO_SLAVE;
Only try to allocate and use the SATA resource if they are enabled by the BIOS. It seems some BIOS's doesn't get this right, and that would result in ATA panic'ing.
2005-04-25 07:50:51 +00:00
/* XXX SOS PHY handling awkward in ALI chip not supported yet */
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ata_pci_hw(dev);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
return 0;
}
Apply fix for "pr82261 DMA-support on Sparc64 broken" The Acer chip or wiring that SUN uses has problems that this patch tries to work around. Original patch by Marius Strobl, hacked into shape by me..
2005-08-23 08:53:01 +00:00
static void
ata_ali_reset(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(device_get_parent(dev));
struct ata_channel *ch = device_get_softc(dev);
device_t *children;
int nchildren, i;
ata_generic_reset(dev);
/*
* workaround for datacorruption bug found on at least SUN Blade-100
* find the ISA function on the southbridge and disable then enable
* the ATA channel tristate buffer
*/
if (ctlr->chip->chiprev == 0xc3 || ctlr->chip->chiprev == 0xc2) {
if (!device_get_children(GRANDPARENT(dev), &children, &nchildren)) {
for (i = 0; i < nchildren; i++) {
if (pci_get_devid(children[i]) == ATA_ALI_1533) {
pci_write_config(children[i], 0x58,
pci_read_config(children[i], 0x58, 1) &
~(0x04 << ch->unit), 1);
pci_write_config(children[i], 0x58,
pci_read_config(children[i], 0x58, 1) |
(0x04 << ch->unit), 1);
break;
}
}
free(children, M_TEMP);
}
}
}
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
static void
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_ali_setmode(device_t dev, int mode)
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
{
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
device_t gparent = GRANDPARENT(dev);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
struct ata_pci_controller *ctlr = device_get_softc(gparent);
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
struct ata_channel *ch = device_get_softc(device_get_parent(dev));
struct ata_device *atadev = device_get_softc(dev);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
int devno = (ch->unit << 1) + ATA_DEV(atadev->unit);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
int error;
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
mode = ata_limit_mode(dev, mode, ctlr->chip->max_dma);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
Fix support for Acer chips. The UDMA enable regs wasn't set properly.
2003-02-25 21:22:27 +00:00
if (ctlr->chip->cfg2 & ALINEW) {
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (mode > ATA_UDMA2 &&
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_read_config(gparent, 0x4a, 1) & (1 << ch->unit)) {
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_print_cable(dev, "controller");
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
mode = ATA_UDMA2;
}
}
else
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
mode = ata_check_80pin(dev, mode);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (ctlr->chip->cfg2 & ALIOLD) {
/* doesn't support ATAPI DMA on write */
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
ch->flags |= ATA_ATAPI_DMA_RO;
if (ch->devices & ATA_ATAPI_MASTER && ch->devices & ATA_ATAPI_SLAVE) {
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
/* doesn't support ATAPI DMA on two ATAPI devices */
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
device_printf(dev, "two atapi devices on this channel, no DMA\n");
mode = ata_limit_mode(dev, mode, ATA_PIO_MAX);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
}
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
error = ata_controlcmd(dev, ATA_SETFEATURES, ATA_SF_SETXFER, 0, mode);
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (bootverbose)
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
device_printf(dev, "%ssetting %s on %s chip\n",
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
(error) ? "FAILURE " : "",
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
ata_mode2str(mode), ctlr->chip->text);
if (!error) {
if (mode >= ATA_UDMA0) {
Add firstshot support for Acer Labs Inc SATA parts Update ALi PATA support to handle ATA133 as well
2005-04-05 14:51:43 +00:00
u_int8_t udma[] = {0x0c, 0x0b, 0x0a, 0x09, 0x08, 0x0f, 0x0d};
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
u_int32_t word54 = pci_read_config(gparent, 0x54, 4);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
Fix chipset setup problem on older ALI chips.
2003-02-23 16:36:21 +00:00
word54 &= ~(0x000f000f << (devno << 2));
word54 |= (((udma[mode&ATA_MODE_MASK]<<16)|0x05)<<(devno<<2));
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, 0x54, word54, 4);
pci_write_config(gparent, 0x58 + (ch->unit << 2),
Fix support for Acer chips. The UDMA enable regs wasn't set properly.
2003-02-25 21:22:27 +00:00
0x00310001, 4);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
else {
u_int32_t piotimings[] =
{ 0x006d0003, 0x00580002, 0x00440001, 0x00330001,
0x00310001, 0x00440001, 0x00330001, 0x00310001};
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, 0x54, pci_read_config(gparent, 0x54, 4) &
~(0x0008000f << (devno << 2)), 4);
pci_write_config(gparent, 0x58 + (ch->unit << 2),
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
piotimings[ata_mode2idx(mode)], 4);
}
atadev->mode = mode;
}
}
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
/*
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
* American Micro Devices (AMD) chipset support functions
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
*/
int
ata_amd_ident(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
struct ata_chip_id *idx;
static struct ata_chip_id ids[] =
Cleanup the strings for printing chipset names, saves quite some space. Add a couble new nVidia chips now we are here.
2005-12-16 08:12:13 +00:00
{{ ATA_AMD756, 0x00, AMDNVIDIA, 0x00, ATA_UDMA4, "756" },
{ ATA_AMD766, 0x00, AMDNVIDIA, AMDCABLE|AMDBUG, ATA_UDMA5, "766" },
{ ATA_AMD768, 0x00, AMDNVIDIA, AMDCABLE, ATA_UDMA5, "768" },
{ ATA_AMD8111, 0x00, AMDNVIDIA, AMDCABLE, ATA_UDMA6, "8111" },
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
{ 0, 0, 0, 0, 0, 0}};
char buffer[64];
Support the ICH5 SATA part. Fix HPT374 UDMA133 timing. Fix Promise ID. Cosmetics on probe print for Promise & HPT. Approved by: re
2003-05-18 16:45:48 +00:00
if (!(idx = ata_match_chip(dev, ids)))
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
return ENXIO;
Cleanup the strings for printing chipset names, saves quite some space. Add a couble new nVidia chips now we are here.
2005-12-16 08:12:13 +00:00
sprintf(buffer, "AMD %s %s controller",
idx->text, ata_mode2str(idx->max_dma));
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
device_set_desc_copy(dev, buffer);
ctlr->chip = idx;
ctlr->chipinit = ata_amd_chipinit;
return 0;
}
static int
ata_amd_chipinit(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
if (ata_setup_interrupt(dev))
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
return ENXIO;
Cleanup the defines used for various chipsets. Add new nVidia controllers.
2004-09-03 07:37:53 +00:00
/* disable/set prefetch, postwrite */
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (ctlr->chip->cfg2 & AMDBUG)
pci_write_config(dev, 0x41, pci_read_config(dev, 0x41, 1) & 0x0f, 1);
else
pci_write_config(dev, 0x41, pci_read_config(dev, 0x41, 1) | 0xf0, 1);
ctlr->setmode = ata_via_family_setmode;
return 0;
}
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
Add support for the ATI IXP[234]00 series chipsets. HW donated by: sentex
2005-10-12 20:00:26 +00:00
/*
* ATI chipset support functions
*/
int
ata_ati_ident(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
struct ata_chip_id *idx;
static struct ata_chip_id ids[] =
Cleanup the strings for printing chipset names, saves quite some space. Add a couble new nVidia chips now we are here.
2005-12-16 08:12:13 +00:00
{{ ATA_ATI_IXP200, 0x00, 0, 0, ATA_UDMA5, "IXP200" },
{ ATA_ATI_IXP300, 0x00, 0, 0, ATA_UDMA6, "IXP300" },
{ ATA_ATI_IXP400, 0x00, 0, 0, ATA_UDMA6, "IXP400" },
{ ATA_ATI_IXP300_S1, 0x00, SIIMEMIO, 0, ATA_SA150, "IXP300" },
{ ATA_ATI_IXP400_S1, 0x00, SIIMEMIO, 0, ATA_SA150, "IXP400" },
{ ATA_ATI_IXP400_S2, 0x00, SIIMEMIO, 0, ATA_SA150, "IXP400" },
Add support for the ATI IXP[234]00 series chipsets. HW donated by: sentex
2005-10-12 20:00:26 +00:00
{ 0, 0, 0, 0, 0, 0}};
char buffer[64];
if (!(idx = ata_match_chip(dev, ids)))
return ENXIO;
Cleanup the strings for printing chipset names, saves quite some space. Add a couble new nVidia chips now we are here.
2005-12-16 08:12:13 +00:00
sprintf(buffer, "ATI %s %s controller",
idx->text, ata_mode2str(idx->max_dma));
Add support for the ATI IXP[234]00 series chipsets. HW donated by: sentex
2005-10-12 20:00:26 +00:00
device_set_desc_copy(dev, buffer);
ctlr->chip = idx;
/* the ATI SATA controller is actually a SiI 3112 controller*/
if (ctlr->chip->cfg1 & SIIMEMIO)
ctlr->chipinit = ata_sii_chipinit;
else
ctlr->chipinit = ata_ati_chipinit;
return 0;
}
static int
ata_ati_chipinit(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
if (ata_setup_interrupt(dev))
return ENXIO;
ctlr->setmode = ata_ati_setmode;
return 0;
}
static void
ata_ati_setmode(device_t dev, int mode)
{
device_t gparent = GRANDPARENT(dev);
struct ata_pci_controller *ctlr = device_get_softc(gparent);
struct ata_channel *ch = device_get_softc(device_get_parent(dev));
struct ata_device *atadev = device_get_softc(dev);
int devno = (ch->unit << 1) + ATA_DEV(atadev->unit);
int offset = (devno ^ 0x01) << 3;
int error;
u_int8_t piotimings[] = { 0x5d, 0x47, 0x34, 0x22, 0x20, 0x34, 0x22, 0x20,
0x20, 0x20, 0x20, 0x20, 0x20, 0x20 };
u_int8_t dmatimings[] = { 0x77, 0x21, 0x20 };
mode = ata_limit_mode(dev, mode, ctlr->chip->max_dma);
mode = ata_check_80pin(dev, mode);
error = ata_controlcmd(dev, ATA_SETFEATURES, ATA_SF_SETXFER, 0, mode);
if (bootverbose)
device_printf(dev, "%ssetting %s on %s chip\n",
(error) ? "FAILURE " : "",
ata_mode2str(mode), ctlr->chip->text);
if (!error) {
if (mode >= ATA_UDMA0) {
pci_write_config(gparent, 0x56,
(pci_read_config(gparent, 0x56, 2) &
~(0xf << (devno << 2))) |
((mode & ATA_MODE_MASK) << (devno << 2)), 2);
pci_write_config(gparent, 0x54,
pci_read_config(gparent, 0x54, 1) |
(0x01 << devno), 1);
pci_write_config(gparent, 0x44,
(pci_read_config(gparent, 0x44, 4) &
~(0xff << offset)) |
(dmatimings[2] << offset), 4);
}
else if (mode >= ATA_WDMA0) {
pci_write_config(gparent, 0x54,
pci_read_config(gparent, 0x54, 1) &
~(0x01 << devno), 1);
pci_write_config(gparent, 0x44,
(pci_read_config(gparent, 0x44, 4) &
~(0xff << offset)) |
(dmatimings[mode & ATA_MODE_MASK] << offset), 4);
}
else
pci_write_config(gparent, 0x54,
pci_read_config(gparent, 0x54, 1) &
~(0x01 << devno), 1);
pci_write_config(gparent, 0x4a,
(pci_read_config(gparent, 0x4a, 2) &
~(0xf << (devno << 2))) |
(((mode - ATA_PIO0) & ATA_MODE_MASK) << (devno<<2)),2);
pci_write_config(gparent, 0x40,
(pci_read_config(gparent, 0x40, 4) &
~(0xff << offset)) |
(piotimings[ata_mode2idx(mode)] << offset), 4);
atadev->mode = mode;
}
}
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
/*
* Cyrix chipset support functions
*/
int
ata_cyrix_ident(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
if (pci_get_devid(dev) == ATA_CYRIX_5530) {
device_set_desc(dev, "Cyrix 5530 ATA33 controller");
ctlr->chipinit = ata_cyrix_chipinit;
return 0;
}
return ENXIO;
}
static int
ata_cyrix_chipinit(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
if (ata_setup_interrupt(dev))
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
return ENXIO;
Add support for detaching PCI controllers. This adds support for cardbus ATA/SATA controllers. I get roughly the same transfer speeds as on true PCI controllers. Nice to be able to add a couble of "real" disks to a laptop :)
2004-03-15 12:03:48 +00:00
if (ctlr->r_res1)
Dont use r_bmio on Cyrix chips if not set from BIOS.
2003-03-10 08:20:14 +00:00
ctlr->setmode = ata_cyrix_setmode;
else
ctlr->setmode = ata_generic_setmode;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
return 0;
}
static void
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_cyrix_setmode(device_t dev, int mode)
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
{
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
struct ata_channel *ch = device_get_softc(device_get_parent(dev));
struct ata_device *atadev = device_get_softc(dev);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
int devno = (ch->unit << 1) + ATA_DEV(atadev->unit);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
u_int32_t piotiming[] =
{ 0x00009172, 0x00012171, 0x00020080, 0x00032010, 0x00040010 };
u_int32_t dmatiming[] = { 0x00077771, 0x00012121, 0x00002020 };
u_int32_t udmatiming[] = { 0x00921250, 0x00911140, 0x00911030 };
int error;
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
ch->dma->alignment = 16;
ch->dma->max_iosize = 126 * DEV_BSIZE;
Up the alignment requirement of the SC1100 to 16 bytes, it has all the bugs and then some of its Cyrix inheritance. Set the max_iosize for the Cyrix to 63K as the SC1100.
2003-10-21 08:53:29 +00:00
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
mode = ata_limit_mode(dev, mode, ATA_UDMA2);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
error = ata_controlcmd(dev, ATA_SETFEATURES, ATA_SF_SETXFER, 0, mode);
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (bootverbose)
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
device_printf(dev, "%ssetting %s on Cyrix chip\n",
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
(error) ? "FAILURE " : "", ata_mode2str(mode));
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (!error) {
if (mode >= ATA_UDMA0) {
Second round of updates to the ATA driver. Clean up the DMA interface too much unneeded stuff crept in with the busdma code back when. Modify the ATA_IN* / ATA_OUT* macros so that resource and offset are gotten from a table. That allows for new chipsets that doesn't nessesarily have things ordered the good old way. This also removes the need for the wierd PC98 resource functions. Tested on: i386, PC98, Alpha, Sparc64
2003-03-29 13:37:09 +00:00
ATA_OUTL(ch->r_io[ATA_BMCMD_PORT].res,
Fix typo (that even got cut/pasted 2 times) Found by: phk Approved by: re@
2003-05-12 16:43:13 +00:00
0x24 + (devno << 3), udmatiming[mode & ATA_MODE_MASK]);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
else if (mode >= ATA_WDMA0) {
Second round of updates to the ATA driver. Clean up the DMA interface too much unneeded stuff crept in with the busdma code back when. Modify the ATA_IN* / ATA_OUT* macros so that resource and offset are gotten from a table. That allows for new chipsets that doesn't nessesarily have things ordered the good old way. This also removes the need for the wierd PC98 resource functions. Tested on: i386, PC98, Alpha, Sparc64
2003-03-29 13:37:09 +00:00
ATA_OUTL(ch->r_io[ATA_BMCMD_PORT].res,
Fix typo (that even got cut/pasted 2 times) Found by: phk Approved by: re@
2003-05-12 16:43:13 +00:00
0x24 + (devno << 3), dmatiming[mode & ATA_MODE_MASK]);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
else {
Second round of updates to the ATA driver. Clean up the DMA interface too much unneeded stuff crept in with the busdma code back when. Modify the ATA_IN* / ATA_OUT* macros so that resource and offset are gotten from a table. That allows for new chipsets that doesn't nessesarily have things ordered the good old way. This also removes the need for the wierd PC98 resource functions. Tested on: i386, PC98, Alpha, Sparc64
2003-03-29 13:37:09 +00:00
ATA_OUTL(ch->r_io[ATA_BMCMD_PORT].res,
Fix typo (that even got cut/pasted 2 times) Found by: phk Approved by: re@
2003-05-12 16:43:13 +00:00
0x20 + (devno << 3), piotiming[mode & ATA_MODE_MASK]);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
atadev->mode = mode;
}
}
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
/*
* Cypress chipset support functions
*/
int
ata_cypress_ident(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
/*
* the Cypress chip is a mess, it contains two ATA functions, but
* both channels are visible on the first one.
* simply ignore the second function for now, as the right
* solution (ignoring the second channel on the first function)
* doesn't work with the crappy ATA interrupt setup on the alpha.
*/
if (pci_get_devid(dev) == ATA_CYPRESS_82C693 &&
Only use function 1 not 0 on cypress chips.
2003-02-27 09:42:57 +00:00
pci_get_function(dev) == 1 &&
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
pci_get_subclass(dev) == PCIS_STORAGE_IDE) {
device_set_desc(dev, "Cypress 82C693 ATA controller");
ctlr->chipinit = ata_cypress_chipinit;
return 0;
}
return ENXIO;
}
static int
ata_cypress_chipinit(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
if (ata_setup_interrupt(dev))
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
return ENXIO;
ctlr->setmode = ata_cypress_setmode;
return 0;
}
static void
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_cypress_setmode(device_t dev, int mode)
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
{
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
device_t gparent = GRANDPARENT(dev);
struct ata_channel *ch = device_get_softc(device_get_parent(dev));
struct ata_device *atadev = device_get_softc(dev);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
int error;
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
mode = ata_limit_mode(dev, mode, ATA_WDMA2);
Only use function 1 not 0 on cypress chips.
2003-02-27 09:42:57 +00:00
Cleanup comments
2005-05-13 10:25:19 +00:00
/* XXX SOS missing WDMA0+1 + PIO modes */
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (mode == ATA_WDMA2) {
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
error = ata_controlcmd(dev, ATA_SETFEATURES, ATA_SF_SETXFER, 0, mode);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (bootverbose)
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
device_printf(dev, "%ssetting WDMA2 on Cypress chip\n",
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
error ? "FAILURE " : "");
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (!error) {
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, ch->unit ? 0x4e : 0x4c, 0x2020, 2);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
atadev->mode = mode;
return;
}
}
/* we could set PIO mode timings, but we assume the BIOS did that */
}
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
/*
* HighPoint chipset support functions
*/
int
ata_highpoint_ident(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
struct ata_chip_id *idx;
static struct ata_chip_id ids[] =
Cleanup the strings for printing chipset names, saves quite some space. Add a couble new nVidia chips now we are here.
2005-12-16 08:12:13 +00:00
{{ ATA_HPT374, 0x07, HPT374, 0x00, ATA_UDMA6, "HPT374" },
{ ATA_HPT372, 0x02, HPT372, 0x00, ATA_UDMA6, "HPT372N" },
{ ATA_HPT372, 0x01, HPT372, 0x00, ATA_UDMA6, "HPT372" },
{ ATA_HPT371, 0x01, HPT372, 0x00, ATA_UDMA6, "HPT371" },
{ ATA_HPT366, 0x05, HPT372, 0x00, ATA_UDMA6, "HPT372" },
{ ATA_HPT366, 0x03, HPT370, 0x00, ATA_UDMA5, "HPT370" },
{ ATA_HPT366, 0x02, HPT366, 0x00, ATA_UDMA4, "HPT368" },
{ ATA_HPT366, 0x00, HPT366, HPTOLD, ATA_UDMA4, "HPT366" },
{ ATA_HPT302, 0x01, HPT372, 0x00, ATA_UDMA6, "HPT302" },
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
{ 0, 0, 0, 0, 0, 0}};
char buffer[64];
Support the ICH5 SATA part. Fix HPT374 UDMA133 timing. Fix Promise ID. Cosmetics on probe print for Promise & HPT. Approved by: re
2003-05-18 16:45:48 +00:00
if (!(idx = ata_match_chip(dev, ids)))
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
return ENXIO;
Fix probe printing of highpoint and promise controllers.
2006-05-11 10:43:08 +00:00
strcpy(buffer, "HighPoint ");
strcat(buffer, idx->text);
Support the ICH5 SATA part. Fix HPT374 UDMA133 timing. Fix Promise ID. Cosmetics on probe print for Promise & HPT. Approved by: re
2003-05-18 16:45:48 +00:00
if (idx->cfg1 == HPT374) {
if (pci_get_function(dev) == 0)
strcat(buffer, " (channel 0+1)");
Fix probe printing of highpoint and promise controllers.
2006-05-11 10:43:08 +00:00
if (pci_get_function(dev) == 1)
Support the ICH5 SATA part. Fix HPT374 UDMA133 timing. Fix Promise ID. Cosmetics on probe print for Promise & HPT. Approved by: re
2003-05-18 16:45:48 +00:00
strcat(buffer, " (channel 2+3)");
}
Fix probe printing of highpoint and promise controllers.
2006-05-11 10:43:08 +00:00
sprintf(buffer, "%s %s controller", buffer, ata_mode2str(idx->max_dma));
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
device_set_desc_copy(dev, buffer);
ctlr->chip = idx;
ctlr->chipinit = ata_highpoint_chipinit;
return 0;
}
static int
ata_highpoint_chipinit(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
if (ata_setup_interrupt(dev))
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
return ENXIO;
if (ctlr->chip->cfg2 == HPTOLD) {
Increase robustness of SATA handling.
2004-06-15 11:02:09 +00:00
/* disable interrupt prediction */
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
pci_write_config(dev, 0x51, (pci_read_config(dev, 0x51, 1) & ~0x80), 1);
}
else {
Increase robustness of SATA handling.
2004-06-15 11:02:09 +00:00
/* disable interrupt prediction */
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
pci_write_config(dev, 0x51, (pci_read_config(dev, 0x51, 1) & ~0x03), 1);
pci_write_config(dev, 0x55, (pci_read_config(dev, 0x55, 1) & ~0x03), 1);
Increase robustness of SATA handling.
2004-06-15 11:02:09 +00:00
/* enable interrupts */
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
pci_write_config(dev, 0x5a, (pci_read_config(dev, 0x5a, 1) & ~0x10), 1);
/* set clocks etc */
if (ctlr->chip->cfg1 < HPT372)
pci_write_config(dev, 0x5b, 0x22, 1);
else
pci_write_config(dev, 0x5b,
(pci_read_config(dev, 0x5b, 1) & 0x01) | 0x20, 1);
}
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ctlr->allocate = ata_highpoint_allocate;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
ctlr->setmode = ata_highpoint_setmode;
return 0;
}
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static int
ata_highpoint_allocate(device_t dev)
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
{
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
struct ata_channel *ch = device_get_softc(dev);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
/* setup the usual register normal pci style */
if (ata_pci_allocate(dev))
return ENXIO;
ch->flags |= ATA_ALWAYS_DMASTAT;
return 0;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
static void
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_highpoint_setmode(device_t dev, int mode)
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
{
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
device_t gparent = GRANDPARENT(dev);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
struct ata_pci_controller *ctlr = device_get_softc(gparent);
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
struct ata_channel *ch = device_get_softc(device_get_parent(dev));
struct ata_device *atadev = device_get_softc(dev);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
int devno = (ch->unit << 1) + ATA_DEV(atadev->unit);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
int error;
u_int32_t timings33[][4] = {
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
/* HPT366 HPT370 HPT372 HPT374 mode */
{ 0x40d0a7aa, 0x06914e57, 0x0d029d5e, 0x0ac1f48a }, /* PIO 0 */
{ 0x40d0a7a3, 0x06914e43, 0x0d029d26, 0x0ac1f465 }, /* PIO 1 */
{ 0x40d0a753, 0x06514e33, 0x0c829ca6, 0x0a81f454 }, /* PIO 2 */
{ 0x40c8a742, 0x06514e22, 0x0c829c84, 0x0a81f443 }, /* PIO 3 */
{ 0x40c8a731, 0x06514e21, 0x0c829c62, 0x0a81f442 }, /* PIO 4 */
{ 0x20c8a797, 0x26514e97, 0x2c82922e, 0x228082ea }, /* MWDMA 0 */
{ 0x20c8a732, 0x26514e33, 0x2c829266, 0x22808254 }, /* MWDMA 1 */
{ 0x20c8a731, 0x26514e21, 0x2c829262, 0x22808242 }, /* MWDMA 2 */
{ 0x10c8a731, 0x16514e31, 0x1c829c62, 0x121882ea }, /* UDMA 0 */
{ 0x10cba731, 0x164d4e31, 0x1c9a9c62, 0x12148254 }, /* UDMA 1 */
{ 0x10caa731, 0x16494e31, 0x1c929c62, 0x120c8242 }, /* UDMA 2 */
{ 0x10cfa731, 0x166d4e31, 0x1c8e9c62, 0x128c8242 }, /* UDMA 3 */
{ 0x10c9a731, 0x16454e31, 0x1c8a9c62, 0x12ac8242 }, /* UDMA 4 */
{ 0, 0x16454e31, 0x1c8a9c62, 0x12848242 }, /* UDMA 5 */
{ 0, 0, 0x1c869c62, 0x12808242 } /* UDMA 6 */
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
};
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
mode = ata_limit_mode(dev, mode, ctlr->chip->max_dma);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
if (ctlr->chip->cfg1 == HPT366 && ata_atapi(dev))
mode = ata_limit_mode(dev, mode, ATA_PIO_MAX);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
mode = ata_highpoint_check_80pin(dev, mode);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
/*
* most if not all HPT chips cant really handle that the device is
* running at ATA_UDMA6/ATA133 speed, so we cheat at set the device to
* a max of ATA_UDMA5/ATA100 to guard against suboptimal performance
*/
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
error = ata_controlcmd(dev, ATA_SETFEATURES, ATA_SF_SETXFER, 0,
ata_limit_mode(dev, mode, ATA_UDMA5));
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (bootverbose)
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
device_printf(dev, "%ssetting %s on HighPoint chip\n",
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
(error) ? "FAILURE " : "", ata_mode2str(mode));
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (!error)
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, 0x40 + (devno << 2),
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
timings33[ata_mode2idx(mode)][ctlr->chip->cfg1], 4);
atadev->mode = mode;
}
static int
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_highpoint_check_80pin(device_t dev, int mode)
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
{
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
device_t gparent = GRANDPARENT(dev);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
struct ata_pci_controller *ctlr = device_get_softc(gparent);
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
struct ata_channel *ch = device_get_softc(device_get_parent(dev));
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
u_int8_t reg, val, res;
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
if (ctlr->chip->cfg1 == HPT374 && pci_get_function(gparent) == 1) {
reg = ch->unit ? 0x57 : 0x53;
val = pci_read_config(gparent, reg, 1);
pci_write_config(gparent, reg, val | 0x80, 1);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
else {
reg = 0x5b;
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
val = pci_read_config(gparent, reg, 1);
pci_write_config(gparent, reg, val & 0xfe, 1);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
res = pci_read_config(gparent, 0x5a, 1) & (ch->unit ? 0x1:0x2);
pci_write_config(gparent, reg, val, 1);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (mode > ATA_UDMA2 && res) {
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_print_cable(dev, "controller");
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
mode = ATA_UDMA2;
}
return mode;
}
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
/*
* Intel chipset support functions
*/
int
ata_intel_ident(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
struct ata_chip_id *idx;
static struct ata_chip_id ids[] =
Add support for the ICH8 and ESB2 chips, also add a few other missing ICH7 parts. HW donated by: JMicron
2006-08-14 19:39:33 +00:00
{{ ATA_I82371FB, 0, 0, 0x00, ATA_WDMA2, "PIIX" },
{ ATA_I82371SB, 0, 0, 0x00, ATA_WDMA2, "PIIX3" },
{ ATA_I82371AB, 0, 0, 0x00, ATA_UDMA2, "PIIX4" },
{ ATA_I82443MX, 0, 0, 0x00, ATA_UDMA2, "PIIX4" },
{ ATA_I82451NX, 0, 0, 0x00, ATA_UDMA2, "PIIX4" },
{ ATA_I82801AB, 0, 0, 0x00, ATA_UDMA2, "ICH0" },
{ ATA_I82801AA, 0, 0, 0x00, ATA_UDMA4, "ICH" },
{ ATA_I82372FB, 0, 0, 0x00, ATA_UDMA4, "ICH" },
{ ATA_I82801BA, 0, 0, 0x00, ATA_UDMA5, "ICH2" },
{ ATA_I82801BA_1, 0, 0, 0x00, ATA_UDMA5, "ICH2" },
{ ATA_I82801CA, 0, 0, 0x00, ATA_UDMA5, "ICH3" },
{ ATA_I82801CA_1, 0, 0, 0x00, ATA_UDMA5, "ICH3" },
{ ATA_I82801DB, 0, 0, 0x00, ATA_UDMA5, "ICH4" },
{ ATA_I82801DB_1, 0, 0, 0x00, ATA_UDMA5, "ICH4" },
{ ATA_I82801EB, 0, 0, 0x00, ATA_UDMA5, "ICH5" },
{ ATA_I82801EB_S1, 0, 0, 0x00, ATA_SA150, "ICH5" },
{ ATA_I82801EB_R1, 0, 0, 0x00, ATA_SA150, "ICH5" },
{ ATA_I6300ESB, 0, 0, 0x00, ATA_UDMA5, "6300ESB" },
{ ATA_I6300ESB_S1, 0, 0, 0x00, ATA_SA150, "6300ESB" },
{ ATA_I6300ESB_R1, 0, 0, 0x00, ATA_SA150, "6300ESB" },
{ ATA_I82801FB, 0, 0, 0x00, ATA_UDMA5, "ICH6" },
{ ATA_I82801FB_S1, 0, AHCI, 0x00, ATA_SA150, "ICH6" },
{ ATA_I82801FB_R1, 0, AHCI, 0x00, ATA_SA150, "ICH6" },
{ ATA_I82801FBM, 0, AHCI, 0x00, ATA_SA150, "ICH6M" },
{ ATA_I82801GB, 0, 0, 0x00, ATA_UDMA5, "ICH7" },
{ ATA_I82801GB_S1, 0, AHCI, 0x00, ATA_SA300, "ICH7" },
{ ATA_I82801GB_R1, 0, AHCI, 0x00, ATA_SA300, "ICH7" },
{ ATA_I82801GB_AH, 0, AHCI, 0x00, ATA_SA300, "ICH7" },
{ ATA_I82801GBM_S1, 0, AHCI, 0x00, ATA_SA300, "ICH7M" },
{ ATA_I82801GBM_R1, 0, AHCI, 0x00, ATA_SA300, "ICH7M" },
{ ATA_I82801GBM_AH, 0, AHCI, 0x00, ATA_SA300, "ICH7M" },
{ ATA_I63XXESB2, 0, 0, 0x00, ATA_UDMA5, "63XXESB2" },
{ ATA_I63XXESB2_S1, 0, AHCI, 0x00, ATA_SA300, "63XXESB2" },
{ ATA_I63XXESB2_S2, 0, AHCI, 0x00, ATA_SA300, "63XXESB2" },
{ ATA_I63XXESB2_R1, 0, AHCI, 0x00, ATA_SA300, "63XXESB2" },
{ ATA_I63XXESB2_R2, 0, AHCI, 0x00, ATA_SA300, "63XXESB2" },
{ ATA_I82801HB_S1, 0, AHCI, 0x00, ATA_SA300, "ICH8" },
{ ATA_I82801HB_S2, 0, AHCI, 0x00, ATA_SA300, "ICH8" },
{ ATA_I82801HB_R1, 0, AHCI, 0x00, ATA_SA300, "ICH8" },
{ ATA_I82801HB_AH4, 0, AHCI, 0x00, ATA_SA300, "ICH8" },
{ ATA_I82801HB_AH6, 0, AHCI, 0x00, ATA_SA300, "ICH8" },
{ ATA_I82801HBM_S1, 0, AHCI, 0x00, ATA_SA300, "ICH8M" },
{ ATA_I82801HBM_S2, 0, AHCI, 0x00, ATA_SA300, "ICH8M" },
{ ATA_I31244, 0, 0, 0x00, ATA_SA150, "31244" },
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
{ 0, 0, 0, 0, 0, 0}};
char buffer[64];
Support the ICH5 SATA part. Fix HPT374 UDMA133 timing. Fix Promise ID. Cosmetics on probe print for Promise & HPT. Approved by: re
2003-05-18 16:45:48 +00:00
if (!(idx = ata_match_chip(dev, ids)))
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
return ENXIO;
Cleanup the strings for printing chipset names, saves quite some space. Add a couble new nVidia chips now we are here.
2005-12-16 08:12:13 +00:00
sprintf(buffer, "Intel %s %s controller",
idx->text, ata_mode2str(idx->max_dma));
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
device_set_desc_copy(dev, buffer);
ctlr->chip = idx;
ctlr->chipinit = ata_intel_chipinit;
return 0;
}
static int
ata_intel_chipinit(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
if (ata_setup_interrupt(dev))
return ENXIO;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
/* good old PIIX needs special treatment (not implemented) */
Increase robustness of SATA handling.
2004-06-15 11:02:09 +00:00
if (ctlr->chip->chipid == ATA_I82371FB) {
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
ctlr->setmode = ata_intel_old_setmode;
Increase robustness of SATA handling.
2004-06-15 11:02:09 +00:00
}
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
/* the intel 31244 needs special care if in DPA mode */
else if (ctlr->chip->chipid == ATA_I31244) {
if (pci_get_subclass(dev) != PCIS_STORAGE_IDE) {
ctlr->r_type2 = SYS_RES_MEMORY;
ctlr->r_rid2 = PCIR_BAR(0);
if (!(ctlr->r_res2 = bus_alloc_resource_any(dev, ctlr->r_type2,
Whitespace cleanup.
2006-01-18 13:10:17 +00:00
&ctlr->r_rid2,
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
RF_ACTIVE)))
Whitespace cleanup.
2006-01-18 13:10:17 +00:00
return ENXIO;
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
ctlr->channels = 4;
ctlr->allocate = ata_intel_31244_allocate;
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ctlr->reset = ata_intel_31244_reset;
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
}
ctlr->setmode = ata_sata_setmode;
}
/* non SATA intel chips goes here */
Increase robustness of SATA handling.
2004-06-15 11:02:09 +00:00
else if (ctlr->chip->max_dma < ATA_SA150) {
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ctlr->allocate = ata_intel_allocate;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
ctlr->setmode = ata_intel_new_setmode;
Increase robustness of SATA handling.
2004-06-15 11:02:09 +00:00
}
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
/* SATA parts can be either compat or AHCI */
Increase robustness of SATA handling.
2004-06-15 11:02:09 +00:00
else {
Update the ICH7 support so it deals better with chips without AHCI. Update Intel MatrixRAID support to be able to pick up RAID0+1 (RAID10) and RAID5 arrays without panic'ing. This has the side effect of now also supporting multiple volumes on MatrixRAID's now I have the metadata better understood.. HW sponsored by: Mullet Scandinavia AB
2005-12-02 10:13:53 +00:00
/* force all ports active "the legacy way" */
pci_write_config(dev, 0x92, pci_read_config(dev, 0x92, 2) | 0x0f,2);
Add support for AHCI compliant ATA devices. For now just support the Intel ICH6 as that the HW at hand. Sponsored by: pair.com
2005-05-11 16:10:08 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ctlr->allocate = ata_intel_allocate;
Update the ICH7 support so it deals better with chips without AHCI. Update Intel MatrixRAID support to be able to pick up RAID0+1 (RAID10) and RAID5 arrays without panic'ing. This has the side effect of now also supporting multiple volumes on MatrixRAID's now I have the metadata better understood.. HW sponsored by: Mullet Scandinavia AB
2005-12-02 10:13:53 +00:00
ctlr->reset = ata_intel_reset;
Add support for AHCI compliant ATA devices. For now just support the Intel ICH6 as that the HW at hand. Sponsored by: pair.com
2005-05-11 16:10:08 +00:00
Update AHCI support to be more generic. Add support for AHCI on the VIA VT8251.
2006-06-28 09:59:09 +00:00
/*
* if we have AHCI capability and BAR(5) as a memory resource
* and AHCI or RAID mode enabled in BIOS we go for AHCI mode
*/
if ((ctlr->chip->cfg1 == AHCI) &&
(pci_read_config(dev, 0x90, 1) & 0xc0)) {
Update the ICH7 support so it deals better with chips without AHCI. Update Intel MatrixRAID support to be able to pick up RAID0+1 (RAID10) and RAID5 arrays without panic'ing. This has the side effect of now also supporting multiple volumes on MatrixRAID's now I have the metadata better understood.. HW sponsored by: Mullet Scandinavia AB
2005-12-02 10:13:53 +00:00
ctlr->r_type2 = SYS_RES_MEMORY;
ctlr->r_rid2 = PCIR_BAR(5);
if ((ctlr->r_res2 = bus_alloc_resource_any(dev, ctlr->r_type2,
&ctlr->r_rid2,
Update AHCI support to be more generic. Add support for AHCI on the VIA VT8251.
2006-06-28 09:59:09 +00:00
RF_ACTIVE)))
return ata_ahci_chipinit(dev);
Add support for AHCI compliant ATA devices. For now just support the Intel ICH6 as that the HW at hand. Sponsored by: pair.com
2005-05-11 16:10:08 +00:00
}
ctlr->setmode = ata_sata_setmode;
Update the ICH7 support so it deals better with chips without AHCI. Update Intel MatrixRAID support to be able to pick up RAID0+1 (RAID10) and RAID5 arrays without panic'ing. This has the side effect of now also supporting multiple volumes on MatrixRAID's now I have the metadata better understood.. HW sponsored by: Mullet Scandinavia AB
2005-12-02 10:13:53 +00:00
/* enable PCI interrupt */
Increase robustness of SATA handling.
2004-06-15 11:02:09 +00:00
pci_write_config(dev, PCIR_COMMAND,
pci_read_config(dev, PCIR_COMMAND, 2) & ~0x0400, 2);
}
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
return 0;
}
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
static int
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ata_intel_allocate(device_t dev)
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
{
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
struct ata_channel *ch = device_get_softc(dev);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
/* setup the usual register normal pci style */
if (ata_pci_allocate(dev))
return ENXIO;
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ch->flags |= ATA_ALWAYS_DMASTAT;
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
return 0;
}
Increase robustness of SATA handling.
2004-06-15 11:02:09 +00:00
static void
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_intel_reset(device_t dev)
Increase robustness of SATA handling.
2004-06-15 11:02:09 +00:00
{
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
device_t parent = device_get_parent(dev);
Update the ICH6 support so all 4 SATA channels can be used. We still use "normal" mode, AHCI mode is in the works still. HW donated by: Sentex HW donated by: Yahoo!
2004-12-08 18:00:46 +00:00
struct ata_pci_controller *ctlr = device_get_softc(parent);
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
struct ata_channel *ch = device_get_softc(dev);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
int mask, timeout;
Increase robustness of SATA handling.
2004-06-15 11:02:09 +00:00
Update the ICH7 support so it deals better with chips without AHCI. Update Intel MatrixRAID support to be able to pick up RAID0+1 (RAID10) and RAID5 arrays without panic'ing. This has the side effect of now also supporting multiple volumes on MatrixRAID's now I have the metadata better understood.. HW sponsored by: Mullet Scandinavia AB
2005-12-02 10:13:53 +00:00
/* ICH6 & ICH7 in compat mode has 4 SATA ports as master/slave on 2 ch's */
if (ctlr->chip->cfg1) {
Update the ICH6 support so all 4 SATA channels can be used. We still use "normal" mode, AHCI mode is in the works still. HW donated by: Sentex HW donated by: Yahoo!
2004-12-08 18:00:46 +00:00
mask = (0x0005 << ch->unit);
}
else {
/* ICH5 in compat mode has SATA ports as master/slave on 1 channel */
if (pci_read_config(parent, 0x90, 1) & 0x04)
mask = 0x0003;
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
else {
Update the ICH6 support so all 4 SATA channels can be used. We still use "normal" mode, AHCI mode is in the works still. HW donated by: Sentex HW donated by: Yahoo!
2004-12-08 18:00:46 +00:00
mask = (0x0001 << ch->unit);
Update the ICH7 support so it deals better with chips without AHCI. Update Intel MatrixRAID support to be able to pick up RAID0+1 (RAID10) and RAID5 arrays without panic'ing. This has the side effect of now also supporting multiple volumes on MatrixRAID's now I have the metadata better understood.. HW sponsored by: Mullet Scandinavia AB
2005-12-02 10:13:53 +00:00
/* XXX SOS should be in intel_allocate if we grow it */
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
ch->flags |= ATA_NO_SLAVE;
}
Update the ICH6 support so all 4 SATA channels can be used. We still use "normal" mode, AHCI mode is in the works still. HW donated by: Sentex HW donated by: Yahoo!
2004-12-08 18:00:46 +00:00
}
Increase robustness of SATA handling.
2004-06-15 11:02:09 +00:00
pci_write_config(parent, 0x92, pci_read_config(parent, 0x92, 2) & ~mask, 2);
DELAY(10);
pci_write_config(parent, 0x92, pci_read_config(parent, 0x92, 2) | mask, 2);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
/* wait up to 1 sec for "connect well" */
for (timeout = 0; timeout < 100 ; timeout++) {
if (((pci_read_config(parent, 0x92, 2) & (mask << 4)) == (mask << 4)) &&
(ATA_IDX_INB(ch, ATA_STATUS) != 0xff))
break;
Introduce ata_udelay() that uses tsleep instead of DELAY if possible. In places where we have long delays that doesn't depend on too accurate timing, use ata_udelay() instead of DELAY() so we dont uselessly spin the CPU if not nessesary;
2004-09-03 12:10:44 +00:00
ata_udelay(10000);
Increase robustness of SATA handling.
2004-06-15 11:02:09 +00:00
}
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_generic_reset(dev);
Increase robustness of SATA handling.
2004-06-15 11:02:09 +00:00
}
Add support for the Intel 31244. Most code by the submitters, hammered upon to get the right ATA fell by me. Submitted by: Olivier Houchard Submitted by: Jia-Shiun Li
2005-06-08 20:02:55 +00:00
static void
ata_intel_old_setmode(device_t dev, int mode)
{
/* NOT YET */
}
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
static void
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_intel_new_setmode(device_t dev, int mode)
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
{
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
device_t gparent = GRANDPARENT(dev);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
struct ata_pci_controller *ctlr = device_get_softc(gparent);
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
struct ata_channel *ch = device_get_softc(device_get_parent(dev));
struct ata_device *atadev = device_get_softc(dev);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
int devno = (ch->unit << 1) + ATA_DEV(atadev->unit);
u_int32_t reg40 = pci_read_config(gparent, 0x40, 4);
u_int8_t reg44 = pci_read_config(gparent, 0x44, 1);
u_int8_t reg48 = pci_read_config(gparent, 0x48, 1);
u_int16_t reg4a = pci_read_config(gparent, 0x4a, 2);
u_int16_t reg54 = pci_read_config(gparent, 0x54, 2);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
u_int32_t mask40 = 0, new40 = 0;
u_int8_t mask44 = 0, new44 = 0;
int error;
u_int8_t timings[] = { 0x00, 0x00, 0x10, 0x21, 0x23, 0x10, 0x21, 0x23,
0x23, 0x23, 0x23, 0x23, 0x23, 0x23 };
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
mode = ata_limit_mode(dev, mode, ctlr->chip->max_dma);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
Add support for the VIA 8237 (both PATA and SATA part). Cleanup the SATA support a bit now we are here anyways.
2003-10-30 13:16:21 +00:00
if ( mode > ATA_UDMA2 && !(reg54 & (0x10 << devno))) {
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_print_cable(dev, "controller");
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
mode = ATA_UDMA2;
}
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
error = ata_controlcmd(dev, ATA_SETFEATURES, ATA_SF_SETXFER, 0, mode);
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (bootverbose)
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
device_printf(dev, "%ssetting %s on %s chip\n",
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
(error) ? "FAILURE " : "",
ata_mode2str(mode), ctlr->chip->text);
Update the SATA support code to work more correctly with real SATA disks now that I can test it. Add support for the SiI 3112 SATA chip using memory mapped I/O. Update the support for the SiI 0680 to use the memio interface as well. Sponsored by: David Leimbach <leimy2k@mac.com> (3112 based controller) Sponsored by: FreeBSD Systems (www.FreeBSDsystems.com) (SATA disks)
2003-07-02 10:50:44 +00:00
if (error)
return;
Add support for the VIA 8237 (both PATA and SATA part). Cleanup the SATA support a bit now we are here anyways.
2003-10-30 13:16:21 +00:00
if (mode >= ATA_UDMA0) {
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, 0x48, reg48 | (0x0001 << devno), 2);
Fix the long standing problem with poor transferrates on Intel ICHH type chips. The DMA timing value was set on device 0 for all devices :/ Prodded by: Harald Schmalzbauer
2005-06-09 21:13:44 +00:00
pci_write_config(gparent, 0x4a,
(reg4a & ~(0x3 << (devno << 2))) |
((0x01 + !(mode & 0x01)) << (devno << 2)), 2);
Add support for the VIA 8237 (both PATA and SATA part). Cleanup the SATA support a bit now we are here anyways.
2003-10-30 13:16:21 +00:00
}
else {
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, 0x48, reg48 & ~(0x0001 << devno), 2);
pci_write_config(gparent, 0x4a, (reg4a & ~(0x3 << (devno << 2))), 2);
Add support for the VIA 8237 (both PATA and SATA part). Cleanup the SATA support a bit now we are here anyways.
2003-10-30 13:16:21 +00:00
}
Always set half'n'half mode on ICH* chips.
2004-10-01 09:06:22 +00:00
reg54 |= 0x0400;
Add support for the VIA 8237 (both PATA and SATA part). Cleanup the SATA support a bit now we are here anyways.
2003-10-30 13:16:21 +00:00
if (mode >= ATA_UDMA2)
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, 0x54, reg54 | (0x1 << devno), 2);
Add support for the VIA 8237 (both PATA and SATA part). Cleanup the SATA support a bit now we are here anyways.
2003-10-30 13:16:21 +00:00
else
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, 0x54, reg54 & ~(0x1 << devno), 2);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
Add support for the VIA 8237 (both PATA and SATA part). Cleanup the SATA support a bit now we are here anyways.
2003-10-30 13:16:21 +00:00
if (mode >= ATA_UDMA5)
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, 0x54, reg54 | (0x1000 << devno), 2);
Add support for the VIA 8237 (both PATA and SATA part). Cleanup the SATA support a bit now we are here anyways.
2003-10-30 13:16:21 +00:00
else
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, 0x54, reg54 & ~(0x1000 << devno), 2);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
Add support for the VIA 8237 (both PATA and SATA part). Cleanup the SATA support a bit now we are here anyways.
2003-10-30 13:16:21 +00:00
reg40 &= ~0x00ff00ff;
reg40 |= 0x40774077;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
Add support for the VIA 8237 (both PATA and SATA part). Cleanup the SATA support a bit now we are here anyways.
2003-10-30 13:16:21 +00:00
if (atadev->unit == ATA_MASTER) {
mask40 = 0x3300;
new40 = timings[ata_mode2idx(mode)] << 8;
}
else {
mask44 = 0x0f;
new44 = ((timings[ata_mode2idx(mode)] & 0x30) >> 2) |
(timings[ata_mode2idx(mode)] & 0x03);
}
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
if (ch->unit) {
Add support for the VIA 8237 (both PATA and SATA part). Cleanup the SATA support a bit now we are here anyways.
2003-10-30 13:16:21 +00:00
mask40 <<= 16;
new40 <<= 16;
mask44 <<= 4;
new44 <<= 4;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, 0x40, (reg40 & ~mask40) | new40, 4);
pci_write_config(gparent, 0x44, (reg44 & ~mask44) | new44, 1);
Add support for the VIA 8237 (both PATA and SATA part). Cleanup the SATA support a bit now we are here anyways.
2003-10-30 13:16:21 +00:00
Update the SATA support code to work more correctly with real SATA disks now that I can test it. Add support for the SiI 3112 SATA chip using memory mapped I/O. Update the support for the SiI 0680 to use the memio interface as well. Sponsored by: David Leimbach <leimy2k@mac.com> (3112 based controller) Sponsored by: FreeBSD Systems (www.FreeBSDsystems.com) (SATA disks)
2003-07-02 10:50:44 +00:00
atadev->mode = mode;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static int
ata_intel_31244_allocate(device_t dev)
Add support for the ITE IT8212F controller. HW donated by: Yahoo!
2004-12-08 11:17:38 +00:00
{
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
struct ata_pci_controller *ctlr = device_get_softc(device_get_parent(dev));
struct ata_channel *ch = device_get_softc(dev);
int i;
int ch_offset;
Add support for the ITE IT8211F controller.
2005-08-05 13:14:00 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ch_offset = 0x200 + ch->unit * 0x200;
Add support for the ITE IT8212F controller. HW donated by: Yahoo!
2004-12-08 11:17:38 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
for (i = ATA_DATA; i < ATA_MAX_RES; i++)
ch->r_io[i].res = ctlr->r_res2;
Add preliminary support for the Serverworks HT1000 chip. HW sponsored by: Yahoo!
2006-07-04 20:36:03 +00:00
/* setup ATA registers */
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ch->r_io[ATA_DATA].offset = ch_offset + 0x00;
ch->r_io[ATA_FEATURE].offset = ch_offset + 0x06;
ch->r_io[ATA_COUNT].offset = ch_offset + 0x08;
ch->r_io[ATA_SECTOR].offset = ch_offset + 0x0c;
ch->r_io[ATA_CYL_LSB].offset = ch_offset + 0x10;
ch->r_io[ATA_CYL_MSB].offset = ch_offset + 0x14;
ch->r_io[ATA_DRIVE].offset = ch_offset + 0x18;
ch->r_io[ATA_COMMAND].offset = ch_offset + 0x1d;
ch->r_io[ATA_ERROR].offset = ch_offset + 0x04;
ch->r_io[ATA_STATUS].offset = ch_offset + 0x1c;
ch->r_io[ATA_ALTSTAT].offset = ch_offset + 0x28;
ch->r_io[ATA_CONTROL].offset = ch_offset + 0x29;
Add preliminary support for the Serverworks HT1000 chip. HW sponsored by: Yahoo!
2006-07-04 20:36:03 +00:00
/* setup DMA registers */
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ch->r_io[ATA_SSTATUS].offset = ch_offset + 0x100;
ch->r_io[ATA_SERROR].offset = ch_offset + 0x104;
ch->r_io[ATA_SCONTROL].offset = ch_offset + 0x108;
Add preliminary support for the Serverworks HT1000 chip. HW sponsored by: Yahoo!
2006-07-04 20:36:03 +00:00
/* setup SATA registers */
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ch->r_io[ATA_BMCMD_PORT].offset = ch_offset + 0x70;
ch->r_io[ATA_BMSTAT_PORT].offset = ch_offset + 0x72;
ch->r_io[ATA_BMDTP_PORT].offset = ch_offset + 0x74;
ch->flags |= ATA_NO_SLAVE;
ata_pci_hw(dev);
ch->hw.status = ata_intel_31244_status;
ch->hw.command = ata_intel_31244_command;
/* enable PHY state change interrupt */
ATA_OUTL(ctlr->r_res2, 0x4,
ATA_INL(ctlr->r_res2, 0x04) | (0x01 << (ch->unit << 3)));
return 0;
}
static int
ata_intel_31244_status(device_t dev)
{
struct ata_channel *ch = device_get_softc(dev);
u_int32_t status = ATA_IDX_INL(ch, ATA_SSTATUS);
u_int32_t error = ATA_IDX_INL(ch, ATA_SERROR);
struct ata_connect_task *tp;
/* check for PHY related interrupts on SATA capable HW */
if (error) {
/* clear error bits/interrupt */
ATA_IDX_OUTL(ch, ATA_SERROR, error);
/* if we have a connection event deal with it */
if ((error & ATA_SE_PHY_CHANGED) &&
(tp = (struct ata_connect_task *)
malloc(sizeof(struct ata_connect_task),
M_ATA, M_NOWAIT | M_ZERO))) {
if ((status & ATA_SS_CONWELL_MASK) == ATA_SS_CONWELL_GEN1) {
if (bootverbose)
device_printf(ch->dev, "CONNECT requested\n");
tp->action = ATA_C_ATTACH;
}
else {
if (bootverbose)
device_printf(ch->dev, "DISCONNECT requested\n");
tp->action = ATA_C_DETACH;
}
tp->dev = ch->dev;
TASK_INIT(&tp->task, 0, ata_sata_phy_event, tp);
taskqueue_enqueue(taskqueue_thread, &tp->task);
}
}
/* any drive action to take care of ? */
Unbreak DMA dumo on Intel 31224.
2006-02-06 19:17:48 +00:00
return ata_pci_status(dev);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
}
static int
ata_intel_31244_command(struct ata_request *request)
{
struct ata_channel *ch = device_get_softc(device_get_parent(request->dev));
struct ata_device *atadev = device_get_softc(request->dev);
u_int64_t lba;
if (!(atadev->flags & ATA_D_48BIT_ACTIVE))
return (ata_generic_command(request));
lba = request->u.ata.lba;
ATA_IDX_OUTB(ch, ATA_DRIVE, ATA_D_IBM | ATA_D_LBA | atadev->unit);
/* enable interrupt */
ATA_IDX_OUTB(ch, ATA_CONTROL, ATA_A_4BIT);
ATA_IDX_OUTW(ch, ATA_FEATURE, request->u.ata.feature);
ATA_IDX_OUTW(ch, ATA_COUNT, request->u.ata.count);
ATA_IDX_OUTW(ch, ATA_SECTOR, ((lba >> 16) & 0xff00) | (lba & 0x00ff));
ATA_IDX_OUTW(ch, ATA_CYL_LSB, ((lba >> 24) & 0xff00) |
((lba >> 8) & 0x00ff));
ATA_IDX_OUTW(ch, ATA_CYL_MSB, ((lba >> 32) & 0xff00) |
((lba >> 16) & 0x00ff));
/* issue command to controller */
ATA_IDX_OUTB(ch, ATA_COMMAND, request->u.ata.command);
return 0;
}
static void
ata_intel_31244_reset(device_t dev)
{
struct ata_channel *ch = device_get_softc(dev);
ata_sata_phy_enable(ch);
}
/*
* Integrated Technology Express Inc. (ITE) chipset support functions
*/
int
ata_ite_ident(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
struct ata_chip_id *idx;
static struct ata_chip_id ids[] =
{{ ATA_IT8212F, 0x00, 0x00, 0x00, ATA_UDMA6, "IT8212F" },
{ ATA_IT8211F, 0x00, 0x00, 0x00, ATA_UDMA6, "IT8211F" },
{ 0, 0, 0, 0, 0, 0}};
char buffer[64];
if (!(idx = ata_match_chip(dev, ids)))
return ENXIO;
sprintf(buffer, "ITE %s %s controller",
idx->text, ata_mode2str(idx->max_dma));
device_set_desc_copy(dev, buffer);
ctlr->chip = idx;
ctlr->chipinit = ata_ite_chipinit;
return 0;
}
static int
ata_ite_chipinit(device_t dev)
Add support for the ITE IT8212F controller. HW donated by: Yahoo!
2004-12-08 11:17:38 +00:00
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
if (ata_setup_interrupt(dev))
return ENXIO;
ctlr->setmode = ata_ite_setmode;
/* set PCI mode and 66Mhz reference clock */
pci_write_config(dev, 0x50, pci_read_config(dev, 0x50, 1) & ~0x83, 1);
/* set default active & recover timings */
pci_write_config(dev, 0x54, 0x31, 1);
pci_write_config(dev, 0x56, 0x31, 1);
return 0;
}
static void
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_ite_setmode(device_t dev, int mode)
Add support for the ITE IT8212F controller. HW donated by: Yahoo!
2004-12-08 11:17:38 +00:00
{
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
device_t gparent = GRANDPARENT(dev);
struct ata_channel *ch = device_get_softc(device_get_parent(dev));
struct ata_device *atadev = device_get_softc(dev);
Add support for the ITE IT8212F controller. HW donated by: Yahoo!
2004-12-08 11:17:38 +00:00
int devno = (ch->unit << 1) + ATA_DEV(atadev->unit);
int error;
/* correct the mode for what the HW supports */
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
mode = ata_limit_mode(dev, mode, ATA_UDMA6);
Add support for the ITE IT8212F controller. HW donated by: Yahoo!
2004-12-08 11:17:38 +00:00
/* check the CBLID bits for 80 conductor cable detection */
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
if (mode > ATA_UDMA2 && (pci_read_config(gparent, 0x40, 2) &
Add support for the ITE IT8212F controller. HW donated by: Yahoo!
2004-12-08 11:17:38 +00:00
(ch->unit ? (1<<3) : (1<<2)))) {
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_print_cable(dev, "controller");
Add support for the ITE IT8212F controller. HW donated by: Yahoo!
2004-12-08 11:17:38 +00:00
mode = ATA_UDMA2;
}
/* set the wanted mode on the device */
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
error = ata_controlcmd(dev, ATA_SETFEATURES, ATA_SF_SETXFER, 0, mode);
Add support for the ITE IT8212F controller. HW donated by: Yahoo!
2004-12-08 11:17:38 +00:00
if (bootverbose)
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
device_printf(dev, "%s setting %s on ITE8212F chip\n",
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
(error) ? "failed" : "success", ata_mode2str(mode));
Add support for the ITE IT8212F controller. HW donated by: Yahoo!
2004-12-08 11:17:38 +00:00
/* if the device accepted the mode change, setup the HW accordingly */
if (!error) {
if (mode >= ATA_UDMA0) {
u_int8_t udmatiming[] =
{ 0x44, 0x42, 0x31, 0x21, 0x11, 0xa2, 0x91 };
/* enable UDMA mode */
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, 0x50,
pci_read_config(gparent, 0x50, 1) &
Add support for the ITE IT8212F controller. HW donated by: Yahoo!
2004-12-08 11:17:38 +00:00
~(1 << (devno + 3)), 1);
/* set UDMA timing */
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent,
Add support for the ITE IT8212F controller. HW donated by: Yahoo!
2004-12-08 11:17:38 +00:00
0x56 + (ch->unit << 2) + ATA_DEV(atadev->unit),
udmatiming[mode & ATA_MODE_MASK], 1);
}
else {
u_int8_t chtiming[] =
{ 0xaa, 0xa3, 0xa1, 0x33, 0x31, 0x88, 0x32, 0x31 };
/* disable UDMA mode */
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, 0x50,
pci_read_config(gparent, 0x50, 1) |
Add support for the ITE IT8212F controller. HW donated by: Yahoo!
2004-12-08 11:17:38 +00:00
(1 << (devno + 3)), 1);
/* set active and recover timing (shared between master & slave) */
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
if (pci_read_config(gparent, 0x54 + (ch->unit << 2), 1) <
Add support for the ITE IT8212F controller. HW donated by: Yahoo!
2004-12-08 11:17:38 +00:00
chtiming[ata_mode2idx(mode)])
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, 0x54 + (ch->unit << 2),
chtiming[ata_mode2idx(mode)], 1);
Add support for the ITE IT8212F controller. HW donated by: Yahoo!
2004-12-08 11:17:38 +00:00
}
atadev->mode = mode;
}
}
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
Add support for the JMicron JMB360 SATAII controller. Thanks to JMicron for providing needed info. HW donated by: Ralf Folkerts
2006-01-25 23:07:42 +00:00
/*
* JMicron chipset support functions
*/
int
ata_jmicron_ident(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
struct ata_chip_id *idx;
static struct ata_chip_id ids[] =
Add support for the JMicron JMB361, 365 and 366 chips. HW and documentation kindly provided by JMicron.
2006-03-13 14:01:37 +00:00
{{ ATA_JMB360, 0, 1, 0, ATA_SA300, "JMB360" },
{ ATA_JMB361, 0, 1, 1, ATA_SA300, "JMB361" },
{ ATA_JMB363, 0, 2, 1, ATA_SA300, "JMB363" },
{ ATA_JMB365, 0, 1, 2, ATA_SA300, "JMB365" },
{ ATA_JMB366, 0, 2, 2, ATA_SA300, "JMB366" },
Add support for the JMicron JMB360 SATAII controller. Thanks to JMicron for providing needed info. HW donated by: Ralf Folkerts
2006-01-25 23:07:42 +00:00
{ 0, 0, 0, 0, 0, 0}};
char buffer[64];
if (!(idx = ata_match_chip(dev, ids)))
return ENXIO;
Update JMicron support to also work with chips where the PATA and SATA parts are on individual PCI functions. HW donated by: JMicron
2006-08-18 09:56:12 +00:00
if ((pci_read_config(dev, 0xdf, 1) & 0x40) &&
(pci_get_function(dev) == (pci_read_config(dev, 0x40, 1) & 0x02 >> 1)))
sprintf(buffer, "JMicron %s %s controller",
idx->text, ata_mode2str(ATA_UDMA6));
else
sprintf(buffer, "JMicron %s %s controller",
idx->text, ata_mode2str(idx->max_dma));
Add support for the JMicron JMB360 SATAII controller. Thanks to JMicron for providing needed info. HW donated by: Ralf Folkerts
2006-01-25 23:07:42 +00:00
device_set_desc_copy(dev, buffer);
ctlr->chip = idx;
ctlr->chipinit = ata_jmicron_chipinit;
return 0;
}
static int
ata_jmicron_chipinit(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
Update AHCI support to be more generic. Add support for AHCI on the VIA VT8251.
2006-06-28 09:59:09 +00:00
int error;
Add support for the JMicron JMB360 SATAII controller. Thanks to JMicron for providing needed info. HW donated by: Ralf Folkerts
2006-01-25 23:07:42 +00:00
if (ata_setup_interrupt(dev))
return ENXIO;
Update JMicron support to also work with chips where the PATA and SATA parts are on individual PCI functions. HW donated by: JMicron
2006-08-18 09:56:12 +00:00
/* do we have multiple PCI functions ? */
if (pci_read_config(dev, 0xdf, 1) & 0x40) {
/* if we have a memory BAR(5) we are on the AHCI part */
ctlr->r_type2 = SYS_RES_MEMORY;
ctlr->r_rid2 = PCIR_BAR(5);
if ((ctlr->r_res2 = bus_alloc_resource_any(dev, ctlr->r_type2,
&ctlr->r_rid2, RF_ACTIVE)))
return ata_ahci_chipinit(dev);
/* otherwise we are on the PATA part */
ctlr->allocate = ata_pci_allocate;
ctlr->reset = ata_generic_reset;
ctlr->dmainit = ata_pci_dmainit;
ctlr->setmode = ata_jmicron_setmode;
ctlr->channels = ctlr->chip->cfg2;
Add support for the JMicron JMB360 SATAII controller. Thanks to JMicron for providing needed info. HW donated by: Ralf Folkerts
2006-01-25 23:07:42 +00:00
}
Update JMicron support to also work with chips where the PATA and SATA parts are on individual PCI functions. HW donated by: JMicron
2006-08-18 09:56:12 +00:00
else {
/* set controller configuration to a combined setup we support */
pci_write_config(dev, 0x40, 0x80c0a131, 4);
pci_write_config(dev, 0x80, 0x01200000, 4);
ctlr->r_type2 = SYS_RES_MEMORY;
ctlr->r_rid2 = PCIR_BAR(5);
if ((ctlr->r_res2 = bus_alloc_resource_any(dev, ctlr->r_type2,
&ctlr->r_rid2, RF_ACTIVE))){
if ((error = ata_ahci_chipinit(dev)))
return error;
}
Add support for the JMicron JMB360 SATAII controller. Thanks to JMicron for providing needed info. HW donated by: Ralf Folkerts
2006-01-25 23:07:42 +00:00
Update JMicron support to also work with chips where the PATA and SATA parts are on individual PCI functions. HW donated by: JMicron
2006-08-18 09:56:12 +00:00
ctlr->allocate = ata_jmicron_allocate;
ctlr->reset = ata_jmicron_reset;
ctlr->dmainit = ata_jmicron_dmainit;
ctlr->setmode = ata_jmicron_setmode;
/* set the number of HW channels */
ctlr->channels = ctlr->chip->cfg1 + ctlr->chip->cfg2;
}
Add support for the JMicron JMB363 dual SATA + single PATA controller. Documentation and HW kindly provided by JMicron.
2006-02-16 17:09:24 +00:00
return 0;
}
Add support for the JMicron JMB360 SATAII controller. Thanks to JMicron for providing needed info. HW donated by: Ralf Folkerts
2006-01-25 23:07:42 +00:00
Add support for the JMicron JMB363 dual SATA + single PATA controller. Documentation and HW kindly provided by JMicron.
2006-02-16 17:09:24 +00:00
static int
ata_jmicron_allocate(device_t dev)
{
Add support for the JMicron JMB361, 365 and 366 chips. HW and documentation kindly provided by JMicron.
2006-03-13 14:01:37 +00:00
struct ata_pci_controller *ctlr = device_get_softc(device_get_parent(dev));
Add support for the JMicron JMB363 dual SATA + single PATA controller. Documentation and HW kindly provided by JMicron.
2006-02-16 17:09:24 +00:00
struct ata_channel *ch = device_get_softc(dev);
int error;
Add support for the JMicron JMB360 SATAII controller. Thanks to JMicron for providing needed info. HW donated by: Ralf Folkerts
2006-01-25 23:07:42 +00:00
Add support for the JMicron JMB361, 365 and 366 chips. HW and documentation kindly provided by JMicron.
2006-03-13 14:01:37 +00:00
if (ch->unit >= ctlr->chip->cfg1) {
ch->unit -= ctlr->chip->cfg1;
Add support for the JMicron JMB363 dual SATA + single PATA controller. Documentation and HW kindly provided by JMicron.
2006-02-16 17:09:24 +00:00
error = ata_pci_allocate(dev);
Add support for the JMicron JMB361, 365 and 366 chips. HW and documentation kindly provided by JMicron.
2006-03-13 14:01:37 +00:00
ch->unit += ctlr->chip->cfg1;
Add support for the JMicron JMB363 dual SATA + single PATA controller. Documentation and HW kindly provided by JMicron.
2006-02-16 17:09:24 +00:00
}
else
error = ata_ahci_allocate(dev);
return error;
}
Add support for the JMicron JMB360 SATAII controller. Thanks to JMicron for providing needed info. HW donated by: Ralf Folkerts
2006-01-25 23:07:42 +00:00
Add support for the JMicron JMB363 dual SATA + single PATA controller. Documentation and HW kindly provided by JMicron.
2006-02-16 17:09:24 +00:00
static void
ata_jmicron_reset(device_t dev)
{
Add support for the JMicron JMB361, 365 and 366 chips. HW and documentation kindly provided by JMicron.
2006-03-13 14:01:37 +00:00
struct ata_pci_controller *ctlr = device_get_softc(device_get_parent(dev));
Add support for the JMicron JMB363 dual SATA + single PATA controller. Documentation and HW kindly provided by JMicron.
2006-02-16 17:09:24 +00:00
struct ata_channel *ch = device_get_softc(dev);
Add support for the JMicron JMB360 SATAII controller. Thanks to JMicron for providing needed info. HW donated by: Ralf Folkerts
2006-01-25 23:07:42 +00:00
Add support for the JMicron JMB361, 365 and 366 chips. HW and documentation kindly provided by JMicron.
2006-03-13 14:01:37 +00:00
if (ch->unit >= ctlr->chip->cfg1)
Add support for the JMicron JMB363 dual SATA + single PATA controller. Documentation and HW kindly provided by JMicron.
2006-02-16 17:09:24 +00:00
ata_generic_reset(dev);
else
ata_ahci_reset(dev);
}
Add support for the JMicron JMB360 SATAII controller. Thanks to JMicron for providing needed info. HW donated by: Ralf Folkerts
2006-01-25 23:07:42 +00:00
Add support for the JMicron JMB363 dual SATA + single PATA controller. Documentation and HW kindly provided by JMicron.
2006-02-16 17:09:24 +00:00
static void
ata_jmicron_dmainit(device_t dev)
{
Add support for the JMicron JMB361, 365 and 366 chips. HW and documentation kindly provided by JMicron.
2006-03-13 14:01:37 +00:00
struct ata_pci_controller *ctlr = device_get_softc(device_get_parent(dev));
Add support for the JMicron JMB363 dual SATA + single PATA controller. Documentation and HW kindly provided by JMicron.
2006-02-16 17:09:24 +00:00
struct ata_channel *ch = device_get_softc(dev);
Add support for the JMicron JMB361, 365 and 366 chips. HW and documentation kindly provided by JMicron.
2006-03-13 14:01:37 +00:00
if (ch->unit >= ctlr->chip->cfg1)
Add support for the JMicron JMB363 dual SATA + single PATA controller. Documentation and HW kindly provided by JMicron.
2006-02-16 17:09:24 +00:00
ata_pci_dmainit(dev);
else
ata_ahci_dmainit(dev);
}
static void
ata_jmicron_setmode(device_t dev, int mode)
{
Add support for the JMicron JMB361, 365 and 366 chips. HW and documentation kindly provided by JMicron.
2006-03-13 14:01:37 +00:00
struct ata_pci_controller *ctlr = device_get_softc(GRANDPARENT(dev));
Add support for the JMicron JMB363 dual SATA + single PATA controller. Documentation and HW kindly provided by JMicron.
2006-02-16 17:09:24 +00:00
struct ata_channel *ch = device_get_softc(device_get_parent(dev));
Update JMicron support to also work with chips where the PATA and SATA parts are on individual PCI functions. HW donated by: JMicron
2006-08-18 09:56:12 +00:00
if (pci_read_config(dev, 0xdf, 1) & 0x40 || ch->unit >= ctlr->chip->cfg1) {
Add support for the JMicron JMB363 dual SATA + single PATA controller. Documentation and HW kindly provided by JMicron.
2006-02-16 17:09:24 +00:00
struct ata_device *atadev = device_get_softc(dev);
/* check for 80pin cable present */
if (pci_read_config(dev, 0x40, 1) & 0x08)
mode = ata_limit_mode(dev, mode, ATA_UDMA2);
else
mode = ata_limit_mode(dev, mode, ATA_UDMA6);
if (!ata_controlcmd(dev, ATA_SETFEATURES, ATA_SF_SETXFER, 0, mode))
atadev->mode = mode;
}
else
ata_sata_setmode(dev, mode);
Add support for the JMicron JMB360 SATAII controller. Thanks to JMicron for providing needed info. HW donated by: Ralf Folkerts
2006-01-25 23:07:42 +00:00
}
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
/*
* Marvell chipset support functions
*/
#define ATA_MV_HOST_BASE(ch) \
((ch->unit & 3) * 0x0100) + (ch->unit > 3 ? 0x30000 : 0x20000)
#define ATA_MV_EDMA_BASE(ch) \
((ch->unit & 3) * 0x2000) + (ch->unit > 3 ? 0x30000 : 0x20000)
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
struct ata_marvell_response {
Whitespace cleanup.
2006-01-18 13:10:17 +00:00
u_int16_t tag;
u_int8_t edma_status;
u_int8_t dev_status;
u_int32_t timestamp;
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
};
struct ata_marvell_dma_prdentry {
u_int32_t addrlo;
u_int32_t count;
u_int32_t addrhi;
u_int32_t reserved;
};
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
int
ata_marvell_ident(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
struct ata_chip_id *idx;
static struct ata_chip_id ids[] =
{{ ATA_M88SX5040, 0, 4, MV5XXX, ATA_SA150, "88SX5040" },
{ ATA_M88SX5041, 0, 4, MV5XXX, ATA_SA150, "88SX5041" },
{ ATA_M88SX5080, 0, 8, MV5XXX, ATA_SA150, "88SX5080" },
{ ATA_M88SX5081, 0, 8, MV5XXX, ATA_SA150, "88SX5081" },
{ ATA_M88SX6041, 0, 4, MV6XXX, ATA_SA300, "88SX6041" },
{ ATA_M88SX6081, 0, 8, MV6XXX, ATA_SA300, "88SX6081" },
{ 0, 0, 0, 0, 0, 0}};
char buffer[64];
if (!(idx = ata_match_chip(dev, ids)))
return ENXIO;
sprintf(buffer, "Marvell %s %s controller",
idx->text, ata_mode2str(idx->max_dma));
device_set_desc_copy(dev, buffer);
ctlr->chip = idx;
ctlr->chipinit = ata_marvell_chipinit;
return 0;
}
static int
ata_marvell_chipinit(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
if (ata_setup_interrupt(dev))
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
return ENXIO;
ctlr->r_type1 = SYS_RES_MEMORY;
ctlr->r_rid1 = PCIR_BAR(0);
if (!(ctlr->r_res1 = bus_alloc_resource_any(dev, ctlr->r_type1,
&ctlr->r_rid1, RF_ACTIVE)))
return ENXIO;
/* mask all host controller interrupts */
ATA_OUTL(ctlr->r_res1, 0x01d64, 0x00000000);
/* mask all PCI interrupts */
ATA_OUTL(ctlr->r_res1, 0x01d5c, 0x00000000);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ctlr->allocate = ata_marvell_allocate;
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
ctlr->reset = ata_marvell_reset;
ctlr->dmainit = ata_marvell_dmainit;
ctlr->setmode = ata_sata_setmode;
ctlr->channels = ctlr->chip->cfg1;
/* clear host controller interrupts */
ATA_OUTL(ctlr->r_res1, 0x20014, 0x00000000);
if (ctlr->chip->cfg1 > 4)
ATA_OUTL(ctlr->r_res1, 0x30014, 0x00000000);
/* clear PCI interrupts */
ATA_OUTL(ctlr->r_res1, 0x01d58, 0x00000000);
/* unmask PCI interrupts we want */
ATA_OUTL(ctlr->r_res1, 0x01d5c, 0x007fffff);
/* unmask host controller interrupts we want */
ATA_OUTL(ctlr->r_res1, 0x01d64, 0x000000ff/*HC0*/ | 0x0001fe00/*HC1*/ |
/*(1<<19) | (1<<20) | (1<<21) |*/(1<<22) | (1<<24) | (0x7f << 25));
Add support for the JMicron JMB360 SATAII controller. Thanks to JMicron for providing needed info. HW donated by: Ralf Folkerts
2006-01-25 23:07:42 +00:00
/* enable PCI interrupt */
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
pci_write_config(dev, PCIR_COMMAND,
pci_read_config(dev, PCIR_COMMAND, 2) & ~0x0400, 2);
return 0;
}
static int
ata_marvell_allocate(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(device_get_parent(dev));
struct ata_channel *ch = device_get_softc(dev);
Fix type casts so that we get at the right response queue. Submitted by: glebius+ru
2006-06-19 11:47:57 +00:00
bus_addr_t work = ch->dma->work_bus;
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
int i;
Fix type casts so that we get at the right response queue. Submitted by: glebius+ru
2006-06-19 11:47:57 +00:00
/* clear work area */
bzero(ch->dma->work, 1024+256);
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
/* set legacy ATA resources */
for (i = ATA_DATA; i <= ATA_COMMAND; i++) {
ch->r_io[i].res = ctlr->r_res1;
ch->r_io[i].offset = 0x02100 + (i << 2) + ATA_MV_EDMA_BASE(ch);
}
ch->r_io[ATA_CONTROL].res = ctlr->r_res1;
ch->r_io[ATA_CONTROL].offset = 0x02120 + ATA_MV_EDMA_BASE(ch);
ch->r_io[ATA_IDX_ADDR].res = ctlr->r_res1;
ata_default_registers(dev);
/* set SATA resources */
switch (ctlr->chip->cfg2) {
case MV5XXX:
ch->r_io[ATA_SSTATUS].res = ctlr->r_res1;
ch->r_io[ATA_SSTATUS].offset = 0x00100 + ATA_MV_HOST_BASE(ch);
ch->r_io[ATA_SERROR].res = ctlr->r_res1;
ch->r_io[ATA_SERROR].offset = 0x00104 + ATA_MV_HOST_BASE(ch);
ch->r_io[ATA_SCONTROL].res = ctlr->r_res1;
ch->r_io[ATA_SCONTROL].offset = 0x00108 + ATA_MV_HOST_BASE(ch);
break;
case MV6XXX:
ch->r_io[ATA_SSTATUS].res = ctlr->r_res1;
ch->r_io[ATA_SSTATUS].offset = 0x02300 + ATA_MV_EDMA_BASE(ch);
ch->r_io[ATA_SERROR].res = ctlr->r_res1;
ch->r_io[ATA_SERROR].offset = 0x02304 + ATA_MV_EDMA_BASE(ch);
ch->r_io[ATA_SCONTROL].res = ctlr->r_res1;
ch->r_io[ATA_SCONTROL].offset = 0x02308 + ATA_MV_EDMA_BASE(ch);
ch->r_io[ATA_SACTIVE].res = ctlr->r_res1;
ch->r_io[ATA_SACTIVE].offset = 0x02350 + ATA_MV_EDMA_BASE(ch);
break;
}
ch->flags |= ATA_NO_SLAVE;
ch->flags |= ATA_USE_16BIT; /* XXX SOS needed ? */
ata_generic_hw(dev);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ch->hw.begin_transaction = ata_marvell_begin_transaction;
ch->hw.end_transaction = ata_marvell_end_transaction;
ch->hw.status = ata_marvell_status;
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
/* disable the EDMA machinery */
ATA_OUTL(ctlr->r_res1, 0x02028 + ATA_MV_EDMA_BASE(ch), 0x00000002);
DELAY(100000); /* SOS should poll for disabled */
/* set configuration to non-queued 128b read transfers stop on error */
ATA_OUTL(ctlr->r_res1, 0x02000 + ATA_MV_EDMA_BASE(ch), (1<<11) | (1<<13));
/* request queue base high */
Fix type casts so that we get at the right response queue. Submitted by: glebius+ru
2006-06-19 11:47:57 +00:00
ATA_OUTL(ctlr->r_res1, 0x02010 + ATA_MV_EDMA_BASE(ch), (work >> 16) >> 16);
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
/* request queue in ptr */
Fix type casts so that we get at the right response queue. Submitted by: glebius+ru
2006-06-19 11:47:57 +00:00
ATA_OUTL(ctlr->r_res1, 0x02014 + ATA_MV_EDMA_BASE(ch), work & 0xffffffff);
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
/* request queue out ptr */
ATA_OUTL(ctlr->r_res1, 0x02018 + ATA_MV_EDMA_BASE(ch), 0x0);
/* response queue base high */
Fix type casts so that we get at the right response queue. Submitted by: glebius+ru
2006-06-19 11:47:57 +00:00
work += 1024;
ATA_OUTL(ctlr->r_res1, 0x0201c + ATA_MV_EDMA_BASE(ch), (work >> 16) >> 16);
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
/* response queue in ptr */
ATA_OUTL(ctlr->r_res1, 0x02020 + ATA_MV_EDMA_BASE(ch), 0x0);
/* response queue out ptr */
Fix type casts so that we get at the right response queue. Submitted by: glebius+ru
2006-06-19 11:47:57 +00:00
ATA_OUTL(ctlr->r_res1, 0x02024 + ATA_MV_EDMA_BASE(ch), work & 0xffffffff);
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
/* clear SATA error register */
ATA_IDX_OUTL(ch, ATA_SERROR, ATA_IDX_INL(ch, ATA_SERROR));
/* clear any outstanding error interrupts */
ATA_OUTL(ctlr->r_res1, 0x02008 + ATA_MV_EDMA_BASE(ch), 0x0);
/* unmask all error interrupts */
ATA_OUTL(ctlr->r_res1, 0x0200c + ATA_MV_EDMA_BASE(ch), ~0x0);
/* enable EDMA machinery */
ATA_OUTL(ctlr->r_res1, 0x02028 + ATA_MV_EDMA_BASE(ch), 0x00000001);
return 0;
}
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static int
ata_marvell_status(device_t dev)
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
{
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
struct ata_pci_controller *ctlr = device_get_softc(device_get_parent(dev));
struct ata_channel *ch = device_get_softc(dev);
u_int32_t cause = ATA_INL(ctlr->r_res1, 0x01d60);
int shift = (ch->unit << 1) + (ch->unit > 3);
/* do we have any errors flagged ? */
if (cause & (1 << shift)) {
struct ata_connect_task *tp;
u_int32_t error =
ATA_INL(ctlr->r_res1, 0x02008 + ATA_MV_EDMA_BASE(ch));
/* check for and handle disconnect events */
if ((error & 0x00000008) &&
(tp = (struct ata_connect_task *)
malloc(sizeof(struct ata_connect_task),
M_ATA, M_NOWAIT | M_ZERO))) {
if (bootverbose)
device_printf(ch->dev, "DISCONNECT requested\n");
tp->action = ATA_C_DETACH;
tp->dev = ch->dev;
TASK_INIT(&tp->task, 0, ata_sata_phy_event, tp);
taskqueue_enqueue(taskqueue_thread, &tp->task);
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
}
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
/* check for and handle connect events */
if ((error & 0x00000010) &&
(tp = (struct ata_connect_task *)
malloc(sizeof(struct ata_connect_task),
M_ATA, M_NOWAIT | M_ZERO))) {
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
if (bootverbose)
device_printf(ch->dev, "CONNECT requested\n");
tp->action = ATA_C_ATTACH;
tp->dev = ch->dev;
TASK_INIT(&tp->task, 0, ata_sata_phy_event, tp);
taskqueue_enqueue(taskqueue_thread, &tp->task);
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
}
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
/* clear SATA error register */
ATA_IDX_OUTL(ch, ATA_SERROR, ATA_IDX_INL(ch, ATA_SERROR));
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
/* clear any outstanding error interrupts */
ATA_OUTL(ctlr->r_res1, 0x02008 + ATA_MV_EDMA_BASE(ch), 0x0);
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
}
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
/* do we have any device action ? */
return (cause & (2 << shift));
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
}
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
/* must be called with ATA channel locked and state_mtx held */
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
static int
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ata_marvell_begin_transaction(struct ata_request *request)
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
{
struct ata_pci_controller *ctlr=device_get_softc(GRANDPARENT(request->dev));
struct ata_channel *ch = device_get_softc(device_get_parent(request->dev));
Further updates to the Marvell support code. Should be reliable enough for actual use by now.
2005-12-28 11:55:43 +00:00
u_int32_t req_in;
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
u_int8_t *bytep;
u_int16_t *wordp;
u_int32_t *quadp;
int i, tag = 0x07;
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
int dummy, error, slot;
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
/* only DMA R/W goes through the EMDA machine */
if (request->u.ata.command != ATA_READ_DMA &&
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
request->u.ata.command != ATA_WRITE_DMA) {
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
Whitespace cleanup.
2006-01-18 13:10:17 +00:00
/* disable the EDMA machinery */
if (ATA_INL(ctlr->r_res1, 0x02028 + ATA_MV_EDMA_BASE(ch)) & 0x00000001)
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
ATA_OUTL(ctlr->r_res1, 0x02028 + ATA_MV_EDMA_BASE(ch), 0x00000002);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
return ata_begin_transaction(request);
}
/* check for 48 bit access and convert if needed */
ata_modify_if_48bit(request);
/* check sanity, setup SG list and DMA engine */
if ((error = ch->dma->load(ch->dev, request->data, request->bytecount,
Whitespace cleanup.
2006-01-18 13:10:17 +00:00
request->flags & ATA_R_READ, ch->dma->sg,
&dummy))) {
device_printf(request->dev, "setting up DMA failed\n");
request->result = error;
return ATA_OP_FINISHED;
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
}
/* get next free request queue slot */
req_in = ATA_INL(ctlr->r_res1, 0x02014 + ATA_MV_EDMA_BASE(ch));
slot = (((req_in & ~0xfffffc00) >> 5) + 0) & 0x1f;
bytep = (u_int8_t *)(ch->dma->work);
bytep += (slot << 5);
wordp = (u_int16_t *)bytep;
quadp = (u_int32_t *)bytep;
Further updates to the Marvell support code. Should be reliable enough for actual use by now.
2005-12-28 11:55:43 +00:00
/* fill in this request */
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
quadp[0] = (long)ch->dma->sg_bus & 0xffffffff;
Appened ull to a really large integer constant to fix build on i386
2006-08-18 00:01:29 +00:00
quadp[1] = (ch->dma->sg_bus & 0xffffffff00000000ull) >> 32;
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
wordp[4] = (request->flags & ATA_R_READ ? 0x01 : 0x00) | (tag<<1);
i = 10;
bytep[i++] = (request->u.ata.count >> 8) & 0xff;
bytep[i++] = 0x10 | ATA_COUNT;
bytep[i++] = request->u.ata.count & 0xff;
bytep[i++] = 0x10 | ATA_COUNT;
bytep[i++] = (request->u.ata.lba >> 24) & 0xff;
bytep[i++] = 0x10 | ATA_SECTOR;
bytep[i++] = request->u.ata.lba & 0xff;
bytep[i++] = 0x10 | ATA_SECTOR;
bytep[i++] = (request->u.ata.lba >> 32) & 0xff;
bytep[i++] = 0x10 | ATA_CYL_LSB;
bytep[i++] = (request->u.ata.lba >> 8) & 0xff;
bytep[i++] = 0x10 | ATA_CYL_LSB;
bytep[i++] = (request->u.ata.lba >> 40) & 0xff;
bytep[i++] = 0x10 | ATA_CYL_MSB;
bytep[i++] = (request->u.ata.lba >> 16) & 0xff;
bytep[i++] = 0x10 | ATA_CYL_MSB;
bytep[i++] = ATA_D_LBA | ATA_D_IBM | ((request->u.ata.lba >> 24) & 0xf);
bytep[i++] = 0x10 | ATA_DRIVE;
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
bytep[i++] = request->u.ata.command;
bytep[i++] = 0x90 | ATA_COMMAND;
/* enable EDMA machinery if needed */
if (!(ATA_INL(ctlr->r_res1, 0x02028 + ATA_MV_EDMA_BASE(ch)) & 0x00000001)) {
ATA_OUTL(ctlr->r_res1, 0x02028 + ATA_MV_EDMA_BASE(ch), 0x00000001);
while (!(ATA_INL(ctlr->r_res1,
0x02028 + ATA_MV_EDMA_BASE(ch)) & 0x00000001))
DELAY(10);
}
Further updates to the Marvell support code. Should be reliable enough for actual use by now.
2005-12-28 11:55:43 +00:00
/* tell EDMA it has a new request */
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
slot = (((req_in & ~0xfffffc00) >> 5) + 1) & 0x1f;
req_in &= 0xfffffc00;
req_in += (slot << 5);
ATA_OUTL(ctlr->r_res1, 0x02014 + ATA_MV_EDMA_BASE(ch), req_in);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
return ATA_OP_CONTINUES;
}
/* must be called with ATA channel locked and state_mtx held */
static int
ata_marvell_end_transaction(struct ata_request *request)
{
struct ata_pci_controller *ctlr=device_get_softc(GRANDPARENT(request->dev));
struct ata_channel *ch = device_get_softc(device_get_parent(request->dev));
int offset = (ch->unit > 3 ? 0x30014 : 0x20014);
u_int32_t icr = ATA_INL(ctlr->r_res1, offset);
int res;
/* EDMA interrupt */
if ((icr & (0x0001 << (ch->unit & 3)))) {
struct ata_marvell_response *response;
u_int32_t rsp_in, rsp_out;
int slot;
/* unload SG list */
ch->dma->unload(ch->dev);
/* stop timeout */
callout_stop(&request->callout);
/* get response ptr's */
rsp_in = ATA_INL(ctlr->r_res1, 0x02020 + ATA_MV_EDMA_BASE(ch));
Whitespace cleanup.
2006-01-18 13:10:17 +00:00
rsp_out = ATA_INL(ctlr->r_res1, 0x02024 + ATA_MV_EDMA_BASE(ch));
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
slot = (((rsp_in & ~0xffffff00) >> 3)) & 0x1f;
rsp_out &= 0xffffff00;
rsp_out += (slot << 3);
response = (struct ata_marvell_response *)
Fix type casts so that we get at the right response queue. Submitted by: glebius+ru
2006-06-19 11:47:57 +00:00
(ch->dma->work + 1024 + (slot << 3));
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
/* record status for this request */
request->status = response->dev_status;
request->error = 0;
/* ack response */
Whitespace cleanup.
2006-01-18 13:10:17 +00:00
ATA_OUTL(ctlr->r_res1, 0x02024 + ATA_MV_EDMA_BASE(ch), rsp_out);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
/* update progress */
if (!(request->status & ATA_S_ERROR) &&
!(request->flags & ATA_R_TIMEOUT))
request->donecount = request->bytecount;
res = ATA_OP_FINISHED;
}
/* legacy ATA interrupt */
else {
res = ata_end_transaction(request);
}
/* ack interrupt */
ATA_OUTL(ctlr->r_res1, offset, ~(icr & (0x0101 << (ch->unit & 3))));
return res;
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
}
static void
ata_marvell_reset(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(device_get_parent(dev));
struct ata_channel *ch = device_get_softc(dev);
/* disable the EDMA machinery */
ATA_OUTL(ctlr->r_res1, 0x02028 + ATA_MV_EDMA_BASE(ch), 0x00000002);
Further updates to the Marvell support code. Should be reliable enough for actual use by now.
2005-12-28 11:55:43 +00:00
while ((ATA_INL(ctlr->r_res1, 0x02028 + ATA_MV_EDMA_BASE(ch)) & 0x00000001))
DELAY(10);
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
/* clear SATA error register */
ATA_IDX_OUTL(ch, ATA_SERROR, ATA_IDX_INL(ch, ATA_SERROR));
/* clear any outstanding error interrupts */
ATA_OUTL(ctlr->r_res1, 0x02008 + ATA_MV_EDMA_BASE(ch), 0x0);
/* unmask all error interrupts */
ATA_OUTL(ctlr->r_res1, 0x0200c + ATA_MV_EDMA_BASE(ch), ~0x0);
/* enable channel and test for devices */
ata_sata_phy_enable(ch);
/* enable EDMA machinery */
ATA_OUTL(ctlr->r_res1, 0x02028 + ATA_MV_EDMA_BASE(ch), 0x00000001);
}
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static void
ata_marvell_dmasetprd(void *xsc, bus_dma_segment_t *segs, int nsegs, int error)
{
struct ata_dmasetprd_args *args = xsc;
struct ata_marvell_dma_prdentry *prd = args->dmatab;
int i;
if ((args->error = error))
return;
for (i = 0; i < nsegs; i++) {
prd[i].addrlo = htole32(segs[i].ds_addr);
prd[i].addrhi = 0;
prd[i].count = htole32(segs[i].ds_len);
}
prd[i - 1].count |= htole32(ATA_DMA_EOT);
}
static void
ata_marvell_dmainit(device_t dev)
{
struct ata_channel *ch = device_get_softc(dev);
ata_dmainit(dev);
if (ch->dma) {
/* note start and stop are not used here */
Whitespace cleanup.
2006-01-18 13:10:17 +00:00
ch->dma->setprd = ata_marvell_dmasetprd;
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
}
}
Add initial support for the Marvell 88SX[56]0[48][01] series of SATA chips. Hardware donated by: Matthew Jacob Hardware donated by: Y!
2005-12-27 17:09:52 +00:00
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
/*
* National chipset support functions
*/
int
ata_national_ident(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
/* this chip is a clone of the Cyrix chip, bugs and all */
if (pci_get_devid(dev) == ATA_SC1100) {
device_set_desc(dev, "National Geode SC1100 ATA33 controller");
ctlr->chipinit = ata_national_chipinit;
return 0;
}
return ENXIO;
}
static int
ata_national_chipinit(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
if (ata_setup_interrupt(dev))
return ENXIO;
ctlr->setmode = ata_national_setmode;
return 0;
}
static void
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_national_setmode(device_t dev, int mode)
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
{
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
device_t gparent = GRANDPARENT(dev);
struct ata_channel *ch = device_get_softc(device_get_parent(dev));
struct ata_device *atadev = device_get_softc(dev);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
int devno = (ch->unit << 1) + ATA_DEV(atadev->unit);
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
u_int32_t piotiming[] =
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
{ 0x9172d132, 0x21717121, 0x00803020, 0x20102010, 0x00100010,
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
0x00803020, 0x20102010, 0x00100010,
0x00100010, 0x00100010, 0x00100010 };
u_int32_t dmatiming[] = { 0x80077771, 0x80012121, 0x80002020 };
u_int32_t udmatiming[] = { 0x80921250, 0x80911140, 0x80911030 };
int error;
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
ch->dma->alignment = 16;
ch->dma->max_iosize = 126 * DEV_BSIZE;
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
mode = ata_limit_mode(dev, mode, ATA_UDMA2);
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
error = ata_controlcmd(dev, ATA_SETFEATURES, ATA_SF_SETXFER, 0, mode);
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
if (bootverbose)
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
device_printf(dev, "%s setting %s on National chip\n",
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
(error) ? "failed" : "success", ata_mode2str(mode));
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
if (!error) {
if (mode >= ATA_UDMA0) {
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, 0x44 + (devno << 3),
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
udmatiming[mode & ATA_MODE_MASK], 4);
}
else if (mode >= ATA_WDMA0) {
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, 0x44 + (devno << 3),
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
dmatiming[mode & ATA_MODE_MASK], 4);
}
else {
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, 0x44 + (devno << 3),
pci_read_config(gparent, 0x44 + (devno << 3), 4) |
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
0x80000000, 4);
}
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, 0x40 + (devno << 3),
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
piotiming[ata_mode2idx(mode)], 4);
atadev->mode = mode;
}
}
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
/*
* nVidia chipset support functions
*/
int
ata_nvidia_ident(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
struct ata_chip_id *idx;
static struct ata_chip_id ids[] =
Finally fix support for the newer MCP51/MCP55 nVidia chipsets. The register layout has changed since the original NV4 - sigh. Hotplug support has been fixed for all nVidia chipsets that supports it (including the MCP51/55). HW donated by: Kingsley College
2006-07-24 10:44:50 +00:00
{{ ATA_NFORCE1, 0, AMDNVIDIA, NVIDIA, ATA_UDMA5, "nForce" },
{ ATA_NFORCE2, 0, AMDNVIDIA, NVIDIA, ATA_UDMA6, "nForce2" },
{ ATA_NFORCE2_PRO, 0, AMDNVIDIA, NVIDIA, ATA_UDMA6, "nForce2 Pro" },
{ ATA_NFORCE2_PRO_S1, 0, 0, 0, ATA_SA150, "nForce2 Pro" },
{ ATA_NFORCE3, 0, AMDNVIDIA, NVIDIA, ATA_UDMA6, "nForce3" },
{ ATA_NFORCE3_PRO, 0, AMDNVIDIA, NVIDIA, ATA_UDMA6, "nForce3 Pro" },
{ ATA_NFORCE3_PRO_S1, 0, 0, 0, ATA_SA150, "nForce3 Pro" },
{ ATA_NFORCE3_PRO_S2, 0, 0, 0, ATA_SA150, "nForce3 Pro" },
{ ATA_NFORCE_MCP04, 0, AMDNVIDIA, NVIDIA, ATA_UDMA6, "nForce MCP" },
Unbreak nForce3 SATA support.
2006-08-05 11:41:46 +00:00
{ ATA_NFORCE_MCP04_S1, 0, 0, NV4, ATA_SA150, "nForce MCP" },
{ ATA_NFORCE_MCP04_S2, 0, 0, NV4, ATA_SA150, "nForce MCP" },
Finally fix support for the newer MCP51/MCP55 nVidia chipsets. The register layout has changed since the original NV4 - sigh. Hotplug support has been fixed for all nVidia chipsets that supports it (including the MCP51/55). HW donated by: Kingsley College
2006-07-24 10:44:50 +00:00
{ ATA_NFORCE_CK804, 0, AMDNVIDIA, NVIDIA, ATA_UDMA6, "nForce CK804" },
Unbreak nForce3 SATA support.
2006-08-05 11:41:46 +00:00
{ ATA_NFORCE_CK804_S1, 0, 0, NV4, ATA_SA300, "nForce CK804" },
{ ATA_NFORCE_CK804_S2, 0, 0, NV4, ATA_SA300, "nForce CK804" },
Finally fix support for the newer MCP51/MCP55 nVidia chipsets. The register layout has changed since the original NV4 - sigh. Hotplug support has been fixed for all nVidia chipsets that supports it (including the MCP51/55). HW donated by: Kingsley College
2006-07-24 10:44:50 +00:00
{ ATA_NFORCE_MCP51, 0, AMDNVIDIA, NVIDIA, ATA_UDMA6, "nForce MCP51" },
Unbreak nForce4 SATA support. Hopefully I dont break something else this time.....
2006-08-11 21:19:29 +00:00
{ ATA_NFORCE_MCP51_S1, 0, 0, NV4|NVQ, ATA_SA300, "nForce MCP51" },
{ ATA_NFORCE_MCP51_S2, 0, 0, NV4|NVQ, ATA_SA300, "nForce MCP51" },
Finally fix support for the newer MCP51/MCP55 nVidia chipsets. The register layout has changed since the original NV4 - sigh. Hotplug support has been fixed for all nVidia chipsets that supports it (including the MCP51/55). HW donated by: Kingsley College
2006-07-24 10:44:50 +00:00
{ ATA_NFORCE_MCP55, 0, AMDNVIDIA, NVIDIA, ATA_UDMA6, "nForce MCP55" },
Unbreak nForce4 SATA support. Hopefully I dont break something else this time.....
2006-08-11 21:19:29 +00:00
{ ATA_NFORCE_MCP55_S1, 0, 0, NV4|NVQ, ATA_SA300, "nForce MCP55" },
{ ATA_NFORCE_MCP55_S2, 0, 0, NV4|NVQ, ATA_SA300, "nForce MCP55" },
Cleanup the strings for printing chipset names, saves quite some space. Add a couble new nVidia chips now we are here.
2005-12-16 08:12:13 +00:00
{ 0, 0, 0, 0, 0, 0}} ;
char buffer[64] ;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
Support the ICH5 SATA part. Fix HPT374 UDMA133 timing. Fix Promise ID. Cosmetics on probe print for Promise & HPT. Approved by: re
2003-05-18 16:45:48 +00:00
if (!(idx = ata_match_chip(dev, ids)))
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
return ENXIO;
Cleanup the strings for printing chipset names, saves quite some space. Add a couble new nVidia chips now we are here.
2005-12-16 08:12:13 +00:00
sprintf(buffer, "nVidia %s %s controller",
idx->text, ata_mode2str(idx->max_dma));
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
device_set_desc_copy(dev, buffer);
ctlr->chip = idx;
ctlr->chipinit = ata_nvidia_chipinit;
return 0;
}
static int
ata_nvidia_chipinit(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
if (ata_setup_interrupt(dev))
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
return ENXIO;
Add firstshot support for Acer Labs Inc SATA parts Update ALi PATA support to handle ATA133 as well
2005-04-05 14:51:43 +00:00
if (ctlr->chip->max_dma >= ATA_SA150) {
Only try to allocate and use the SATA resource if they are enabled by the BIOS. It seems some BIOS's doesn't get this right, and that would result in ATA panic'ing.
2005-04-25 07:50:51 +00:00
if (pci_read_config(dev, PCIR_BAR(5), 1) & 1)
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ctlr->r_type2 = SYS_RES_IOPORT;
Only try to allocate and use the SATA resource if they are enabled by the BIOS. It seems some BIOS's doesn't get this right, and that would result in ATA panic'ing.
2005-04-25 07:50:51 +00:00
else
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ctlr->r_type2 = SYS_RES_MEMORY;
ctlr->r_rid2 = PCIR_BAR(5);
if ((ctlr->r_res2 = bus_alloc_resource_any(dev, ctlr->r_type2,
Only try to allocate and use the SATA resource if they are enabled by the BIOS. It seems some BIOS's doesn't get this right, and that would result in ATA panic'ing.
2005-04-25 07:50:51 +00:00
&ctlr->r_rid2, RF_ACTIVE))) {
Finally fix support for the newer MCP51/MCP55 nVidia chipsets. The register layout has changed since the original NV4 - sigh. Hotplug support has been fixed for all nVidia chipsets that supports it (including the MCP51/55). HW donated by: Kingsley College
2006-07-24 10:44:50 +00:00
int offset = ctlr->chip->cfg2 & NV4 ? 0x0440 : 0x0010;
Make nVidia SATA hotplug support DTRT. HW sponsored by: Yahoo!
2005-06-05 21:18:26 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ctlr->allocate = ata_nvidia_allocate;
ctlr->reset = ata_nvidia_reset;
Make nVidia SATA hotplug support DTRT. HW sponsored by: Yahoo!
2005-06-05 21:18:26 +00:00
/* enable control access */
pci_write_config(dev, 0x50, pci_read_config(dev, 0x50, 1) | 0x04,1);
Unbreak nForce4 SATA support. Hopefully I dont break something else this time.....
2006-08-11 21:19:29 +00:00
if (ctlr->chip->cfg2 & NVQ) {
Finally fix support for the newer MCP51/MCP55 nVidia chipsets. The register layout has changed since the original NV4 - sigh. Hotplug support has been fixed for all nVidia chipsets that supports it (including the MCP51/55). HW donated by: Kingsley College
2006-07-24 10:44:50 +00:00
/* clear interrupt status */
Unbreak nForce3 SATA support.
2006-08-05 11:41:46 +00:00
ATA_OUTL(ctlr->r_res2, offset, 0x00ff00ff);
Make nVidia SATA hotplug support DTRT. HW sponsored by: Yahoo!
2005-06-05 21:18:26 +00:00
Finally fix support for the newer MCP51/MCP55 nVidia chipsets. The register layout has changed since the original NV4 - sigh. Hotplug support has been fixed for all nVidia chipsets that supports it (including the MCP51/55). HW donated by: Kingsley College
2006-07-24 10:44:50 +00:00
/* enable device and PHY state change interrupts */
Properly initialize the nVidia MCP5X interrupts etc. Suggested by: nVidia
2006-08-30 11:50:22 +00:00
ATA_OUTL(ctlr->r_res2, offset + 4, 0x000d000d);
/* disable NCQ support */
ATA_OUTL(ctlr->r_res2, 0x0400,
ATA_INL(ctlr->r_res2, 0x0400) & 0xfffffff9);
Finally fix support for the newer MCP51/MCP55 nVidia chipsets. The register layout has changed since the original NV4 - sigh. Hotplug support has been fixed for all nVidia chipsets that supports it (including the MCP51/55). HW donated by: Kingsley College
2006-07-24 10:44:50 +00:00
}
else {
/* clear interrupt status */
Unbreak nForce3 SATA support.
2006-08-05 11:41:46 +00:00
ATA_OUTB(ctlr->r_res2, offset, 0xff);
Finally fix support for the newer MCP51/MCP55 nVidia chipsets. The register layout has changed since the original NV4 - sigh. Hotplug support has been fixed for all nVidia chipsets that supports it (including the MCP51/55). HW donated by: Kingsley College
2006-07-24 10:44:50 +00:00
/* enable device and PHY state change interrupts */
Unbreak nForce3 SATA support.
2006-08-05 11:41:46 +00:00
ATA_OUTB(ctlr->r_res2, offset + 1, 0xdd);
Finally fix support for the newer MCP51/MCP55 nVidia chipsets. The register layout has changed since the original NV4 - sigh. Hotplug support has been fixed for all nVidia chipsets that supports it (including the MCP51/55). HW donated by: Kingsley College
2006-07-24 10:44:50 +00:00
}
Make nVidia SATA hotplug support DTRT. HW sponsored by: Yahoo!
2005-06-05 21:18:26 +00:00
/* enable PCI interrupt */
Only try to allocate and use the SATA resource if they are enabled by the BIOS. It seems some BIOS's doesn't get this right, and that would result in ATA panic'ing.
2005-04-25 07:50:51 +00:00
pci_write_config(dev, PCIR_COMMAND,
pci_read_config(dev, PCIR_COMMAND, 2) & ~0x0400,2);
Make nVidia SATA hotplug support DTRT. HW sponsored by: Yahoo!
2005-06-05 21:18:26 +00:00
Only try to allocate and use the SATA resource if they are enabled by the BIOS. It seems some BIOS's doesn't get this right, and that would result in ATA panic'ing.
2005-04-25 07:50:51 +00:00
}
Add firstshot support for Acer Labs Inc SATA parts Update ALi PATA support to handle ATA133 as well
2005-04-05 14:51:43 +00:00
ctlr->setmode = ata_sata_setmode;
}
else {
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
/* disable prefetch, postwrite */
pci_write_config(dev, 0x51, pci_read_config(dev, 0x51, 1) & 0x0f, 1);
ctlr->setmode = ata_via_family_setmode;
}
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
return 0;
}
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
static int
ata_nvidia_allocate(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(device_get_parent(dev));
struct ata_channel *ch = device_get_softc(dev);
/* setup the usual register normal pci style */
Properly react to allocation failures. Found by: imp@
2005-10-10 19:12:43 +00:00
if (ata_pci_allocate(dev))
return ENXIO;
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
ch->r_io[ATA_SSTATUS].res = ctlr->r_res2;
ch->r_io[ATA_SSTATUS].offset = (ch->unit << 6);
ch->r_io[ATA_SERROR].res = ctlr->r_res2;
ch->r_io[ATA_SERROR].offset = 0x04 + (ch->unit << 6);
ch->r_io[ATA_SCONTROL].res = ctlr->r_res2;
ch->r_io[ATA_SCONTROL].offset = 0x08 + (ch->unit << 6);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ch->hw.status = ata_nvidia_status;
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
ch->flags |= ATA_NO_SLAVE;
return 0;
}
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static int
ata_nvidia_status(device_t dev)
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
{
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
struct ata_pci_controller *ctlr = device_get_softc(device_get_parent(dev));
struct ata_channel *ch = device_get_softc(dev);
Finally fix support for the newer MCP51/MCP55 nVidia chipsets. The register layout has changed since the original NV4 - sigh. Hotplug support has been fixed for all nVidia chipsets that supports it (including the MCP51/55). HW donated by: Kingsley College
2006-07-24 10:44:50 +00:00
int offset = ctlr->chip->cfg2 & NV4 ? 0x0440 : 0x0010;
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
struct ata_connect_task *tp;
Unbreak nForce4 SATA support. Hopefully I dont break something else this time.....
2006-08-11 21:19:29 +00:00
int shift = ch->unit << (ctlr->chip->cfg2 & NVQ ? 4 : 2);
Finally fix support for the newer MCP51/MCP55 nVidia chipsets. The register layout has changed since the original NV4 - sigh. Hotplug support has been fixed for all nVidia chipsets that supports it (including the MCP51/55). HW donated by: Kingsley College
2006-07-24 10:44:50 +00:00
u_int32_t status;
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
Finally fix support for the newer MCP51/MCP55 nVidia chipsets. The register layout has changed since the original NV4 - sigh. Hotplug support has been fixed for all nVidia chipsets that supports it (including the MCP51/55). HW donated by: Kingsley College
2006-07-24 10:44:50 +00:00
/* get and clear interrupt status */
Unbreak nForce4 SATA support. Hopefully I dont break something else this time.....
2006-08-11 21:19:29 +00:00
if (ctlr->chip->cfg2 & NVQ) {
Finally fix support for the newer MCP51/MCP55 nVidia chipsets. The register layout has changed since the original NV4 - sigh. Hotplug support has been fixed for all nVidia chipsets that supports it (including the MCP51/55). HW donated by: Kingsley College
2006-07-24 10:44:50 +00:00
status = ATA_INL(ctlr->r_res2, offset);
Properly initialize the nVidia MCP5X interrupts etc. Suggested by: nVidia
2006-08-30 11:50:22 +00:00
ATA_OUTL(ctlr->r_res2, offset, (0x0f << shift) | 0x00f000f0);
Finally fix support for the newer MCP51/MCP55 nVidia chipsets. The register layout has changed since the original NV4 - sigh. Hotplug support has been fixed for all nVidia chipsets that supports it (including the MCP51/55). HW donated by: Kingsley College
2006-07-24 10:44:50 +00:00
}
Unbreak nForce3 SATA support.
2006-08-05 11:41:46 +00:00
else {
status = ATA_INB(ctlr->r_res2, offset);
ATA_OUTB(ctlr->r_res2, offset, (0x0f << shift));
}
Make nVidia SATA hotplug support DTRT. HW sponsored by: Yahoo!
2005-06-05 21:18:26 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
/* check for and handle connect events */
if (((status & (0x0c << shift)) == (0x04 << shift)) &&
(tp = (struct ata_connect_task *)
malloc(sizeof(struct ata_connect_task),
M_ATA, M_NOWAIT | M_ZERO))) {
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
if (bootverbose)
device_printf(ch->dev, "CONNECT requested\n");
tp->action = ATA_C_ATTACH;
tp->dev = ch->dev;
TASK_INIT(&tp->task, 0, ata_sata_phy_event, tp);
taskqueue_enqueue(taskqueue_thread, &tp->task);
}
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
/* check for and handle disconnect events */
if ((status & (0x08 << shift)) &&
Properly initialize the nVidia MCP5X interrupts etc. Suggested by: nVidia
2006-08-30 11:50:22 +00:00
!((status & (0x04 << shift) && ATA_IDX_INL(ch, ATA_SSTATUS))) &&
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
(tp = (struct ata_connect_task *)
malloc(sizeof(struct ata_connect_task),
M_ATA, M_NOWAIT | M_ZERO))) {
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
if (bootverbose)
device_printf(ch->dev, "DISCONNECT requested\n");
tp->action = ATA_C_DETACH;
tp->dev = ch->dev;
TASK_INIT(&tp->task, 0, ata_sata_phy_event, tp);
taskqueue_enqueue(taskqueue_thread, &tp->task);
}
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
/* do we have any device action ? */
return (status & (0x01 << shift));
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
}
static void
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_nvidia_reset(device_t dev)
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
{
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
struct ata_channel *ch = device_get_softc(dev);
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
ata_sata_phy_enable(ch);
}
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
/*
* Promise chipset support functions
*/
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
#define ATA_PDC_APKT_OFFSET 0x00000010
#define ATA_PDC_HPKT_OFFSET 0x00000040
#define ATA_PDC_ASG_OFFSET 0x00000080
#define ATA_PDC_LSG_OFFSET 0x000000c0
#define ATA_PDC_HSG_OFFSET 0x00000100
#define ATA_PDC_CHN_OFFSET 0x00000400
#define ATA_PDC_BUF_BASE 0x00400000
#define ATA_PDC_BUF_OFFSET 0x00100000
#define ATA_PDC_MAX_HPKT 8
#define ATA_PDC_WRITE_REG 0x00
#define ATA_PDC_WRITE_CTL 0x0e
#define ATA_PDC_WRITE_END 0x08
#define ATA_PDC_WAIT_NBUSY 0x10
#define ATA_PDC_WAIT_READY 0x18
#define ATA_PDC_1B 0x20
#define ATA_PDC_2B 0x40
struct host_packet {
u_int32_t addr;
TAILQ_ENTRY(host_packet) chain;
};
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
struct ata_promise_sx4 {
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
struct mtx mtx;
TAILQ_HEAD(, host_packet) queue;
int busy;
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
};
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
int
ata_promise_ident(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
struct ata_chip_id *idx;
static struct ata_chip_id ids[] =
Cleanup the strings for printing chipset names, saves quite some space. Add a couble new nVidia chips now we are here.
2005-12-16 08:12:13 +00:00
{{ ATA_PDC20246, 0, PROLD, 0x00, ATA_UDMA2, "PDC20246" },
{ ATA_PDC20262, 0, PRNEW, 0x00, ATA_UDMA4, "PDC20262" },
{ ATA_PDC20263, 0, PRNEW, 0x00, ATA_UDMA4, "PDC20263" },
{ ATA_PDC20265, 0, PRNEW, 0x00, ATA_UDMA5, "PDC20265" },
{ ATA_PDC20267, 0, PRNEW, 0x00, ATA_UDMA5, "PDC20267" },
{ ATA_PDC20268, 0, PRTX, PRTX4, ATA_UDMA5, "PDC20268" },
{ ATA_PDC20269, 0, PRTX, 0x00, ATA_UDMA6, "PDC20269" },
{ ATA_PDC20270, 0, PRTX, PRTX4, ATA_UDMA5, "PDC20270" },
{ ATA_PDC20271, 0, PRTX, 0x00, ATA_UDMA6, "PDC20271" },
{ ATA_PDC20275, 0, PRTX, 0x00, ATA_UDMA6, "PDC20275" },
{ ATA_PDC20276, 0, PRTX, PRSX6K, ATA_UDMA6, "PDC20276" },
{ ATA_PDC20277, 0, PRTX, 0x00, ATA_UDMA6, "PDC20277" },
{ ATA_PDC20318, 0, PRMIO, PRSATA, ATA_SA150, "PDC20318" },
{ ATA_PDC20319, 0, PRMIO, PRSATA, ATA_SA150, "PDC20319" },
{ ATA_PDC20371, 0, PRMIO, PRCMBO, ATA_SA150, "PDC20371" },
{ ATA_PDC20375, 0, PRMIO, PRCMBO, ATA_SA150, "PDC20375" },
{ ATA_PDC20376, 0, PRMIO, PRCMBO, ATA_SA150, "PDC20376" },
{ ATA_PDC20377, 0, PRMIO, PRCMBO, ATA_SA150, "PDC20377" },
{ ATA_PDC20378, 0, PRMIO, PRCMBO, ATA_SA150, "PDC20378" },
{ ATA_PDC20379, 0, PRMIO, PRCMBO, ATA_SA150, "PDC20379" },
Unbreak Promise SATAII/150 controllers caused by the DMA dump changes.
2006-02-09 20:54:42 +00:00
{ ATA_PDC20571, 0, PRMIO, PRCMBO2, ATA_SA150, "PDC20571" },
Cleanup the strings for printing chipset names, saves quite some space. Add a couble new nVidia chips now we are here.
2005-12-16 08:12:13 +00:00
{ ATA_PDC20575, 0, PRMIO, PRCMBO2, ATA_SA150, "PDC20575" },
{ ATA_PDC20579, 0, PRMIO, PRCMBO2, ATA_SA150, "PDC20579" },
Unbreak Promise SATAII/150 controllers caused by the DMA dump changes.
2006-02-09 20:54:42 +00:00
{ ATA_PDC20771, 0, PRMIO, PRCMBO2, ATA_SA300, "PDC20771" },
Whitespace cleanup.
2006-01-18 13:10:17 +00:00
{ ATA_PDC40775, 0, PRMIO, PRCMBO2, ATA_SA300, "PDC40775" },
Cleanup the strings for printing chipset names, saves quite some space. Add a couble new nVidia chips now we are here.
2005-12-16 08:12:13 +00:00
{ ATA_PDC20617, 0, PRMIO, PRPATA, ATA_UDMA6, "PDC20617" },
{ ATA_PDC20618, 0, PRMIO, PRPATA, ATA_UDMA6, "PDC20618" },
{ ATA_PDC20619, 0, PRMIO, PRPATA, ATA_UDMA6, "PDC20619" },
{ ATA_PDC20620, 0, PRMIO, PRPATA, ATA_UDMA6, "PDC20620" },
{ ATA_PDC20621, 0, PRMIO, PRSX4X, ATA_UDMA5, "PDC20621" },
{ ATA_PDC20622, 0, PRMIO, PRSX4X, ATA_SA150, "PDC20622" },
{ ATA_PDC40518, 0, PRMIO, PRSATA2, ATA_SA150, "PDC40518" },
{ ATA_PDC40519, 0, PRMIO, PRSATA2, ATA_SA150, "PDC40519" },
{ ATA_PDC40718, 0, PRMIO, PRSATA2, ATA_SA300, "PDC40718" },
{ ATA_PDC40719, 0, PRMIO, PRSATA2, ATA_SA300, "PDC40719" },
Whitespace cleanup.
2006-01-18 13:10:17 +00:00
{ ATA_PDC40779, 0, PRMIO, PRSATA2, ATA_SA300, "PDC40779" },
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
{ 0, 0, 0, 0, 0, 0}};
Support the ICH5 SATA part. Fix HPT374 UDMA133 timing. Fix Promise ID. Cosmetics on probe print for Promise & HPT. Approved by: re
2003-05-18 16:45:48 +00:00
char buffer[64];
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
uintptr_t devid = 0;
Support the ICH5 SATA part. Fix HPT374 UDMA133 timing. Fix Promise ID. Cosmetics on probe print for Promise & HPT. Approved by: re
2003-05-18 16:45:48 +00:00
if (!(idx = ata_match_chip(dev, ids)))
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
return ENXIO;
/* if we are on a SuperTrak SX6000 dont attach */
Minor cleanup of the Promise code.
2003-03-03 11:51:08 +00:00
if ((idx->cfg2 & PRSX6K) && pci_get_class(GRANDPARENT(dev))==PCIC_BRIDGE &&
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
!BUS_READ_IVAR(device_get_parent(GRANDPARENT(dev)),
GRANDPARENT(dev), PCI_IVAR_DEVID, &devid) &&
Minor cleanup of the Promise code.
2003-03-03 11:51:08 +00:00
devid == ATA_I960RM)
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
return ENXIO;
Fix promise probe printing.
2006-01-04 08:20:25 +00:00
strcpy(buffer, "Promise ");
strcat(buffer, idx->text);
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
Minor cleanup of the Promise code.
2003-03-03 11:51:08 +00:00
/* if we are on a FastTrak TX4, adjust the interrupt resource */
if ((idx->cfg2 & PRTX4) && pci_get_class(GRANDPARENT(dev))==PCIC_BRIDGE &&
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
!BUS_READ_IVAR(device_get_parent(GRANDPARENT(dev)),
GRANDPARENT(dev), PCI_IVAR_DEVID, &devid) &&
Fix support for the Promise TX4 on amd64. Somehow the bridge on there shows up with another PCI id than it does on x86, no idea why...
2004-02-02 14:05:57 +00:00
((devid == ATA_DEC_21150) || (devid == ATA_DEC_21150_1))) {
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
static long start = 0, end = 0;
if (pci_get_slot(dev) == 1) {
bus_get_resource(dev, SYS_RES_IRQ, 0, &start, &end);
Support the ICH5 SATA part. Fix HPT374 UDMA133 timing. Fix Promise ID. Cosmetics on probe print for Promise & HPT. Approved by: re
2003-05-18 16:45:48 +00:00
strcat(buffer, " (channel 0+1)");
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
else if (pci_get_slot(dev) == 2 && start && end) {
bus_set_resource(dev, SYS_RES_IRQ, 0, start, end);
Support the ICH5 SATA part. Fix HPT374 UDMA133 timing. Fix Promise ID. Cosmetics on probe print for Promise & HPT. Approved by: re
2003-05-18 16:45:48 +00:00
strcat(buffer, " (channel 2+3)");
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
else {
start = end = 0;
}
}
Fix promise probe printing.
2006-01-04 08:20:25 +00:00
sprintf(buffer, "%s %s controller", buffer, ata_mode2str(idx->max_dma));
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
device_set_desc_copy(dev, buffer);
ctlr->chip = idx;
ctlr->chipinit = ata_promise_chipinit;
return 0;
}
static int
ata_promise_chipinit(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
Unbreak Promise SATAII/150 controllers caused by the DMA dump changes.
2006-02-09 20:54:42 +00:00
int fake_reg, stat_reg;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
if (ata_setup_interrupt(dev))
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
return ENXIO;
Further update to the Promise code. Support should now cover all old and current products.
2005-01-01 19:24:25 +00:00
Second round of updates to the ATA driver. Clean up the DMA interface too much unneeded stuff crept in with the busdma code back when. Modify the ATA_IN* / ATA_OUT* macros so that resource and offset are gotten from a table. That allows for new chipsets that doesn't nessesarily have things ordered the good old way. This also removes the need for the wierd PC98 resource functions. Tested on: i386, PC98, Alpha, Sparc64
2003-03-29 13:37:09 +00:00
switch (ctlr->chip->cfg1) {
case PRNEW:
Fix the setup of old Promise controllers, clocks was not setup right causing way too low transfer rates. Enable interrupts on old CMD64[89] chips, apparently some bogus BIOS's doesn't get this right.
2003-06-07 15:19:16 +00:00
/* setup clocks */
Add support for detaching PCI controllers. This adds support for cardbus ATA/SATA controllers. I get roughly the same transfer speeds as on true PCI controllers. Nice to be able to add a couble of "real" disks to a laptop :)
2004-03-15 12:03:48 +00:00
ATA_OUTB(ctlr->r_res1, 0x11, ATA_INB(ctlr->r_res1, 0x11) | 0x0a);
Fix the setup of old Promise controllers, clocks was not setup right causing way too low transfer rates. Enable interrupts on old CMD64[89] chips, apparently some bogus BIOS's doesn't get this right.
2003-06-07 15:19:16 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ctlr->dmainit = ata_promise_dmainit;
Second round of updates to the ATA driver. Clean up the DMA interface too much unneeded stuff crept in with the busdma code back when. Modify the ATA_IN* / ATA_OUT* macros so that resource and offset are gotten from a table. That allows for new chipsets that doesn't nessesarily have things ordered the good old way. This also removes the need for the wierd PC98 resource functions. Tested on: i386, PC98, Alpha, Sparc64
2003-03-29 13:37:09 +00:00
/* FALLTHROUGH */
case PROLD:
Fix the setup of old Promise controllers, clocks was not setup right causing way too low transfer rates. Enable interrupts on old CMD64[89] chips, apparently some bogus BIOS's doesn't get this right.
2003-06-07 15:19:16 +00:00
/* enable burst mode */
Add support for detaching PCI controllers. This adds support for cardbus ATA/SATA controllers. I get roughly the same transfer speeds as on true PCI controllers. Nice to be able to add a couble of "real" disks to a laptop :)
2004-03-15 12:03:48 +00:00
ATA_OUTB(ctlr->r_res1, 0x1f, ATA_INB(ctlr->r_res1, 0x1f) | 0x01);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ctlr->allocate = ata_promise_allocate;
ctlr->setmode = ata_promise_setmode;
return 0;
Second round of updates to the ATA driver. Clean up the DMA interface too much unneeded stuff crept in with the busdma code back when. Modify the ATA_IN* / ATA_OUT* macros so that resource and offset are gotten from a table. That allows for new chipsets that doesn't nessesarily have things ordered the good old way. This also removes the need for the wierd PC98 resource functions. Tested on: i386, PC98, Alpha, Sparc64
2003-03-29 13:37:09 +00:00
case PRTX:
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ctlr->allocate = ata_promise_tx2_allocate;
ctlr->setmode = ata_promise_setmode;
return 0;
I'm pleased to announce that Promise is now supporting the FreeBSD project by providing documentation (under NDA) and hardware for testing. This commit is the first result of the cooperation, and adds support for several of their new controllers that we didn't support before (and probably newer would have without this arrangement). Add support for the Promise SATA150 TX2/TX4 and the Promise TX4000 controllers. This also adds support for various motherboard fitted Promise SATA/ATA chips. Note that this code uses memory mapped registers to minimize overhead. I belive FreeBSD has made another first in the Open Source world by being able to release support for this :)
2003-05-01 06:20:50 +00:00
case PRMIO:
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
ctlr->r_type1 = SYS_RES_MEMORY;
Try to narrow down the race window on HW that does not have ways to poll for which channel actually pulled the irq line.
2004-08-05 21:13:41 +00:00
ctlr->r_rid1 = PCIR_BAR(4);
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
if (!(ctlr->r_res1 = bus_alloc_resource_any(dev, ctlr->r_type1,
&ctlr->r_rid1, RF_ACTIVE)))
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
goto failnfree;
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
Add support for detaching PCI controllers. This adds support for cardbus ATA/SATA controllers. I get roughly the same transfer speeds as on true PCI controllers. Nice to be able to add a couble of "real" disks to a laptop :)
2004-03-15 12:03:48 +00:00
ctlr->r_type2 = SYS_RES_MEMORY;
Try to narrow down the race window on HW that does not have ways to poll for which channel actually pulled the irq line.
2004-08-05 21:13:41 +00:00
ctlr->r_rid2 = PCIR_BAR(3);
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
if (!(ctlr->r_res2 = bus_alloc_resource_any(dev, ctlr->r_type2,
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
&ctlr->r_rid2, RF_ACTIVE)))
goto failnfree;
I'm pleased to announce that Promise is now supporting the FreeBSD project by providing documentation (under NDA) and hardware for testing. This commit is the first result of the cooperation, and adds support for several of their new controllers that we didn't support before (and probably newer would have without this arrangement). Add support for the Promise SATA150 TX2/TX4 and the Promise TX4000 controllers. This also adds support for various motherboard fitted Promise SATA/ATA chips. Note that this code uses memory mapped registers to minimize overhead. I belive FreeBSD has made another first in the Open Source world by being able to release support for this :)
2003-05-01 06:20:50 +00:00
Unbreak Promise SATAII/150 controllers caused by the DMA dump changes.
2006-02-09 20:54:42 +00:00
if (ctlr->chip->cfg2 == PRSX4X) {
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
struct ata_promise_sx4 *hpkt;
u_int32_t dimm = ATA_INL(ctlr->r_res2, 0x000c0080);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
if (bus_teardown_intr(dev, ctlr->r_irq, ctlr->handle) ||
Whitespace cleanup.
2006-01-18 13:10:17 +00:00
bus_setup_intr(dev, ctlr->r_irq, ATA_INTR_FLAGS,
ata_promise_sx4_intr, ctlr, &ctlr->handle)) {
device_printf(dev, "unable to setup interrupt\n");
goto failnfree;
}
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
/* print info about cache memory */
device_printf(dev, "DIMM size %dMB @ 0x%08x%s\n",
Increase robustness of SATA handling.
2004-06-15 11:02:09 +00:00
(((dimm >> 16) & 0xff)-((dimm >> 24) & 0xff)+1) << 4,
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
((dimm >> 24) & 0xff),
ATA_INL(ctlr->r_res2, 0x000c0088) & (1<<16) ?
" ECC enabled" : "" );
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
/* adjust cache memory parameters */
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
ATA_OUTL(ctlr->r_res2, 0x000c000c,
(ATA_INL(ctlr->r_res2, 0x000c000c) & 0xffff0000));
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
/* setup host packet controls */
hpkt = malloc(sizeof(struct ata_promise_sx4),
M_TEMP, M_NOWAIT | M_ZERO);
Use the right ordering of args on mtx_init(). No functional changes since the args in question was all zero's. Found by: Jimmy Olgeni <olgeni@FreeBSD.org>
2004-06-22 11:18:25 +00:00
mtx_init(&hpkt->mtx, "ATA promise HPKT lock", NULL, MTX_DEF);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
TAILQ_INIT(&hpkt->queue);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
hpkt->busy = 0;
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
device_set_ivars(dev, hpkt);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ctlr->allocate = ata_promise_mio_allocate;
ctlr->reset = ata_promise_mio_reset;
ctlr->dmainit = ata_promise_mio_dmainit;
ctlr->setmode = ata_promise_setmode;
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
ctlr->channels = 4;
Further update to the Promise code. Support should now cover all old and current products.
2005-01-01 19:24:25 +00:00
return 0;
Unbreak Promise SATAII/150 controllers caused by the DMA dump changes.
2006-02-09 20:54:42 +00:00
}
/* mio type controllers need an interrupt intercept */
if (bus_teardown_intr(dev, ctlr->r_irq, ctlr->handle) ||
Whitespace cleanup.
2006-01-18 13:10:17 +00:00
bus_setup_intr(dev, ctlr->r_irq, ATA_INTR_FLAGS,
ata_promise_mio_intr, ctlr, &ctlr->handle)) {
device_printf(dev, "unable to setup interrupt\n");
goto failnfree;
I'm pleased to announce that Promise is now supporting the FreeBSD project by providing documentation (under NDA) and hardware for testing. This commit is the first result of the cooperation, and adds support for several of their new controllers that we didn't support before (and probably newer would have without this arrangement). Add support for the Promise SATA150 TX2/TX4 and the Promise TX4000 controllers. This also adds support for various motherboard fitted Promise SATA/ATA chips. Note that this code uses memory mapped registers to minimize overhead. I belive FreeBSD has made another first in the Open Source world by being able to release support for this :)
2003-05-01 06:20:50 +00:00
}
Further update to the Promise code. Support should now cover all old and current products.
2005-01-01 19:24:25 +00:00
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
switch (ctlr->chip->cfg2) {
case PRPATA:
ctlr->channels = ((ATA_INL(ctlr->r_res2, 0x48) & 0x01) > 0) +
((ATA_INL(ctlr->r_res2, 0x48) & 0x02) > 0) + 2;
Unbreak Promise SATAII/150 controllers caused by the DMA dump changes.
2006-02-09 20:54:42 +00:00
goto sata150;
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
case PRCMBO:
Unbreak Promise SATAII/150 controllers caused by the DMA dump changes.
2006-02-09 20:54:42 +00:00
ctlr->channels = 3;
goto sata150;
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
case PRSATA:
ctlr->channels = 4;
Unbreak Promise SATAII/150 controllers caused by the DMA dump changes.
2006-02-09 20:54:42 +00:00
sata150:
fake_reg = 0x60;
stat_reg = 0x6c;
break;
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
Unbreak Promise SATAII/150 controllers caused by the DMA dump changes.
2006-02-09 20:54:42 +00:00
case PRCMBO2:
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
ctlr->channels = 3;
Unbreak Promise SATAII/150 controllers caused by the DMA dump changes.
2006-02-09 20:54:42 +00:00
goto sataii;
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
case PRSATA2:
Unbreak Promise SATAII/150 controllers caused by the DMA dump changes.
2006-02-09 20:54:42 +00:00
default:
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
ctlr->channels = 4;
Unbreak Promise SATAII/150 controllers caused by the DMA dump changes.
2006-02-09 20:54:42 +00:00
sataii:
fake_reg = 0x54;
stat_reg = 0x60;
break;
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
}
Unbreak Promise SATAII/150 controllers caused by the DMA dump changes.
2006-02-09 20:54:42 +00:00
/* prime fake interrupt register */
ATA_OUTL(ctlr->r_res2, fake_reg, 0xffffffff);
/* clear SATA status */
ATA_OUTL(ctlr->r_res2, stat_reg, 0x000000ff);
ctlr->allocate = ata_promise_mio_allocate;
ctlr->reset = ata_promise_mio_reset;
ctlr->dmainit = ata_promise_mio_dmainit;
ctlr->setmode = ata_promise_mio_setmode;
return 0;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
failnfree:
if (ctlr->r_res2)
bus_release_resource(dev, ctlr->r_type2, ctlr->r_rid2, ctlr->r_res2);
if (ctlr->r_res1)
bus_release_resource(dev, ctlr->r_type1, ctlr->r_rid1, ctlr->r_res1);
return ENXIO;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
I'm pleased to announce that Promise is now supporting the FreeBSD project by providing documentation (under NDA) and hardware for testing. This commit is the first result of the cooperation, and adds support for several of their new controllers that we didn't support before (and probably newer would have without this arrangement). Add support for the Promise SATA150 TX2/TX4 and the Promise TX4000 controllers. This also adds support for various motherboard fitted Promise SATA/ATA chips. Note that this code uses memory mapped registers to minimize overhead. I belive FreeBSD has made another first in the Open Source world by being able to release support for this :)
2003-05-01 06:20:50 +00:00
static int
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ata_promise_allocate(device_t dev)
I'm pleased to announce that Promise is now supporting the FreeBSD project by providing documentation (under NDA) and hardware for testing. This commit is the first result of the cooperation, and adds support for several of their new controllers that we didn't support before (and probably newer would have without this arrangement). Add support for the Promise SATA150 TX2/TX4 and the Promise TX4000 controllers. This also adds support for various motherboard fitted Promise SATA/ATA chips. Note that this code uses memory mapped registers to minimize overhead. I belive FreeBSD has made another first in the Open Source world by being able to release support for this :)
2003-05-01 06:20:50 +00:00
{
Add firstshot support for Acer Labs Inc SATA parts Update ALi PATA support to handle ATA133 as well
2005-04-05 14:51:43 +00:00
struct ata_channel *ch = device_get_softc(dev);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
if (ata_pci_allocate(dev))
return ENXIO;
ch->hw.status = ata_promise_status;
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
return 0;
}
I'm pleased to announce that Promise is now supporting the FreeBSD project by providing documentation (under NDA) and hardware for testing. This commit is the first result of the cooperation, and adds support for several of their new controllers that we didn't support before (and probably newer would have without this arrangement). Add support for the Promise SATA150 TX2/TX4 and the Promise TX4000 controllers. This also adds support for various motherboard fitted Promise SATA/ATA chips. Note that this code uses memory mapped registers to minimize overhead. I belive FreeBSD has made another first in the Open Source world by being able to release support for this :)
2003-05-01 06:20:50 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static int
ata_promise_status(device_t dev)
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
{
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
struct ata_pci_controller *ctlr = device_get_softc(device_get_parent(dev));
struct ata_channel *ch = device_get_softc(dev);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
if (ATA_INL(ctlr->r_res1, 0x1c) & (ch->unit ? 0x00004000 : 0x00000400)) {
return ata_pci_status(dev);
Add first shot on support for the new Promise SATAII chips. HW donated by: pil.dk
2004-12-08 10:02:41 +00:00
}
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
return 0;
}
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static int
ata_promise_dmastart(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(GRANDPARENT(dev));
struct ata_channel *ch = device_get_softc(device_get_parent(dev));
struct ata_device *atadev = device_get_softc(dev);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
if (atadev->flags & ATA_D_48BIT_ACTIVE) {
ATA_OUTB(ctlr->r_res1, 0x11,
ATA_INB(ctlr->r_res1, 0x11) | (ch->unit ? 0x08 : 0x02));
ATA_OUTL(ctlr->r_res1, ch->unit ? 0x24 : 0x20,
((ch->dma->flags & ATA_DMA_READ) ? 0x05000000 : 0x06000000) |
(ch->dma->cur_iosize >> 1));
}
ATA_IDX_OUTB(ch, ATA_BMSTAT_PORT, (ATA_IDX_INB(ch, ATA_BMSTAT_PORT) |
(ATA_BMSTAT_INTERRUPT | ATA_BMSTAT_ERROR)));
ATA_IDX_OUTL(ch, ATA_BMDTP_PORT, ch->dma->sg_bus);
ATA_IDX_OUTB(ch, ATA_BMCMD_PORT,
((ch->dma->flags & ATA_DMA_READ) ? ATA_BMCMD_WRITE_READ : 0) |
ATA_BMCMD_START_STOP);
ch->flags |= ATA_DMA_ACTIVE;
return 0;
}
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static int
ata_promise_dmastop(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(GRANDPARENT(dev));
struct ata_channel *ch = device_get_softc(device_get_parent(dev));
struct ata_device *atadev = device_get_softc(dev);
int error;
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
if (atadev->flags & ATA_D_48BIT_ACTIVE) {
ATA_OUTB(ctlr->r_res1, 0x11,
ATA_INB(ctlr->r_res1, 0x11) & ~(ch->unit ? 0x08 : 0x02));
ATA_OUTL(ctlr->r_res1, ch->unit ? 0x24 : 0x20, 0);
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
}
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
error = ATA_IDX_INB(ch, ATA_BMSTAT_PORT);
ATA_IDX_OUTB(ch, ATA_BMCMD_PORT,
ATA_IDX_INB(ch, ATA_BMCMD_PORT) & ~ATA_BMCMD_START_STOP);
ATA_IDX_OUTB(ch, ATA_BMSTAT_PORT, ATA_BMSTAT_INTERRUPT | ATA_BMSTAT_ERROR);
ch->flags &= ~ATA_DMA_ACTIVE;
return error;
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
}
static void
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ata_promise_dmareset(device_t dev)
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
{
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
struct ata_channel *ch = device_get_softc(dev);
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ATA_IDX_OUTB(ch, ATA_BMCMD_PORT,
ATA_IDX_INB(ch, ATA_BMCMD_PORT) & ~ATA_BMCMD_START_STOP);
ATA_IDX_OUTB(ch, ATA_BMSTAT_PORT, ATA_BMSTAT_INTERRUPT | ATA_BMSTAT_ERROR);
ch->flags &= ~ATA_DMA_ACTIVE;
}
static void
ata_promise_dmainit(device_t dev)
{
struct ata_channel *ch = device_get_softc(dev);
ata_dmainit(dev);
if (ch->dma) {
ch->dma->start = ata_promise_dmastart;
ch->dma->stop = ata_promise_dmastop;
ch->dma->reset = ata_promise_dmareset;
}
}
static void
ata_promise_setmode(device_t dev, int mode)
{
device_t gparent = GRANDPARENT(dev);
struct ata_pci_controller *ctlr = device_get_softc(gparent);
struct ata_channel *ch = device_get_softc(device_get_parent(dev));
struct ata_device *atadev = device_get_softc(dev);
int devno = (ch->unit << 1) + ATA_DEV(atadev->unit);
int error;
u_int32_t timings[][2] = {
/* PROLD PRNEW mode */
{ 0x004ff329, 0x004fff2f }, /* PIO 0 */
{ 0x004fec25, 0x004ff82a }, /* PIO 1 */
{ 0x004fe823, 0x004ff026 }, /* PIO 2 */
{ 0x004fe622, 0x004fec24 }, /* PIO 3 */
{ 0x004fe421, 0x004fe822 }, /* PIO 4 */
{ 0x004567f3, 0x004acef6 }, /* MWDMA 0 */
{ 0x004467f3, 0x0048cef6 }, /* MWDMA 1 */
{ 0x004367f3, 0x0046cef6 }, /* MWDMA 2 */
{ 0x004367f3, 0x0046cef6 }, /* UDMA 0 */
{ 0x004247f3, 0x00448ef6 }, /* UDMA 1 */
{ 0x004127f3, 0x00436ef6 }, /* UDMA 2 */
{ 0, 0x00424ef6 }, /* UDMA 3 */
{ 0, 0x004127f3 }, /* UDMA 4 */
{ 0, 0x004127f3 } /* UDMA 5 */
};
mode = ata_limit_mode(dev, mode, ctlr->chip->max_dma);
switch (ctlr->chip->cfg1) {
case PROLD:
case PRNEW:
if (mode > ATA_UDMA2 && (pci_read_config(gparent, 0x50, 2) &
(ch->unit ? 1 << 11 : 1 << 10))) {
ata_print_cable(dev, "controller");
mode = ATA_UDMA2;
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
}
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
if (ata_atapi(dev) && mode > ATA_PIO_MAX)
mode = ata_limit_mode(dev, mode, ATA_PIO_MAX);
break;
case PRTX:
ATA_IDX_OUTB(ch, ATA_BMDEVSPEC_0, 0x0b);
if (mode > ATA_UDMA2 &&
ATA_IDX_INB(ch, ATA_BMDEVSPEC_1) & 0x04) {
ata_print_cable(dev, "controller");
mode = ATA_UDMA2;
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
}
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
break;
case PRMIO:
if (mode > ATA_UDMA2 &&
(ATA_INL(ctlr->r_res2,
(ctlr->chip->cfg2 & PRSX4X ? 0x000c0260 : 0x0260) +
(ch->unit << 7)) & 0x01000000)) {
ata_print_cable(dev, "controller");
mode = ATA_UDMA2;
}
break;
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
}
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
error = ata_controlcmd(dev, ATA_SETFEATURES, ATA_SF_SETXFER, 0, mode);
if (bootverbose)
device_printf(dev, "%ssetting %s on %s chip\n",
(error) ? "FAILURE " : "",
ata_mode2str(mode), ctlr->chip->text);
if (!error) {
if (ctlr->chip->cfg1 < PRTX)
pci_write_config(gparent, 0x60 + (devno << 2),
timings[ata_mode2idx(mode)][ctlr->chip->cfg1], 4);
atadev->mode = mode;
}
return;
}
static int
ata_promise_tx2_allocate(device_t dev)
{
struct ata_channel *ch = device_get_softc(dev);
if (ata_pci_allocate(dev))
return ENXIO;
ch->hw.status = ata_promise_tx2_status;
return 0;
}
static int
ata_promise_tx2_status(device_t dev)
{
struct ata_channel *ch = device_get_softc(dev);
ATA_IDX_OUTB(ch, ATA_BMDEVSPEC_0, 0x0b);
if (ATA_IDX_INB(ch, ATA_BMDEVSPEC_1) & 0x20) {
return ata_pci_status(dev);
}
return 0;
}
static int
ata_promise_mio_allocate(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(device_get_parent(dev));
struct ata_channel *ch = device_get_softc(dev);
int offset = (ctlr->chip->cfg2 & PRSX4X) ? 0x000c0000 : 0;
int i;
for (i = ATA_DATA; i <= ATA_COMMAND; i++) {
ch->r_io[i].res = ctlr->r_res2;
ch->r_io[i].offset = offset + 0x0200 + (i << 2) + (ch->unit << 7);
}
ch->r_io[ATA_CONTROL].res = ctlr->r_res2;
ch->r_io[ATA_CONTROL].offset = offset + 0x0238 + (ch->unit << 7);
ch->r_io[ATA_IDX_ADDR].res = ctlr->r_res2;
ata_default_registers(dev);
if ((ctlr->chip->cfg2 & (PRSATA | PRSATA2)) ||
((ctlr->chip->cfg2 & (PRCMBO | PRCMBO2)) && ch->unit < 2)) {
ch->r_io[ATA_SSTATUS].res = ctlr->r_res2;
ch->r_io[ATA_SSTATUS].offset = 0x400 + (ch->unit << 8);
ch->r_io[ATA_SERROR].res = ctlr->r_res2;
ch->r_io[ATA_SERROR].offset = 0x404 + (ch->unit << 8);
ch->r_io[ATA_SCONTROL].res = ctlr->r_res2;
ch->r_io[ATA_SCONTROL].offset = 0x408 + (ch->unit << 8);
ch->flags |= ATA_NO_SLAVE;
}
ch->flags |= ATA_USE_16BIT;
ata_generic_hw(dev);
if (ctlr->chip->cfg2 & PRSX4X) {
ch->hw.command = ata_promise_sx4_command;
}
else {
ch->hw.command = ata_promise_mio_command;
ch->hw.status = ata_promise_mio_status;
}
return 0;
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
}
Fix Sii3114 support.
2004-05-10 20:23:25 +00:00
static void
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ata_promise_mio_intr(void *data)
Fix Sii3114 support.
2004-05-10 20:23:25 +00:00
{
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
struct ata_pci_controller *ctlr = data;
struct ata_channel *ch;
Unbreak Promise SATAII/150 controllers caused by the DMA dump changes.
2006-02-09 20:54:42 +00:00
u_int32_t vector;
int unit, fake_reg;
switch (ctlr->chip->cfg2) {
case PRPATA:
case PRCMBO:
case PRSATA:
fake_reg = 0x60;
break;
case PRCMBO2:
case PRSATA2:
default:
fake_reg = 0x54;
break;
}
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
/*
* since reading interrupt status register on early "mio" chips
* clears the status bits we cannot read it for each channel later on
* in the generic interrupt routine.
* store the bits in an unused register in the chip so we can read
* it from there safely to get around this "feature".
*/
Unbreak Promise SATAII/150 controllers caused by the DMA dump changes.
2006-02-09 20:54:42 +00:00
vector = ATA_INL(ctlr->r_res2, 0x040);
ATA_OUTL(ctlr->r_res2, 0x040, vector);
ATA_OUTL(ctlr->r_res2, fake_reg, vector);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
for (unit = 0; unit < ctlr->channels; unit++) {
Whitespace cleanup.
2006-01-18 13:10:17 +00:00
if ((ch = ctlr->interrupt[unit].argument))
ctlr->interrupt[unit].function(ch);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
}
Unbreak Promise SATAII/150 controllers caused by the DMA dump changes.
2006-02-09 20:54:42 +00:00
ATA_OUTL(ctlr->r_res2, fake_reg, 0xffffffff);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
}
static int
ata_promise_mio_status(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(device_get_parent(dev));
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
struct ata_channel *ch = device_get_softc(dev);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
struct ata_connect_task *tp;
Unbreak Promise SATAII/150 controllers caused by the DMA dump changes.
2006-02-09 20:54:42 +00:00
u_int32_t fake_reg, stat_reg, vector, status;
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
switch (ctlr->chip->cfg2) {
case PRPATA:
case PRCMBO:
Unbreak Promise SATAII/150 controllers caused by the DMA dump changes.
2006-02-09 20:54:42 +00:00
case PRSATA:
fake_reg = 0x60;
stat_reg = 0x6c;
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
break;
Unbreak Promise SATAII/150 controllers caused by the DMA dump changes.
2006-02-09 20:54:42 +00:00
case PRCMBO2:
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
case PRSATA2:
default:
Unbreak Promise SATAII/150 controllers caused by the DMA dump changes.
2006-02-09 20:54:42 +00:00
fake_reg = 0x54;
stat_reg = 0x60;
break;
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
}
Unbreak Promise SATAII/150 controllers caused by the DMA dump changes.
2006-02-09 20:54:42 +00:00
/* read and acknowledge interrupt */
vector = ATA_INL(ctlr->r_res2, fake_reg);
/* read and clear interface status */
status = ATA_INL(ctlr->r_res2, stat_reg);
ATA_OUTL(ctlr->r_res2, stat_reg, status & (0x00000011 << ch->unit));
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
/* check for and handle disconnect events */
if ((status & (0x00000001 << ch->unit)) &&
(tp = (struct ata_connect_task *)
malloc(sizeof(struct ata_connect_task),
M_ATA, M_NOWAIT | M_ZERO))) {
if (bootverbose)
device_printf(ch->dev, "DISCONNECT requested\n");
tp->action = ATA_C_DETACH;
tp->dev = ch->dev;
TASK_INIT(&tp->task, 0, ata_sata_phy_event, tp);
taskqueue_enqueue(taskqueue_thread, &tp->task);
}
/* check for and handle connect events */
if ((status & (0x00000010 << ch->unit)) &&
(tp = (struct ata_connect_task *)
malloc(sizeof(struct ata_connect_task),
M_ATA, M_NOWAIT | M_ZERO))) {
if (bootverbose)
device_printf(ch->dev, "CONNECT requested\n");
tp->action = ATA_C_ATTACH;
tp->dev = ch->dev;
TASK_INIT(&tp->task, 0, ata_sata_phy_event, tp);
taskqueue_enqueue(taskqueue_thread, &tp->task);
}
/* do we have any device action ? */
return (vector & (1 << (ch->unit + 1)));
}
static int
ata_promise_mio_command(struct ata_request *request)
{
struct ata_pci_controller *ctlr=device_get_softc(GRANDPARENT(request->dev));
struct ata_channel *ch = device_get_softc(device_get_parent(request->dev));
u_int32_t *wordp = (u_int32_t *)ch->dma->work;
ATA_OUTL(ctlr->r_res2, (ch->unit + 1) << 2, 0x00000001);
/* XXX SOS add ATAPI commands support later */
switch (request->u.ata.command) {
default:
return ata_generic_command(request);
case ATA_READ_DMA:
case ATA_READ_DMA48:
wordp[0] = htole32(0x04 | ((ch->unit + 1) << 16) | (0x00 << 24));
break;
case ATA_WRITE_DMA:
case ATA_WRITE_DMA48:
wordp[0] = htole32(0x00 | ((ch->unit + 1) << 16) | (0x00 << 24));
break;
Fix Sii3114 support.
2004-05-10 20:23:25 +00:00
}
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
wordp[1] = htole32(ch->dma->sg_bus);
wordp[2] = 0;
ata_promise_apkt((u_int8_t*)wordp, request);
ATA_OUTL(ctlr->r_res2, 0x0240 + (ch->unit << 7), ch->dma->work_bus);
return 0;
Fix Sii3114 support.
2004-05-10 20:23:25 +00:00
}
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
static void
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_promise_mio_reset(device_t dev)
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
{
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
struct ata_pci_controller *ctlr = device_get_softc(device_get_parent(dev));
struct ata_channel *ch = device_get_softc(dev);
Generalise the SATA PHY handling code so it wont be duplicated for each SATA chip. Promise and Silicon Image are the current candidates for this.
2005-04-08 09:37:47 +00:00
struct ata_promise_sx4 *hpktp;
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
Further update to the Promise code. Support should now cover all old and current products.
2005-01-01 19:24:25 +00:00
switch (ctlr->chip->cfg2) {
Generalise the SATA PHY handling code so it wont be duplicated for each SATA chip. Promise and Silicon Image are the current candidates for this.
2005-04-08 09:37:47 +00:00
case PRSX4X:
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
Further update to the Promise code. Support should now cover all old and current products.
2005-01-01 19:24:25 +00:00
/* softreset channel ATA module */
Generalise the SATA PHY handling code so it wont be duplicated for each SATA chip. Promise and Silicon Image are the current candidates for this.
2005-04-08 09:37:47 +00:00
hpktp = device_get_ivars(ctlr->dev);
Fix the reset sequence for sx4/sx4000 type controllers, it could hang the chip in an indefinite state.
2004-07-30 13:33:09 +00:00
ATA_OUTL(ctlr->r_res2, 0xc0260 + (ch->unit << 7), ch->unit + 1);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
ata_udelay(1000);
Increase robustness of SATA handling.
2004-06-15 11:02:09 +00:00
ATA_OUTL(ctlr->r_res2, 0xc0260 + (ch->unit << 7),
(ATA_INL(ctlr->r_res2, 0xc0260 + (ch->unit << 7)) &
~0x00003f9f) | (ch->unit + 1));
Cleanup comments
2005-05-13 10:25:19 +00:00
/* softreset HOST module */ /* XXX SOS what about other outstandings */
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
mtx_lock(&hpktp->mtx);
ATA_OUTL(ctlr->r_res2, 0xc012c,
(ATA_INL(ctlr->r_res2, 0xc012c) & ~0x00000f9f) | (1 << 11));
DELAY(10);
ATA_OUTL(ctlr->r_res2, 0xc012c,
(ATA_INL(ctlr->r_res2, 0xc012c) & ~0x00000f9f));
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
hpktp->busy = 0;
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
mtx_unlock(&hpktp->mtx);
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_generic_reset(dev);
Further update to the Promise code. Support should now cover all old and current products.
2005-01-01 19:24:25 +00:00
break;
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
case PRPATA:
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
case PRCMBO:
Generalise the SATA PHY handling code so it wont be duplicated for each SATA chip. Promise and Silicon Image are the current candidates for this.
2005-04-08 09:37:47 +00:00
case PRSATA:
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
if ((ctlr->chip->cfg2 == PRSATA) ||
((ctlr->chip->cfg2 == PRCMBO) && (ch->unit < 2))) {
/* mask plug/unplug intr */
ATA_OUTL(ctlr->r_res2, 0x06c, (0x00110000 << ch->unit));
}
Add first shot on support for the new Promise SATAII chips. HW donated by: pil.dk
2004-12-08 10:02:41 +00:00
/* softreset channels ATA module */
ATA_OUTL(ctlr->r_res2, 0x0260 + (ch->unit << 7), (1 << 11));
ata_udelay(10000);
ATA_OUTL(ctlr->r_res2, 0x0260 + (ch->unit << 7),
(ATA_INL(ctlr->r_res2, 0x0260 + (ch->unit << 7)) &
~0x00003f9f) | (ch->unit + 1));
Increase robustness of SATA handling.
2004-06-15 11:02:09 +00:00
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
if ((ctlr->chip->cfg2 == PRSATA) ||
((ctlr->chip->cfg2 == PRCMBO) && (ch->unit < 2))) {
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
ata_sata_phy_enable(ch);
Increase robustness of SATA handling.
2004-06-15 11:02:09 +00:00
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
/* reset and enable plug/unplug intr */
ATA_OUTL(ctlr->r_res2, 0x06c, (0x00000011 << ch->unit));
}
Rearrange the way the reset code is called. Prepare for different looking controllers.
2005-04-28 22:08:08 +00:00
else
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_generic_reset(dev);
Further update to the Promise code. Support should now cover all old and current products.
2005-01-01 19:24:25 +00:00
break;
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
case PRCMBO2:
Generalise the SATA PHY handling code so it wont be duplicated for each SATA chip. Promise and Silicon Image are the current candidates for this.
2005-04-08 09:37:47 +00:00
case PRSATA2:
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
if ((ctlr->chip->cfg2 == PRSATA2) ||
((ctlr->chip->cfg2 == PRCMBO2) && (ch->unit < 2))) {
/* set portmultiplier port */
ATA_OUTL(ctlr->r_res2, 0x4e8 + (ch->unit << 8), 0x0f);
Increase robustness of SATA handling.
2004-06-15 11:02:09 +00:00
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
/* mask plug/unplug intr */
ATA_OUTL(ctlr->r_res2, 0x060, (0x00110000 << ch->unit));
}
Add first shot on support for the new Promise SATAII chips. HW donated by: pil.dk
2004-12-08 10:02:41 +00:00
/* softreset channels ATA module */
ATA_OUTL(ctlr->r_res2, 0x0260 + (ch->unit << 7), (1 << 11));
ata_udelay(10000);
Increase robustness of SATA handling.
2004-06-15 11:02:09 +00:00
ATA_OUTL(ctlr->r_res2, 0x0260 + (ch->unit << 7),
(ATA_INL(ctlr->r_res2, 0x0260 + (ch->unit << 7)) &
~0x00003f9f) | (ch->unit + 1));
Add first shot on support for the new Promise SATAII chips. HW donated by: pil.dk
2004-12-08 10:02:41 +00:00
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
if ((ctlr->chip->cfg2 == PRSATA2) ||
((ctlr->chip->cfg2 == PRCMBO2) && (ch->unit < 2))) {
Add first shot on support for the new Promise SATAII chips. HW donated by: pil.dk
2004-12-08 10:02:41 +00:00
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
/* set PHY mode to "improved" */
ATA_OUTL(ctlr->r_res2, 0x414 + (ch->unit << 8),
(ATA_INL(ctlr->r_res2, 0x414 + (ch->unit << 8)) &
~0x00000003) | 0x00000001);
Add first shot on support for the new Promise SATAII chips. HW donated by: pil.dk
2004-12-08 10:02:41 +00:00
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
ata_sata_phy_enable(ch);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
/* reset and enable plug/unplug intr */
ATA_OUTL(ctlr->r_res2, 0x060, (0x00000011 << ch->unit));
Add first shot on support for the new Promise SATAII chips. HW donated by: pil.dk
2004-12-08 10:02:41 +00:00
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
/* set portmultiplier port */
ATA_OUTL(ctlr->r_res2, 0x4e8 + (ch->unit << 8), 0x00);
}
Rearrange the way the reset code is called. Prepare for different looking controllers.
2005-04-28 22:08:08 +00:00
else
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_generic_reset(dev);
Further update to the Promise code. Support should now cover all old and current products.
2005-01-01 19:24:25 +00:00
break;
Rearrange the way the reset code is called. Prepare for different looking controllers.
2005-04-28 22:08:08 +00:00
Increase robustness of SATA handling.
2004-06-15 11:02:09 +00:00
}
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
}
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static void
ata_promise_mio_dmainit(device_t dev)
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
{
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
/* note start and stop are not used here */
ata_dmainit(dev);
}
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static void
ata_promise_mio_setmode(device_t dev, int mode)
{
device_t gparent = GRANDPARENT(dev);
struct ata_pci_controller *ctlr = device_get_softc(gparent);
struct ata_channel *ch = device_get_softc(device_get_parent(dev));
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
if ( (ctlr->chip->cfg2 == PRSATA) ||
((ctlr->chip->cfg2 == PRCMBO) && (ch->unit < 2)) ||
(ctlr->chip->cfg2 == PRSATA2) ||
((ctlr->chip->cfg2 == PRCMBO2) && (ch->unit < 2)))
ata_sata_setmode(dev, mode);
else
ata_promise_setmode(dev, mode);
}
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static void
ata_promise_sx4_intr(void *data)
{
struct ata_pci_controller *ctlr = data;
struct ata_channel *ch;
u_int32_t vector = ATA_INL(ctlr->r_res2, 0x000c0480);
int unit;
Sync promise_mio_command with WIP on the Promise line of SATA controllers.
2004-09-03 08:23:04 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
for (unit = 0; unit < ctlr->channels; unit++) {
if (vector & (1 << (unit + 1)))
if ((ch = ctlr->interrupt[unit].argument))
ctlr->interrupt[unit].function(ch);
if (vector & (1 << (unit + 5)))
if ((ch = ctlr->interrupt[unit].argument))
ata_promise_queue_hpkt(ctlr,
htole32((ch->unit * ATA_PDC_CHN_OFFSET) +
ATA_PDC_HPKT_OFFSET));
if (vector & (1 << (unit + 9))) {
ata_promise_next_hpkt(ctlr);
if ((ch = ctlr->interrupt[unit].argument))
ctlr->interrupt[unit].function(ch);
}
if (vector & (1 << (unit + 13))) {
ata_promise_next_hpkt(ctlr);
if ((ch = ctlr->interrupt[unit].argument))
ATA_OUTL(ctlr->r_res2, 0x000c0240 + (ch->unit << 7),
htole32((ch->unit * ATA_PDC_CHN_OFFSET) +
ATA_PDC_APKT_OFFSET));
}
Sync promise_mio_command with WIP on the Promise line of SATA controllers.
2004-09-03 08:23:04 +00:00
}
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
}
static int
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
ata_promise_sx4_command(struct ata_request *request)
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
{
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
device_t gparent = GRANDPARENT(request->dev);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
struct ata_pci_controller *ctlr = device_get_softc(gparent);
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
struct ata_channel *ch = device_get_softc(device_get_parent(request->dev));
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
struct ata_dma_prdentry *prd = ch->dma->sg;
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
caddr_t window = rman_get_virtual(ctlr->r_res1);
u_int32_t *wordp;
int i, idx, length = 0;
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
/* XXX SOS add ATAPI commands support later */
switch (request->u.ata.command) {
Fix Promise sx4/sx4000 support that broke on the latest race fixes.
2004-08-16 09:29:46 +00:00
default:
return -1;
case ATA_ATA_IDENTIFY:
case ATA_READ:
Fix problem on Promise "mio" controllers and 48bit mode caused by last commit
2005-08-22 11:38:53 +00:00
case ATA_READ48:
Fix Promise sx4/sx4000 support that broke on the latest race fixes.
2004-08-16 09:29:46 +00:00
case ATA_READ_MUL:
Fix problem on Promise "mio" controllers and 48bit mode caused by last commit
2005-08-22 11:38:53 +00:00
case ATA_READ_MUL48:
Fix Promise sx4/sx4000 support that broke on the latest race fixes.
2004-08-16 09:29:46 +00:00
case ATA_WRITE:
Fix problem on Promise "mio" controllers and 48bit mode caused by last commit
2005-08-22 11:38:53 +00:00
case ATA_WRITE48:
Fix Promise sx4/sx4000 support that broke on the latest race fixes.
2004-08-16 09:29:46 +00:00
case ATA_WRITE_MUL:
Fix problem on Promise "mio" controllers and 48bit mode caused by last commit
2005-08-22 11:38:53 +00:00
case ATA_WRITE_MUL48:
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
ATA_OUTL(ctlr->r_res2, 0x000c0400 + ((ch->unit + 1) << 2), 0x00000001);
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
return ata_generic_command(request);
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
Fix Promise sx4/sx4000 support that broke on the latest race fixes.
2004-08-16 09:29:46 +00:00
case ATA_SETFEATURES:
case ATA_FLUSHCACHE:
Fix problem on Promise "mio" controllers and 48bit mode caused by last commit
2005-08-22 11:38:53 +00:00
case ATA_FLUSHCACHE48:
Fix Promise sx4/sx4000 support that broke on the latest race fixes.
2004-08-16 09:29:46 +00:00
case ATA_SLEEP:
case ATA_SET_MULTI:
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
wordp = (u_int32_t *)
(window + (ch->unit * ATA_PDC_CHN_OFFSET) + ATA_PDC_APKT_OFFSET);
wordp[0] = htole32(0x08 | ((ch->unit + 1)<<16) | (0x00 << 24));
wordp[1] = 0;
wordp[2] = 0;
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
ata_promise_apkt((u_int8_t *)wordp, request);
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
ATA_OUTL(ctlr->r_res2, 0x000c0484, 0x00000001);
ATA_OUTL(ctlr->r_res2, 0x000c0400 + ((ch->unit + 1) << 2), 0x00000001);
ATA_OUTL(ctlr->r_res2, 0x000c0240 + (ch->unit << 7),
htole32((ch->unit * ATA_PDC_CHN_OFFSET)+ATA_PDC_APKT_OFFSET));
return 0;
Fix Promise sx4/sx4000 support that broke on the latest race fixes.
2004-08-16 09:29:46 +00:00
case ATA_READ_DMA:
Fix problem on Promise "mio" controllers and 48bit mode caused by last commit
2005-08-22 11:38:53 +00:00
case ATA_READ_DMA48:
Fix Promise sx4/sx4000 support that broke on the latest race fixes.
2004-08-16 09:29:46 +00:00
case ATA_WRITE_DMA:
Fix problem on Promise "mio" controllers and 48bit mode caused by last commit
2005-08-22 11:38:53 +00:00
case ATA_WRITE_DMA48:
Fix Promise sx4/sx4000 support that broke on the latest race fixes.
2004-08-16 09:29:46 +00:00
wordp = (u_int32_t *)
(window + (ch->unit * ATA_PDC_CHN_OFFSET) + ATA_PDC_HSG_OFFSET);
i = idx = 0;
do {
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
wordp[idx++] = prd[i].addr;
wordp[idx++] = prd[i].count;
Fix Promise sx4/sx4000 support that broke on the latest race fixes.
2004-08-16 09:29:46 +00:00
length += (prd[i].count & ~ATA_DMA_EOT);
} while (!(prd[i++].count & ATA_DMA_EOT));
wordp = (u_int32_t *)
(window + (ch->unit * ATA_PDC_CHN_OFFSET) + ATA_PDC_LSG_OFFSET);
wordp[0] = htole32((ch->unit * ATA_PDC_BUF_OFFSET) + ATA_PDC_BUF_BASE);
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
wordp[1] = htole32(request->bytecount | ATA_DMA_EOT);
Fix Promise sx4/sx4000 support that broke on the latest race fixes.
2004-08-16 09:29:46 +00:00
wordp = (u_int32_t *)
(window + (ch->unit * ATA_PDC_CHN_OFFSET) + ATA_PDC_ASG_OFFSET);
wordp[0] = htole32((ch->unit * ATA_PDC_BUF_OFFSET) + ATA_PDC_BUF_BASE);
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
wordp[1] = htole32(request->bytecount | ATA_DMA_EOT);
Fix Promise sx4/sx4000 support that broke on the latest race fixes.
2004-08-16 09:29:46 +00:00
wordp = (u_int32_t *)
(window + (ch->unit * ATA_PDC_CHN_OFFSET) + ATA_PDC_HPKT_OFFSET);
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
if (request->flags & ATA_R_READ)
Fix Promise sx4/sx4000 support that broke on the latest race fixes.
2004-08-16 09:29:46 +00:00
wordp[0] = htole32(0x14 | ((ch->unit+9)<<16) | ((ch->unit+5)<<24));
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
if (request->flags & ATA_R_WRITE)
Fix Promise sx4/sx4000 support that broke on the latest race fixes.
2004-08-16 09:29:46 +00:00
wordp[0] = htole32(0x00 | ((ch->unit+13)<<16) | (0x00<<24));
wordp[1] = htole32((ch->unit * ATA_PDC_CHN_OFFSET)+ATA_PDC_HSG_OFFSET);
wordp[2] = htole32((ch->unit * ATA_PDC_CHN_OFFSET)+ATA_PDC_LSG_OFFSET);
wordp[3] = 0;
wordp = (u_int32_t *)
(window + (ch->unit * ATA_PDC_CHN_OFFSET) + ATA_PDC_APKT_OFFSET);
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
if (request->flags & ATA_R_READ)
Fix Promise sx4/sx4000 support that broke on the latest race fixes.
2004-08-16 09:29:46 +00:00
wordp[0] = htole32(0x04 | ((ch->unit+5)<<16) | (0x00<<24));
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
if (request->flags & ATA_R_WRITE)
Fix Promise sx4/sx4000 support that broke on the latest race fixes.
2004-08-16 09:29:46 +00:00
wordp[0] = htole32(0x10 | ((ch->unit+1)<<16) | ((ch->unit+13)<<24));
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
wordp[1] = htole32((ch->unit * ATA_PDC_CHN_OFFSET)+ATA_PDC_ASG_OFFSET);
Fix Promise sx4/sx4000 support that broke on the latest race fixes.
2004-08-16 09:29:46 +00:00
wordp[2] = 0;
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
ata_promise_apkt((u_int8_t *)wordp, request);
Fix Promise sx4/sx4000 support that broke on the latest race fixes.
2004-08-16 09:29:46 +00:00
ATA_OUTL(ctlr->r_res2, 0x000c0484, 0x00000001);
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
if (request->flags & ATA_R_READ) {
Fix Promise sx4/sx4000 support that broke on the latest race fixes.
2004-08-16 09:29:46 +00:00
ATA_OUTL(ctlr->r_res2, 0x000c0400 + ((ch->unit+5)<<2), 0x00000001);
ATA_OUTL(ctlr->r_res2, 0x000c0400 + ((ch->unit+9)<<2), 0x00000001);
ATA_OUTL(ctlr->r_res2, 0x000c0240 + (ch->unit << 7),
htole32((ch->unit * ATA_PDC_CHN_OFFSET) + ATA_PDC_APKT_OFFSET));
}
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
if (request->flags & ATA_R_WRITE) {
Fix Promise sx4/sx4000 support that broke on the latest race fixes.
2004-08-16 09:29:46 +00:00
ATA_OUTL(ctlr->r_res2, 0x000c0400 + ((ch->unit+1)<<2), 0x00000001);
ATA_OUTL(ctlr->r_res2, 0x000c0400 + ((ch->unit+13)<<2), 0x00000001);
ata_promise_queue_hpkt(ctlr,
htole32((ch->unit * ATA_PDC_CHN_OFFSET) + ATA_PDC_HPKT_OFFSET));
}
return 0;
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
}
}
static int
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
ata_promise_apkt(u_int8_t *bytep, struct ata_request *request)
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
{
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
struct ata_device *atadev = device_get_softc(request->dev);
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
int i = 12;
bytep[i++] = ATA_PDC_1B | ATA_PDC_WRITE_REG | ATA_PDC_WAIT_NBUSY|ATA_DRIVE;
bytep[i++] = ATA_D_IBM | ATA_D_LBA | atadev->unit;
bytep[i++] = ATA_PDC_1B | ATA_PDC_WRITE_CTL;
bytep[i++] = ATA_A_4BIT;
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
if (atadev->flags & ATA_D_48BIT_ACTIVE) {
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
bytep[i++] = ATA_PDC_2B | ATA_PDC_WRITE_REG | ATA_FEATURE;
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
bytep[i++] = request->u.ata.feature >> 8;
bytep[i++] = request->u.ata.feature;
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
bytep[i++] = ATA_PDC_2B | ATA_PDC_WRITE_REG | ATA_COUNT;
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
bytep[i++] = request->u.ata.count >> 8;
bytep[i++] = request->u.ata.count;
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
bytep[i++] = ATA_PDC_2B | ATA_PDC_WRITE_REG | ATA_SECTOR;
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
bytep[i++] = request->u.ata.lba >> 24;
bytep[i++] = request->u.ata.lba;
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
bytep[i++] = ATA_PDC_2B | ATA_PDC_WRITE_REG | ATA_CYL_LSB;
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
bytep[i++] = request->u.ata.lba >> 32;
bytep[i++] = request->u.ata.lba >> 8;
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
bytep[i++] = ATA_PDC_2B | ATA_PDC_WRITE_REG | ATA_CYL_MSB;
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
bytep[i++] = request->u.ata.lba >> 40;
bytep[i++] = request->u.ata.lba >> 16;
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
bytep[i++] = ATA_PDC_1B | ATA_PDC_WRITE_REG | ATA_DRIVE;
bytep[i++] = ATA_D_LBA | atadev->unit;
}
else {
bytep[i++] = ATA_PDC_1B | ATA_PDC_WRITE_REG | ATA_FEATURE;
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
bytep[i++] = request->u.ata.feature;
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
bytep[i++] = ATA_PDC_1B | ATA_PDC_WRITE_REG | ATA_COUNT;
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
bytep[i++] = request->u.ata.count;
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
bytep[i++] = ATA_PDC_1B | ATA_PDC_WRITE_REG | ATA_SECTOR;
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
bytep[i++] = request->u.ata.lba;
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
bytep[i++] = ATA_PDC_1B | ATA_PDC_WRITE_REG | ATA_CYL_LSB;
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
bytep[i++] = request->u.ata.lba >> 8;
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
bytep[i++] = ATA_PDC_1B | ATA_PDC_WRITE_REG | ATA_CYL_MSB;
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
bytep[i++] = request->u.ata.lba >> 16;
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
bytep[i++] = ATA_PDC_1B | ATA_PDC_WRITE_REG | ATA_DRIVE;
bytep[i++] = (atadev->flags & ATA_D_USE_CHS ? 0 : ATA_D_LBA) |
Reshape the dma code to be a bit more flexible so it can cope with new HW that has new and different demands. Fix a few nits in former commit in this cleanup crusade. Sponsored by: pair.com
2005-05-03 07:55:07 +00:00
ATA_D_IBM | atadev->unit | ((request->u.ata.lba >> 24)&0xf);
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
}
Add support for controllers that doesn't have the usual taskfile layout. No functional changes.
2005-04-06 10:22:56 +00:00
bytep[i++] = ATA_PDC_1B | ATA_PDC_WRITE_END | ATA_COMMAND;
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
bytep[i++] = request->u.ata.command;
return i;
}
I'm pleased to announce that Promise is now supporting the FreeBSD project by providing documentation (under NDA) and hardware for testing. This commit is the first result of the cooperation, and adds support for several of their new controllers that we didn't support before (and probably newer would have without this arrangement). Add support for the Promise SATA150 TX2/TX4 and the Promise TX4000 controllers. This also adds support for various motherboard fitted Promise SATA/ATA chips. Note that this code uses memory mapped registers to minimize overhead. I belive FreeBSD has made another first in the Open Source world by being able to release support for this :)
2003-05-01 06:20:50 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static void
ata_promise_queue_hpkt(struct ata_pci_controller *ctlr, u_int32_t hpkt)
{
struct ata_promise_sx4 *hpktp = device_get_ivars(ctlr->dev);
mtx_lock(&hpktp->mtx);
if (hpktp->busy) {
struct host_packet *hp =
malloc(sizeof(struct host_packet), M_TEMP, M_NOWAIT | M_ZERO);
hp->addr = hpkt;
TAILQ_INSERT_TAIL(&hpktp->queue, hp, chain);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
else {
hpktp->busy = 1;
ATA_OUTL(ctlr->r_res2, 0x000c0100, hpkt);
}
mtx_unlock(&hpktp->mtx);
}
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static void
ata_promise_next_hpkt(struct ata_pci_controller *ctlr)
{
struct ata_promise_sx4 *hpktp = device_get_ivars(ctlr->dev);
struct host_packet *hp;
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
mtx_lock(&hpktp->mtx);
if ((hp = TAILQ_FIRST(&hpktp->queue))) {
TAILQ_REMOVE(&hpktp->queue, hp, chain);
ATA_OUTL(ctlr->r_res2, 0x000c0100, hp->addr);
free(hp, M_TEMP);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
else
hpktp->busy = 0;
mtx_unlock(&hpktp->mtx);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
/*
* ServerWorks chipset support functions
*/
int
ata_serverworks_ident(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
struct ata_chip_id *idx;
static struct ata_chip_id ids[] =
Add preliminary support for the Serverworks HT1000 chip. HW sponsored by: Yahoo!
2006-07-04 20:36:03 +00:00
{{ ATA_ROSB4, 0x00, SWKS33, 0x00, ATA_UDMA2, "ROSB4" },
{ ATA_CSB5, 0x92, SWKS100, 0x00, ATA_UDMA5, "CSB5" },
{ ATA_CSB5, 0x00, SWKS66, 0x00, ATA_UDMA4, "CSB5" },
{ ATA_CSB6, 0x00, SWKS100, 0x00, ATA_UDMA5, "CSB6" },
{ ATA_CSB6_1, 0x00, SWKS66, 0x00, ATA_UDMA4, "CSB6" },
{ ATA_HT1000, 0x00, SWKS100, 0x00, ATA_UDMA5, "HT1000" },
{ ATA_HT1000_S1, 0x00, SWKS100, 0x00, ATA_SA150, "HT1000 SATA" },
{ ATA_HT1000_S2, 0x00, SWKSMIO, 0x00, ATA_SA150, "HT1000 SATA mmio" },
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
{ 0, 0, 0, 0, 0, 0}};
char buffer[64];
Support the ICH5 SATA part. Fix HPT374 UDMA133 timing. Fix Promise ID. Cosmetics on probe print for Promise & HPT. Approved by: re
2003-05-18 16:45:48 +00:00
if (!(idx = ata_match_chip(dev, ids)))
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
return ENXIO;
Cleanup the strings for printing chipset names, saves quite some space. Add a couble new nVidia chips now we are here.
2005-12-16 08:12:13 +00:00
sprintf(buffer, "ServerWorks %s %s controller",
idx->text, ata_mode2str(idx->max_dma));
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
device_set_desc_copy(dev, buffer);
ctlr->chip = idx;
ctlr->chipinit = ata_serverworks_chipinit;
return 0;
}
static int
ata_serverworks_chipinit(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
if (ata_setup_interrupt(dev))
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
return ENXIO;
Add preliminary support for the Serverworks HT1000 chip. HW sponsored by: Yahoo!
2006-07-04 20:36:03 +00:00
if (ctlr->chip->cfg1 == SWKSMIO) {
ctlr->r_type2 = SYS_RES_MEMORY;
ctlr->r_rid2 = PCIR_BAR(5);
if (!(ctlr->r_res2 = bus_alloc_resource_any(dev, ctlr->r_type2,
&ctlr->r_rid2, RF_ACTIVE)))
return ENXIO;
ctlr->channels = 4;
ctlr->allocate = ata_serverworks_allocate;
ctlr->setmode = ata_sata_setmode;
return 0;
}
else if (ctlr->chip->cfg1 == SWKS33) {
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
device_t *children;
int nchildren, i;
/* locate the ISA part in the southbridge and enable UDMA33 */
if (!device_get_children(device_get_parent(dev), &children,&nchildren)){
for (i = 0; i < nchildren; i++) {
if (pci_get_devid(children[i]) == ATA_ROSB4_ISA) {
pci_write_config(children[i], 0x64,
(pci_read_config(children[i], 0x64, 4) &
~0x00002000) | 0x00004000, 4);
break;
}
}
free(children, M_TEMP);
}
}
else {
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
pci_write_config(dev, 0x5a,
(pci_read_config(dev, 0x5a, 1) & ~0x40) |
(ctlr->chip->cfg1 == SWKS100) ? 0x03 : 0x02, 1);
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
}
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
ctlr->setmode = ata_serverworks_setmode;
return 0;
}
Add preliminary support for the Serverworks HT1000 chip. HW sponsored by: Yahoo!
2006-07-04 20:36:03 +00:00
static int
ata_serverworks_allocate(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(device_get_parent(dev));
struct ata_channel *ch = device_get_softc(dev);
int ch_offset;
int i;
ch_offset = ch->unit * 0x100;
for (i = ATA_DATA; i < ATA_MAX_RES; i++)
ch->r_io[i].res = ctlr->r_res2;
/* setup ATA registers */
ch->r_io[ATA_DATA].offset = ch_offset + 0x00;
ch->r_io[ATA_FEATURE].offset = ch_offset + 0x04;
ch->r_io[ATA_COUNT].offset = ch_offset + 0x08;
ch->r_io[ATA_SECTOR].offset = ch_offset + 0x0c;
ch->r_io[ATA_CYL_LSB].offset = ch_offset + 0x10;
ch->r_io[ATA_CYL_MSB].offset = ch_offset + 0x14;
ch->r_io[ATA_DRIVE].offset = ch_offset + 0x18;
ch->r_io[ATA_COMMAND].offset = ch_offset + 0x1c;
ch->r_io[ATA_CONTROL].offset = ch_offset + 0x20;
ata_default_registers(dev);
/* setup DMA registers */
ch->r_io[ATA_BMCMD_PORT].offset = ch_offset + 0x30;
ch->r_io[ATA_BMSTAT_PORT].offset = ch_offset + 0x32;
ch->r_io[ATA_BMDTP_PORT].offset = ch_offset + 0x34;
/* setup SATA registers */
ch->r_io[ATA_SSTATUS].offset = ch_offset + 0x40;
ch->r_io[ATA_SERROR].offset = ch_offset + 0x44;
ch->r_io[ATA_SCONTROL].offset = ch_offset + 0x48;
ch->flags |= ATA_NO_SLAVE;
ata_pci_hw(dev);
return 0;
}
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
static void
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_serverworks_setmode(device_t dev, int mode)
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
{
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
device_t gparent = GRANDPARENT(dev);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
struct ata_pci_controller *ctlr = device_get_softc(gparent);
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
struct ata_channel *ch = device_get_softc(device_get_parent(dev));
struct ata_device *atadev = device_get_softc(dev);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
int devno = (ch->unit << 1) + ATA_DEV(atadev->unit);
Fix the serverworks modesetting code, of mask offset was wrong.
2004-10-01 09:04:53 +00:00
int offset = (devno ^ 0x01) << 3;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
int error;
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
u_int8_t piotimings[] = { 0x5d, 0x47, 0x34, 0x22, 0x20, 0x34, 0x22, 0x20,
0x20, 0x20, 0x20, 0x20, 0x20, 0x20 };
u_int8_t dmatimings[] = { 0x77, 0x21, 0x20 };
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
mode = ata_limit_mode(dev, mode, ctlr->chip->max_dma);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
mode = ata_check_80pin(dev, mode);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
error = ata_controlcmd(dev, ATA_SETFEATURES, ATA_SF_SETXFER, 0, mode);
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (bootverbose)
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
device_printf(dev, "%ssetting %s on %s chip\n",
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
(error) ? "FAILURE " : "",
ata_mode2str(mode), ctlr->chip->text);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (!error) {
if (mode >= ATA_UDMA0) {
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, 0x56,
(pci_read_config(gparent, 0x56, 2) &
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
~(0xf << (devno << 2))) |
((mode & ATA_MODE_MASK) << (devno << 2)), 2);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, 0x54,
pci_read_config(gparent, 0x54, 1) |
Fix the serverworks modesetting code, of mask offset was wrong.
2004-10-01 09:04:53 +00:00
(0x01 << devno), 1);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, 0x44,
(pci_read_config(gparent, 0x44, 4) &
Fix the serverworks modesetting code, of mask offset was wrong.
2004-10-01 09:04:53 +00:00
~(0xff << offset)) |
(dmatimings[2] << offset), 4);
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
}
else if (mode >= ATA_WDMA0) {
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, 0x54,
pci_read_config(gparent, 0x54, 1) &
Fix the serverworks modesetting code, of mask offset was wrong.
2004-10-01 09:04:53 +00:00
~(0x01 << devno), 1);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, 0x44,
(pci_read_config(gparent, 0x44, 4) &
Fix the serverworks modesetting code, of mask offset was wrong.
2004-10-01 09:04:53 +00:00
~(0xff << offset)) |
Add support for the ATI IXP[234]00 series chipsets. HW donated by: sentex
2005-10-12 20:00:26 +00:00
(dmatimings[mode & ATA_MODE_MASK] << offset), 4);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
else
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, 0x54,
pci_read_config(gparent, 0x54, 1) &
Fix the serverworks modesetting code, of mask offset was wrong.
2004-10-01 09:04:53 +00:00
~(0x01 << devno), 1);
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, 0x40,
(pci_read_config(gparent, 0x40, 4) &
Fix the serverworks modesetting code, of mask offset was wrong.
2004-10-01 09:04:53 +00:00
~(0xff << offset)) |
(piotimings[ata_mode2idx(mode)] << offset), 4);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
atadev->mode = mode;
}
}
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
/*
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
* Silicon Image Inc. (SiI) (former CMD) chipset support functions
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
*/
int
ata_sii_ident(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
struct ata_chip_id *idx;
static struct ata_chip_id ids[] =
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
{{ ATA_SII3114, 0x00, SIIMEMIO, SII4CH, ATA_SA150, "SiI 3114" },
{ ATA_SII3512, 0x02, SIIMEMIO, 0, ATA_SA150, "SiI 3512" },
{ ATA_SII3112, 0x02, SIIMEMIO, 0, ATA_SA150, "SiI 3112" },
{ ATA_SII3112_1, 0x02, SIIMEMIO, 0, ATA_SA150, "SiI 3112" },
{ ATA_SII3512, 0x00, SIIMEMIO, SIIBUG, ATA_SA150, "SiI 3512" },
{ ATA_SII3112, 0x00, SIIMEMIO, SIIBUG, ATA_SA150, "SiI 3112" },
{ ATA_SII3112_1, 0x00, SIIMEMIO, SIIBUG, ATA_SA150, "SiI 3112" },
Add another ID for the SiI3112a SATA chip as used on the Adaptec 1210SA. RAID support is still in the works, so for now just normal ATA ops. Sponsored by: Matt Douhan(www.fruitsalad.org)
2003-09-08 13:55:05 +00:00
{ ATA_SII0680, 0x00, SIIMEMIO, SIISETCLK, ATA_UDMA6, "SiI 0680" },
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
{ ATA_CMD649, 0x00, 0, SIIINTR, ATA_UDMA5, "CMD 649" },
{ ATA_CMD648, 0x00, 0, SIIINTR, ATA_UDMA4, "CMD 648" },
{ ATA_CMD646, 0x07, 0, 0, ATA_UDMA2, "CMD 646U2" },
{ ATA_CMD646, 0x00, 0, 0, ATA_WDMA2, "CMD 646" },
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
{ 0, 0, 0, 0, 0, 0}};
char buffer[64];
Support the ICH5 SATA part. Fix HPT374 UDMA133 timing. Fix Promise ID. Cosmetics on probe print for Promise & HPT. Approved by: re
2003-05-18 16:45:48 +00:00
if (!(idx = ata_match_chip(dev, ids)))
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
return ENXIO;
sprintf(buffer, "%s %s controller", idx->text, ata_mode2str(idx->max_dma));
device_set_desc_copy(dev, buffer);
ctlr->chip = idx;
ctlr->chipinit = ata_sii_chipinit;
return 0;
}
static int
ata_sii_chipinit(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
if (ata_setup_interrupt(dev))
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
return ENXIO;
Update the SATA support code to work more correctly with real SATA disks now that I can test it. Add support for the SiI 3112 SATA chip using memory mapped I/O. Update the support for the SiI 0680 to use the memio interface as well. Sponsored by: David Leimbach <leimy2k@mac.com> (3112 based controller) Sponsored by: FreeBSD Systems (www.FreeBSDsystems.com) (SATA disks)
2003-07-02 10:50:44 +00:00
if (ctlr->chip->cfg1 == SIIMEMIO) {
Add support for detaching PCI controllers. This adds support for cardbus ATA/SATA controllers. I get roughly the same transfer speeds as on true PCI controllers. Nice to be able to add a couble of "real" disks to a laptop :)
2004-03-15 12:03:48 +00:00
ctlr->r_type2 = SYS_RES_MEMORY;
Try to narrow down the race window on HW that does not have ways to poll for which channel actually pulled the irq line.
2004-08-05 21:13:41 +00:00
ctlr->r_rid2 = PCIR_BAR(5);
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
if (!(ctlr->r_res2 = bus_alloc_resource_any(dev, ctlr->r_type2,
&ctlr->r_rid2, RF_ACTIVE)))
Update the SATA support code to work more correctly with real SATA disks now that I can test it. Add support for the SiI 3112 SATA chip using memory mapped I/O. Update the support for the SiI 0680 to use the memio interface as well. Sponsored by: David Leimbach <leimy2k@mac.com> (3112 based controller) Sponsored by: FreeBSD Systems (www.FreeBSDsystems.com) (SATA disks)
2003-07-02 10:50:44 +00:00
return ENXIO;
if (ctlr->chip->cfg2 & SIISETCLK) {
if ((pci_read_config(dev, 0x8a, 1) & 0x30) != 0x10)
pci_write_config(dev, 0x8a,
Add another ID for the SiI3112a SATA chip as used on the Adaptec 1210SA. RAID support is still in the works, so for now just normal ATA ops. Sponsored by: Matt Douhan(www.fruitsalad.org)
2003-09-08 13:55:05 +00:00
(pci_read_config(dev, 0x8a, 1) & 0xcf)|0x10,1);
Update the SATA support code to work more correctly with real SATA disks now that I can test it. Add support for the SiI 3112 SATA chip using memory mapped I/O. Update the support for the SiI 0680 to use the memio interface as well. Sponsored by: David Leimbach <leimy2k@mac.com> (3112 based controller) Sponsored by: FreeBSD Systems (www.FreeBSDsystems.com) (SATA disks)
2003-07-02 10:50:44 +00:00
if ((pci_read_config(dev, 0x8a, 1) & 0x30) != 0x10)
device_printf(dev, "%s could not set ATA133 clock\n",
ctlr->chip->text);
}
Add another ID for the SiI3112a SATA chip as used on the Adaptec 1210SA. RAID support is still in the works, so for now just normal ATA ops. Sponsored by: Matt Douhan(www.fruitsalad.org)
2003-09-08 13:55:05 +00:00
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
/* if we have 4 channels enable the second set */
Fix Sii3114 support.
2004-05-10 20:23:25 +00:00
if (ctlr->chip->cfg2 & SII4CH) {
ATA_OUTL(ctlr->r_res2, 0x0200, 0x00000002);
Add support for detaching PCI controllers. This adds support for cardbus ATA/SATA controllers. I get roughly the same transfer speeds as on true PCI controllers. Nice to be able to add a couble of "real" disks to a laptop :)
2004-03-15 12:03:48 +00:00
ctlr->channels = 4;
Fix Sii3114 support.
2004-05-10 20:23:25 +00:00
}
Add support for detaching PCI controllers. This adds support for cardbus ATA/SATA controllers. I get roughly the same transfer speeds as on true PCI controllers. Nice to be able to add a couble of "real" disks to a laptop :)
2004-03-15 12:03:48 +00:00
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
/* enable PCI interrupt as BIOS might not */
pci_write_config(dev, 0x8a, (pci_read_config(dev, 0x8a, 1) & 0x3f), 1);
/* dont block interrupts from any channel */
pci_write_config(dev, 0x48,
(pci_read_config(dev, 0x48, 4) & ~0x03c00000), 4);
Fix Sii3114 support.
2004-05-10 20:23:25 +00:00
ctlr->allocate = ata_sii_allocate;
Increase robustness of SATA handling.
2004-06-15 11:02:09 +00:00
if (ctlr->chip->max_dma >= ATA_SA150) {
ctlr->reset = ata_sii_reset;
Add support for the VIA 8237 (both PATA and SATA part). Cleanup the SATA support a bit now we are here anyways.
2003-10-30 13:16:21 +00:00
ctlr->setmode = ata_sata_setmode;
Increase robustness of SATA handling.
2004-06-15 11:02:09 +00:00
}
Add support for the VIA 8237 (both PATA and SATA part). Cleanup the SATA support a bit now we are here anyways.
2003-10-30 13:16:21 +00:00
else
ctlr->setmode = ata_sii_setmode;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
Update the SATA support code to work more correctly with real SATA disks now that I can test it. Add support for the SiI 3112 SATA chip using memory mapped I/O. Update the support for the SiI 0680 to use the memio interface as well. Sponsored by: David Leimbach <leimy2k@mac.com> (3112 based controller) Sponsored by: FreeBSD Systems (www.FreeBSDsystems.com) (SATA disks)
2003-07-02 10:50:44 +00:00
else {
if ((pci_read_config(dev, 0x51, 1) & 0x08) != 0x08) {
device_printf(dev, "HW has secondary channel disabled\n");
ctlr->channels = 1;
}
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
Update the SATA support code to work more correctly with real SATA disks now that I can test it. Add support for the SiI 3112 SATA chip using memory mapped I/O. Update the support for the SiI 0680 to use the memio interface as well. Sponsored by: David Leimbach <leimy2k@mac.com> (3112 based controller) Sponsored by: FreeBSD Systems (www.FreeBSDsystems.com) (SATA disks)
2003-07-02 10:50:44 +00:00
/* enable interrupt as BIOS might not */
Fix support for the old CMD 646 chips, interrupt setup was wrong.
2003-02-25 12:56:37 +00:00
pci_write_config(dev, 0x71, 0x01, 1);
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ctlr->allocate = ata_cmd_allocate;
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
ctlr->setmode = ata_cmd_setmode;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
Update the SATA support code to work more correctly with real SATA disks now that I can test it. Add support for the SiI 3112 SATA chip using memory mapped I/O. Update the support for the SiI 0680 to use the memio interface as well. Sponsored by: David Leimbach <leimy2k@mac.com> (3112 based controller) Sponsored by: FreeBSD Systems (www.FreeBSDsystems.com) (SATA disks)
2003-07-02 10:50:44 +00:00
return 0;
}
Prevent probing the secondary channel on CMD646 controllers if the channel has been disabled by BIOS. This prevents a bus timeout machine check on B&W G3 PowerMacs, which have a primary-only CMD646 on the motherboard. Approved by: sos Obtained from: NetBSD
2003-06-27 10:11:54 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static int
ata_cmd_allocate(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(device_get_parent(dev));
struct ata_channel *ch = device_get_softc(dev);
/* setup the usual register normal pci style */
if (ata_pci_allocate(dev))
return ENXIO;
if (ctlr->chip->cfg2 & SIIINTR)
ch->hw.status = ata_cmd_status;
return 0;
}
static int
ata_cmd_status(device_t dev)
{
struct ata_channel *ch = device_get_softc(dev);
u_int8_t reg71;
if (((reg71 = pci_read_config(device_get_parent(ch->dev), 0x71, 1)) &
(ch->unit ? 0x08 : 0x04))) {
pci_write_config(device_get_parent(ch->dev), 0x71,
reg71 & ~(ch->unit ? 0x04 : 0x08), 1);
return ata_pci_status(dev);
}
return 0;
}
static void
ata_cmd_setmode(device_t dev, int mode)
{
device_t gparent = GRANDPARENT(dev);
struct ata_pci_controller *ctlr = device_get_softc(gparent);
struct ata_channel *ch = device_get_softc(device_get_parent(dev));
struct ata_device *atadev = device_get_softc(dev);
int devno = (ch->unit << 1) + ATA_DEV(atadev->unit);
int error;
mode = ata_limit_mode(dev, mode, ctlr->chip->max_dma);
mode = ata_check_80pin(dev, mode);
error = ata_controlcmd(dev, ATA_SETFEATURES, ATA_SF_SETXFER, 0, mode);
if (bootverbose)
device_printf(dev, "%ssetting %s on %s chip\n",
(error) ? "FAILURE " : "",
ata_mode2str(mode), ctlr->chip->text);
if (!error) {
int treg = 0x54 + ((devno < 3) ? (devno << 1) : 7);
int ureg = ch->unit ? 0x7b : 0x73;
if (mode >= ATA_UDMA0) {
int udmatimings[][2] = { { 0x31, 0xc2 }, { 0x21, 0x82 },
{ 0x11, 0x42 }, { 0x25, 0x8a },
{ 0x15, 0x4a }, { 0x05, 0x0a } };
u_int8_t umode = pci_read_config(gparent, ureg, 1);
umode &= ~(atadev->unit == ATA_MASTER ? 0x35 : 0xca);
umode |= udmatimings[mode & ATA_MODE_MASK][ATA_DEV(atadev->unit)];
pci_write_config(gparent, ureg, umode, 1);
}
else if (mode >= ATA_WDMA0) {
int dmatimings[] = { 0x87, 0x32, 0x3f };
pci_write_config(gparent, treg, dmatimings[mode & ATA_MODE_MASK],1);
pci_write_config(gparent, ureg,
pci_read_config(gparent, ureg, 1) &
~(atadev->unit == ATA_MASTER ? 0x35 : 0xca), 1);
}
else {
int piotimings[] = { 0xa9, 0x57, 0x44, 0x32, 0x3f };
pci_write_config(gparent, treg,
piotimings[(mode & ATA_MODE_MASK) - ATA_PIO0], 1);
pci_write_config(gparent, ureg,
pci_read_config(gparent, ureg, 1) &
~(atadev->unit == ATA_MASTER ? 0x35 : 0xca), 1);
}
atadev->mode = mode;
}
}
Update the SATA support code to work more correctly with real SATA disks now that I can test it. Add support for the SiI 3112 SATA chip using memory mapped I/O. Update the support for the SiI 0680 to use the memio interface as well. Sponsored by: David Leimbach <leimy2k@mac.com> (3112 based controller) Sponsored by: FreeBSD Systems (www.FreeBSDsystems.com) (SATA disks)
2003-07-02 10:50:44 +00:00
static int
Add firstshot support for Acer Labs Inc SATA parts Update ALi PATA support to handle ATA133 as well
2005-04-05 14:51:43 +00:00
ata_sii_allocate(device_t dev)
Update the SATA support code to work more correctly with real SATA disks now that I can test it. Add support for the SiI 3112 SATA chip using memory mapped I/O. Update the support for the SiI 0680 to use the memio interface as well. Sponsored by: David Leimbach <leimy2k@mac.com> (3112 based controller) Sponsored by: FreeBSD Systems (www.FreeBSDsystems.com) (SATA disks)
2003-07-02 10:50:44 +00:00
{
struct ata_pci_controller *ctlr = device_get_softc(device_get_parent(dev));
Add firstshot support for Acer Labs Inc SATA parts Update ALi PATA support to handle ATA133 as well
2005-04-05 14:51:43 +00:00
struct ata_channel *ch = device_get_softc(dev);
Add support for the Silicon Image 3114 4 channel SATA controller.
2004-01-17 23:34:13 +00:00
int unit01 = (ch->unit & 1), unit10 = (ch->unit & 2);
Update the SATA support code to work more correctly with real SATA disks now that I can test it. Add support for the SiI 3112 SATA chip using memory mapped I/O. Update the support for the SiI 0680 to use the memio interface as well. Sponsored by: David Leimbach <leimy2k@mac.com> (3112 based controller) Sponsored by: FreeBSD Systems (www.FreeBSDsystems.com) (SATA disks)
2003-07-02 10:50:44 +00:00
int i;
Prevent probing the secondary channel on CMD646 controllers if the channel has been disabled by BIOS. This prevents a bus timeout machine check on B&W G3 PowerMacs, which have a primary-only CMD646 on the motherboard. Approved by: sos Obtained from: NetBSD
2003-06-27 10:11:54 +00:00
Add support for controllers that doesn't have the usual taskfile layout. No functional changes.
2005-04-06 10:22:56 +00:00
for (i = ATA_DATA; i <= ATA_COMMAND; i++) {
Add support for detaching PCI controllers. This adds support for cardbus ATA/SATA controllers. I get roughly the same transfer speeds as on true PCI controllers. Nice to be able to add a couble of "real" disks to a laptop :)
2004-03-15 12:03:48 +00:00
ch->r_io[i].res = ctlr->r_res2;
Fix Sii3114 support.
2004-05-10 20:23:25 +00:00
ch->r_io[i].offset = 0x80 + i + (unit01 << 6) + (unit10 << 8);
Update the SATA support code to work more correctly with real SATA disks now that I can test it. Add support for the SiI 3112 SATA chip using memory mapped I/O. Update the support for the SiI 0680 to use the memio interface as well. Sponsored by: David Leimbach <leimy2k@mac.com> (3112 based controller) Sponsored by: FreeBSD Systems (www.FreeBSDsystems.com) (SATA disks)
2003-07-02 10:50:44 +00:00
}
Add support for controllers that doesn't have the usual taskfile layout. No functional changes.
2005-04-06 10:22:56 +00:00
ch->r_io[ATA_CONTROL].res = ctlr->r_res2;
ch->r_io[ATA_CONTROL].offset = 0x8a + (unit01 << 6) + (unit10 << 8);
ch->r_io[ATA_IDX_ADDR].res = ctlr->r_res2;
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_default_registers(dev);
Add preliminary support for the Serverworks HT1000 chip. HW sponsored by: Yahoo!
2006-07-04 20:36:03 +00:00
Add support for detaching PCI controllers. This adds support for cardbus ATA/SATA controllers. I get roughly the same transfer speeds as on true PCI controllers. Nice to be able to add a couble of "real" disks to a laptop :)
2004-03-15 12:03:48 +00:00
ch->r_io[ATA_BMCMD_PORT].res = ctlr->r_res2;
Fix Sii3114 support.
2004-05-10 20:23:25 +00:00
ch->r_io[ATA_BMCMD_PORT].offset = 0x00 + (unit01 << 3) + (unit10 << 8);
Add support for detaching PCI controllers. This adds support for cardbus ATA/SATA controllers. I get roughly the same transfer speeds as on true PCI controllers. Nice to be able to add a couble of "real" disks to a laptop :)
2004-03-15 12:03:48 +00:00
ch->r_io[ATA_BMSTAT_PORT].res = ctlr->r_res2;
Fix Sii3114 support.
2004-05-10 20:23:25 +00:00
ch->r_io[ATA_BMSTAT_PORT].offset = 0x02 + (unit01 << 3) + (unit10 << 8);
Add support for detaching PCI controllers. This adds support for cardbus ATA/SATA controllers. I get roughly the same transfer speeds as on true PCI controllers. Nice to be able to add a couble of "real" disks to a laptop :)
2004-03-15 12:03:48 +00:00
ch->r_io[ATA_BMDTP_PORT].res = ctlr->r_res2;
Fix Sii3114 support.
2004-05-10 20:23:25 +00:00
ch->r_io[ATA_BMDTP_PORT].offset = 0x04 + (unit01 << 3) + (unit10 << 8);
Add support for detaching PCI controllers. This adds support for cardbus ATA/SATA controllers. I get roughly the same transfer speeds as on true PCI controllers. Nice to be able to add a couble of "real" disks to a laptop :)
2004-03-15 12:03:48 +00:00
ch->r_io[ATA_BMDEVSPEC_0].res = ctlr->r_res2;
Fix Sii3114 support.
2004-05-10 20:23:25 +00:00
ch->r_io[ATA_BMDEVSPEC_0].offset = 0xa1 + (unit01 << 6) + (unit10 << 8);
Update the SATA support code to work more correctly with real SATA disks now that I can test it. Add support for the SiI 3112 SATA chip using memory mapped I/O. Update the support for the SiI 0680 to use the memio interface as well. Sponsored by: David Leimbach <leimy2k@mac.com> (3112 based controller) Sponsored by: FreeBSD Systems (www.FreeBSDsystems.com) (SATA disks)
2003-07-02 10:50:44 +00:00
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
if (ctlr->chip->max_dma >= ATA_SA150) {
Generalise the SATA PHY handling code so it wont be duplicated for each SATA chip. Promise and Silicon Image are the current candidates for this.
2005-04-08 09:37:47 +00:00
ch->r_io[ATA_SSTATUS].res = ctlr->r_res2;
ch->r_io[ATA_SSTATUS].offset = 0x104 + (unit01 << 7) + (unit10 << 8);
ch->r_io[ATA_SERROR].res = ctlr->r_res2;
ch->r_io[ATA_SERROR].offset = 0x108 + (unit01 << 7) + (unit10 << 8);
ch->r_io[ATA_SCONTROL].res = ctlr->r_res2;
ch->r_io[ATA_SCONTROL].offset = 0x100 + (unit01 << 7) + (unit10 << 8);
Update the SATA support code to work more correctly with real SATA disks now that I can test it. Add support for the SiI 3112 SATA chip using memory mapped I/O. Update the support for the SiI 0680 to use the memio interface as well. Sponsored by: David Leimbach <leimy2k@mac.com> (3112 based controller) Sponsored by: FreeBSD Systems (www.FreeBSDsystems.com) (SATA disks)
2003-07-02 10:50:44 +00:00
ch->flags |= ATA_NO_SLAVE;
Overhaul of the timeout/reinit framework. This should clear up most of the leftovers from the old version that really doesn't work anymore. Add a reset function for host-end of the ATA channel. This is needed for the SiI3112 in order to whack it back to reality if a device locks up the SATA interface (thereby preventing that we can reset the device). The result is that ATA now recovers from the timeouts that happens with the SiI3112A and more or less all disks based on old PATA electronics with a Marvell PATA->SATA converter. This includes lots of the popular SATA dongles and the WDC Raptor disks..
2004-01-11 22:08:34 +00:00
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
/* enable PHY state change interrupt */
ATA_OUTL(ctlr->r_res2, 0x148 + (unit01 << 7) + (unit10 << 8),(1 << 16));
}
Reduce the amount of memory reported to busdma. This made the requirements for bouncebuffers too big with PAE. Cleanup the way size defines for transfers are implemented.
2004-09-10 10:31:37 +00:00
if ((ctlr->chip->cfg2 & SIIBUG) && ch->dma) {
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
/* work around errata in early chips */
Reduce the amount of memory reported to busdma. This made the requirements for bouncebuffers too big with PAE. Cleanup the way size defines for transfers are implemented.
2004-09-10 10:31:37 +00:00
ch->dma->boundary = 16 * DEV_BSIZE;
Add support for setting the SG list segment size. Use this for the SiI3112 workaround to get rid of the "oversized DMA" errors. MFC to 6.0 candidate.
2005-10-06 15:44:07 +00:00
ch->dma->segsize = 15 * DEV_BSIZE;
Reduce the amount of memory reported to busdma. This made the requirements for bouncebuffers too big with PAE. Cleanup the way size defines for transfers are implemented.
2004-09-10 10:31:37 +00:00
}
Add support for the Promise command sequencer present on all modern Promise controllers (PDC203** PDC206**). This also adds preliminary support for the Promise SX4/SX4000 but *only* as a "normal" Promise ATA controller (ATA RAID's are supported though but only RAID0, RAID1 and RAID0+1). This cuts off yet another 5-8% of the command overhead on promise controllers, making them the fastest we have ever had support for. Work is now continuing to add support for this in ATA RAID, to accellerate ATA RAID quite a bit on these controllers, and especially the SX4/SX4000 series as they have quite a few tricks in there.. This commit also adds a few fixes to the SATA code needed for proper support.
2004-04-13 09:44:20 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ata_pci_hw(dev);
ch->hw.status = ata_sii_status;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
return 0;
}
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static int
ata_sii_status(device_t dev)
Update the SATA support code to work more correctly with real SATA disks now that I can test it. Add support for the SiI 3112 SATA chip using memory mapped I/O. Update the support for the SiI 0680 to use the memio interface as well. Sponsored by: David Leimbach <leimy2k@mac.com> (3112 based controller) Sponsored by: FreeBSD Systems (www.FreeBSDsystems.com) (SATA disks)
2003-07-02 10:50:44 +00:00
{
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
struct ata_pci_controller *ctlr = device_get_softc(device_get_parent(dev));
struct ata_channel *ch = device_get_softc(dev);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
/* check for PHY related interrupts on SATA capable HW */
if (ctlr->chip->max_dma >= ATA_SA150) {
u_int32_t status = ATA_IDX_INL(ch, ATA_SSTATUS);
u_int32_t error = ATA_IDX_INL(ch, ATA_SERROR);
struct ata_connect_task *tp;
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
if (error) {
/* clear error bits/interrupt */
ATA_IDX_OUTL(ch, ATA_SERROR, error);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
/* if we have a connection event deal with it */
if ((error & ATA_SE_PHY_CHANGED) &&
(tp = (struct ata_connect_task *)
malloc(sizeof(struct ata_connect_task),
M_ATA, M_NOWAIT | M_ZERO))) {
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
if ((status & ATA_SS_CONWELL_MASK) == ATA_SS_CONWELL_GEN1) {
if (bootverbose)
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
device_printf(ch->dev, "CONNECT requested\n");
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
tp->action = ATA_C_ATTACH;
}
else {
if (bootverbose)
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
device_printf(ch->dev, "DISCONNECT requested\n");
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
tp->action = ATA_C_DETACH;
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
}
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
tp->dev = ch->dev;
TASK_INIT(&tp->task, 0, ata_sata_phy_event, tp);
taskqueue_enqueue(taskqueue_thread, &tp->task);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
}
}
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
Fix support for the cmd646 chip. Spotted by: tmm reviewed by: re@
2003-11-21 22:58:56 +00:00
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
/* any drive action to take care of ? */
if (ATA_IDX_INB(ch, ATA_BMDEVSPEC_0) & 0x08)
return ata_pci_status(dev);
else
return 0;
Fix support for the cmd646 chip. Spotted by: tmm reviewed by: re@
2003-11-21 22:58:56 +00:00
}
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
static void
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_sii_reset(device_t dev)
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
{
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
struct ata_pci_controller *ctlr = device_get_softc(device_get_parent(dev));
struct ata_channel *ch = device_get_softc(dev);
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
int offset = ((ch->unit & 1) << 7) + ((ch->unit & 2) << 8);
/* disable PHY state change interrupt */
ATA_OUTL(ctlr->r_res2, 0x148 + offset, ~(1 << 16));
Add support for setting the SG list segment size. Use this for the SiI3112 workaround to get rid of the "oversized DMA" errors. MFC to 6.0 candidate.
2005-10-06 15:44:07 +00:00
/* reset controller part for this channel */
ATA_OUTL(ctlr->r_res2, 0x48,
ATA_INL(ctlr->r_res2, 0x48) | (0xc0 >> ch->unit));
DELAY(1000);
ATA_OUTL(ctlr->r_res2, 0x48,
ATA_INL(ctlr->r_res2, 0x48) & ~(0xc0 >> ch->unit));
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
ata_sata_phy_enable(ch);
/* enable PHY state change interrupt */
ATA_OUTL(ctlr->r_res2, 0x148 + offset, (1 << 16));
}
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
static void
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_sii_setmode(device_t dev, int mode)
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
{
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
device_t gparent = GRANDPARENT(dev);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
struct ata_pci_controller *ctlr = device_get_softc(gparent);
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
struct ata_channel *ch = device_get_softc(device_get_parent(dev));
struct ata_device *atadev = device_get_softc(dev);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
int rego = (ch->unit << 4) + (ATA_DEV(atadev->unit) << 1);
int mreg = ch->unit ? 0x84 : 0x80;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
int mask = 0x03 << (ATA_DEV(atadev->unit) << 2);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
int mval = pci_read_config(gparent, mreg, 1) & ~mask;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
int error;
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
mode = ata_limit_mode(dev, mode, ctlr->chip->max_dma);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
Fix potential cable detection problem on older CMD chips.
2004-01-14 14:24:36 +00:00
if (ctlr->chip->cfg2 & SIISETCLK) {
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
if (mode > ATA_UDMA2 && (pci_read_config(gparent, 0x79, 1) &
(ch->unit ? 0x02 : 0x01))) {
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_print_cable(dev, "controller");
Fix potential cable detection problem on older CMD chips.
2004-01-14 14:24:36 +00:00
mode = ATA_UDMA2;
}
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
}
Fix potential cable detection problem on older CMD chips.
2004-01-14 14:24:36 +00:00
else
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
mode = ata_check_80pin(dev, mode);
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
error = ata_controlcmd(dev, ATA_SETFEATURES, ATA_SF_SETXFER, 0, mode);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (bootverbose)
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
device_printf(dev, "%ssetting %s on %s chip\n",
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
(error) ? "FAILURE " : "",
ata_mode2str(mode), ctlr->chip->text);
Update the SATA support code to work more correctly with real SATA disks now that I can test it. Add support for the SiI 3112 SATA chip using memory mapped I/O. Update the support for the SiI 0680 to use the memio interface as well. Sponsored by: David Leimbach <leimy2k@mac.com> (3112 based controller) Sponsored by: FreeBSD Systems (www.FreeBSDsystems.com) (SATA disks)
2003-07-02 10:50:44 +00:00
if (error)
return;
Add support for the VIA 8237 (both PATA and SATA part). Cleanup the SATA support a bit now we are here anyways.
2003-10-30 13:16:21 +00:00
if (mode >= ATA_UDMA0) {
u_int8_t udmatimings[] = { 0xf, 0xb, 0x7, 0x5, 0x3, 0x2, 0x1 };
u_int8_t ureg = 0xac + rego;
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, mreg,
Add support for the VIA 8237 (both PATA and SATA part). Cleanup the SATA support a bit now we are here anyways.
2003-10-30 13:16:21 +00:00
mval | (0x03 << (ATA_DEV(atadev->unit) << 2)), 1);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, ureg,
(pci_read_config(gparent, ureg, 1) & ~0x3f) |
Add support for the VIA 8237 (both PATA and SATA part). Cleanup the SATA support a bit now we are here anyways.
2003-10-30 13:16:21 +00:00
udmatimings[mode & ATA_MODE_MASK], 1);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
Add support for the VIA 8237 (both PATA and SATA part). Cleanup the SATA support a bit now we are here anyways.
2003-10-30 13:16:21 +00:00
}
else if (mode >= ATA_WDMA0) {
u_int8_t dreg = 0xa8 + rego;
u_int16_t dmatimings[] = { 0x2208, 0x10c2, 0x10c1 };
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, mreg,
Add support for the VIA 8237 (both PATA and SATA part). Cleanup the SATA support a bit now we are here anyways.
2003-10-30 13:16:21 +00:00
mval | (0x02 << (ATA_DEV(atadev->unit) << 2)), 1);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, dreg, dmatimings[mode & ATA_MODE_MASK], 2);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
Add support for the VIA 8237 (both PATA and SATA part). Cleanup the SATA support a bit now we are here anyways.
2003-10-30 13:16:21 +00:00
}
else {
u_int8_t preg = 0xa4 + rego;
u_int16_t piotimings[] = { 0x328a, 0x2283, 0x1104, 0x10c3, 0x10c1 };
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, mreg,
Add support for the VIA 8237 (both PATA and SATA part). Cleanup the SATA support a bit now we are here anyways.
2003-10-30 13:16:21 +00:00
mval | (0x01 << (ATA_DEV(atadev->unit) << 2)), 1);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, preg, piotimings[mode & ATA_MODE_MASK], 2);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
Update the SATA support code to work more correctly with real SATA disks now that I can test it. Add support for the SiI 3112 SATA chip using memory mapped I/O. Update the support for the SiI 0680 to use the memio interface as well. Sponsored by: David Leimbach <leimy2k@mac.com> (3112 based controller) Sponsored by: FreeBSD Systems (www.FreeBSDsystems.com) (SATA disks)
2003-07-02 10:50:44 +00:00
atadev->mode = mode;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
/*
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
* Silicon Integrated Systems Corp. (SiS) chipset support functions
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
*/
int
ata_sis_ident(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
struct ata_chip_id *idx;
static struct ata_chip_id ids[] =
Cleanup the strings for printing chipset names, saves quite some space. Add a couble new nVidia chips now we are here.
2005-12-16 08:12:13 +00:00
{{ ATA_SIS182, 0x00, SISSATA, 0, ATA_SA150, "182" }, /* south */
{ ATA_SIS181, 0x00, SISSATA, 0, ATA_SA150, "181" }, /* south */
{ ATA_SIS180, 0x00, SISSATA, 0, ATA_SA150, "180" }, /* south */
{ ATA_SIS965, 0x00, SIS133NEW, 0, ATA_UDMA6, "965" }, /* south */
{ ATA_SIS964, 0x00, SIS133NEW, 0, ATA_UDMA6, "964" }, /* south */
{ ATA_SIS963, 0x00, SIS133NEW, 0, ATA_UDMA6, "963" }, /* south */
{ ATA_SIS962, 0x00, SIS133NEW, 0, ATA_UDMA6, "962" }, /* south */
{ ATA_SIS745, 0x00, SIS100NEW, 0, ATA_UDMA5, "745" }, /* 1chip */
{ ATA_SIS735, 0x00, SIS100NEW, 0, ATA_UDMA5, "735" }, /* 1chip */
{ ATA_SIS733, 0x00, SIS100NEW, 0, ATA_UDMA5, "733" }, /* 1chip */
{ ATA_SIS730, 0x00, SIS100OLD, 0, ATA_UDMA5, "730" }, /* 1chip */
{ ATA_SIS635, 0x00, SIS100NEW, 0, ATA_UDMA5, "635" }, /* 1chip */
{ ATA_SIS633, 0x00, SIS100NEW, 0, ATA_UDMA5, "633" }, /* unknown */
{ ATA_SIS630, 0x30, SIS100OLD, 0, ATA_UDMA5, "630S"}, /* 1chip */
{ ATA_SIS630, 0x00, SIS66, 0, ATA_UDMA4, "630" }, /* 1chip */
{ ATA_SIS620, 0x00, SIS66, 0, ATA_UDMA4, "620" }, /* 1chip */
{ ATA_SIS550, 0x00, SIS66, 0, ATA_UDMA5, "550" },
{ ATA_SIS540, 0x00, SIS66, 0, ATA_UDMA4, "540" },
{ ATA_SIS530, 0x00, SIS66, 0, ATA_UDMA4, "530" },
{ ATA_SIS5513, 0xc2, SIS33, 1, ATA_UDMA2, "5513" },
{ ATA_SIS5513, 0x00, SIS33, 1, ATA_WDMA2, "5513" },
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
{ 0, 0, 0, 0, 0, 0 }};
char buffer[64];
Update the SiS support to distinguish older southbridges better.
2004-03-16 16:23:28 +00:00
int found = 0;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
Add support for the SiS964 ATA/SATA southbridge. This could not have been done without the support from kuriyama. Approved by: re@
2003-11-18 15:27:28 +00:00
if (!(idx = ata_find_chip(dev, ids, -pci_get_slot(dev))))
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
return ENXIO;
Update the SiS support to distinguish older southbridges better.
2004-03-16 16:23:28 +00:00
if (idx->cfg2 && !found) {
u_int8_t reg57 = pci_read_config(dev, 0x57, 1);
pci_write_config(dev, 0x57, (reg57 & 0x7f), 1);
if (pci_read_config(dev, PCIR_DEVVENDOR, 4) == ATA_SIS5518) {
found = 1;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
idx->cfg1 = SIS133NEW;
Try to simplify the SiS probe. If SiS stays with the current schema for new nothbridges we should even find the southbridge and be able to use it (if its one we know that is).
2003-12-25 19:37:11 +00:00
idx->max_dma = ATA_UDMA6;
sprintf(buffer, "SiS 962/963 %s controller",
ata_mode2str(idx->max_dma));
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
Update the SiS support to distinguish older southbridges better.
2004-03-16 16:23:28 +00:00
pci_write_config(dev, 0x57, reg57, 1);
}
if (idx->cfg2 && !found) {
u_int8_t reg4a = pci_read_config(dev, 0x4a, 1);
pci_write_config(dev, 0x4a, (reg4a | 0x10), 1);
if (pci_read_config(dev, PCIR_DEVVENDOR, 4) == ATA_SIS5517) {
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
struct ata_chip_id id[] =
Update the SiS support to distinguish older southbridges better.
2004-03-16 16:23:28 +00:00
{{ ATA_SISSOUTH, 0x10, 0, 0, 0, "" }, { 0, 0, 0, 0, 0, 0 }};
Support the ICH5 SATA part. Fix HPT374 UDMA133 timing. Fix Promise ID. Cosmetics on probe print for Promise & HPT. Approved by: re
2003-05-18 16:45:48 +00:00
Update the SiS support to distinguish older southbridges better.
2004-03-16 16:23:28 +00:00
found = 1;
Try to simplify the SiS probe. If SiS stays with the current schema for new nothbridges we should even find the southbridge and be able to use it (if its one we know that is).
2003-12-25 19:37:11 +00:00
if (ata_find_chip(dev, id, pci_get_slot(dev))) {
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
idx->cfg1 = SIS133OLD;
Try to simplify the SiS probe. If SiS stays with the current schema for new nothbridges we should even find the southbridge and be able to use it (if its one we know that is).
2003-12-25 19:37:11 +00:00
idx->max_dma = ATA_UDMA6;
}
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
else {
idx->cfg1 = SIS100NEW;
Try to simplify the SiS probe. If SiS stays with the current schema for new nothbridges we should even find the southbridge and be able to use it (if its one we know that is).
2003-12-25 19:37:11 +00:00
idx->max_dma = ATA_UDMA5;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
sprintf(buffer, "SiS 961 %s controller",ata_mode2str(idx->max_dma));
}
Update the SiS support to distinguish older southbridges better.
2004-03-16 16:23:28 +00:00
pci_write_config(dev, 0x4a, reg4a, 1);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
Update the SiS support to distinguish older southbridges better.
2004-03-16 16:23:28 +00:00
if (!found)
Cleanup the strings for printing chipset names, saves quite some space. Add a couble new nVidia chips now we are here.
2005-12-16 08:12:13 +00:00
sprintf(buffer,"SiS %s %s controller",
idx->text, ata_mode2str(idx->max_dma));
Update the SiS support to distinguish older southbridges better.
2004-03-16 16:23:28 +00:00
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
device_set_desc_copy(dev, buffer);
ctlr->chip = idx;
ctlr->chipinit = ata_sis_chipinit;
return 0;
}
static int
ata_sis_chipinit(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
if (ata_setup_interrupt(dev))
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
return ENXIO;
switch (ctlr->chip->cfg1) {
Add a few more older SiS ID's.
2003-04-10 12:24:39 +00:00
case SIS33:
break;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
case SIS66:
case SIS100OLD:
Change setup for SiS device to allow PCI native mode.
2004-04-24 15:54:20 +00:00
pci_write_config(dev, 0x52, pci_read_config(dev, 0x52, 1) & ~0x04, 1);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
break;
case SIS100NEW:
case SIS133OLD:
Change setup for SiS device to allow PCI native mode.
2004-04-24 15:54:20 +00:00
pci_write_config(dev, 0x49, pci_read_config(dev, 0x49, 1) & ~0x01, 1);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
break;
case SIS133NEW:
Change setup for SiS device to allow PCI native mode.
2004-04-24 15:54:20 +00:00
pci_write_config(dev, 0x50, pci_read_config(dev, 0x50, 2) | 0x0008, 2);
pci_write_config(dev, 0x52, pci_read_config(dev, 0x52, 2) | 0x0008, 2);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
break;
Add support for the SiS964 ATA/SATA southbridge. This could not have been done without the support from kuriyama. Approved by: re@
2003-11-18 15:27:28 +00:00
case SISSATA:
Only try to allocate and use the SATA resource if they are enabled by the BIOS. It seems some BIOS's doesn't get this right, and that would result in ATA panic'ing.
2005-04-25 07:50:51 +00:00
ctlr->r_type2 = SYS_RES_IOPORT;
ctlr->r_rid2 = PCIR_BAR(5);
if ((ctlr->r_res2 = bus_alloc_resource_any(dev, ctlr->r_type2,
&ctlr->r_rid2, RF_ACTIVE))) {
ctlr->allocate = ata_sis_allocate;
ctlr->reset = ata_sis_reset;
Add support for the JMicron JMB360 SATAII controller. Thanks to JMicron for providing needed info. HW donated by: Ralf Folkerts
2006-01-25 23:07:42 +00:00
/* enable PCI interrupt */
pci_write_config(dev, PCIR_COMMAND,
pci_read_config(dev, PCIR_COMMAND, 2) & ~0x0400,2);
Only try to allocate and use the SATA resource if they are enabled by the BIOS. It seems some BIOS's doesn't get this right, and that would result in ATA panic'ing.
2005-04-25 07:50:51 +00:00
}
Add support for the SiS964 ATA/SATA southbridge. This could not have been done without the support from kuriyama. Approved by: re@
2003-11-18 15:27:28 +00:00
ctlr->setmode = ata_sata_setmode;
return 0;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
default:
return ENXIO;
}
ctlr->setmode = ata_sis_setmode;
return 0;
}
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
static int
ata_sis_allocate(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(device_get_parent(dev));
struct ata_channel *ch = device_get_softc(dev);
Fix SiS SATA support, the SATA registers was off. Add support for SiS metadata. HW donated by: obrien
2005-11-28 23:08:37 +00:00
int offset = ch->unit << ((ctlr->chip->chipid == ATA_SIS182) ? 5 : 6);
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
/* setup the usual register normal pci style */
Properly react to allocation failures. Found by: imp@
2005-10-10 19:12:43 +00:00
if (ata_pci_allocate(dev))
return ENXIO;
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
ch->r_io[ATA_SSTATUS].res = ctlr->r_res2;
Fix SiS SATA support, the SATA registers was off. Add support for SiS metadata. HW donated by: obrien
2005-11-28 23:08:37 +00:00
ch->r_io[ATA_SSTATUS].offset = 0x00 + offset;
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
ch->r_io[ATA_SERROR].res = ctlr->r_res2;
Fix SiS SATA support, the SATA registers was off. Add support for SiS metadata. HW donated by: obrien
2005-11-28 23:08:37 +00:00
ch->r_io[ATA_SERROR].offset = 0x04 + offset;
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
ch->r_io[ATA_SCONTROL].res = ctlr->r_res2;
Fix SiS SATA support, the SATA registers was off. Add support for SiS metadata. HW donated by: obrien
2005-11-28 23:08:37 +00:00
ch->r_io[ATA_SCONTROL].offset = 0x08 + offset;
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
ch->flags |= ATA_NO_SLAVE;
/* XXX SOS PHY hotplug handling missing in SiS chip ?? */
/* XXX SOS unknown how to enable PHY state change interrupt */
return 0;
}
static void
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_sis_reset(device_t dev)
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
{
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
struct ata_channel *ch = device_get_softc(dev);
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
ata_sata_phy_enable(ch);
}
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
static void
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_sis_setmode(device_t dev, int mode)
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
{
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
device_t gparent = GRANDPARENT(dev);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
struct ata_pci_controller *ctlr = device_get_softc(gparent);
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
struct ata_channel *ch = device_get_softc(device_get_parent(dev));
struct ata_device *atadev = device_get_softc(dev);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
int devno = (ch->unit << 1) + ATA_DEV(atadev->unit);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
int error;
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
mode = ata_limit_mode(dev, mode, ctlr->chip->max_dma);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (ctlr->chip->cfg1 == SIS133NEW) {
if (mode > ATA_UDMA2 &&
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_read_config(gparent, ch->unit ? 0x52 : 0x50,2) & 0x8000) {
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_print_cable(dev, "controller");
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
mode = ATA_UDMA2;
}
}
else {
if (mode > ATA_UDMA2 &&
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_read_config(gparent, 0x48, 1)&(ch->unit ? 0x20 : 0x10)) {
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_print_cable(dev, "controller");
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
mode = ATA_UDMA2;
}
}
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
error = ata_controlcmd(dev, ATA_SETFEATURES, ATA_SF_SETXFER, 0, mode);
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (bootverbose)
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
device_printf(dev, "%ssetting %s on %s chip\n",
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
(error) ? "FAILURE " : "",
ata_mode2str(mode), ctlr->chip->text);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (!error) {
switch (ctlr->chip->cfg1) {
case SIS133NEW: {
u_int32_t timings[] =
{ 0x28269008, 0x0c266008, 0x04263008, 0x0c0a3008, 0x05093008,
Dont use r_bmio on Cyrix chips if not set from BIOS.
2003-03-10 08:20:14 +00:00
0x22196008, 0x0c0a3008, 0x05093008, 0x050939fc, 0x050936ac,
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
0x0509347c, 0x0509325c, 0x0509323c, 0x0509322c, 0x0509321c};
u_int32_t reg;
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
reg = (pci_read_config(gparent, 0x57, 1)&0x40?0x70:0x40)+(devno<<2);
pci_write_config(gparent, reg, timings[ata_mode2idx(mode)], 4);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
break;
}
case SIS133OLD: {
u_int16_t timings[] =
{ 0x00cb, 0x0067, 0x0044, 0x0033, 0x0031, 0x0044, 0x0033, 0x0031,
0x8f31, 0x8a31, 0x8731, 0x8531, 0x8331, 0x8231, 0x8131 };
u_int16_t reg = 0x40 + (devno << 1);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, reg, timings[ata_mode2idx(mode)], 2);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
break;
}
case SIS100NEW: {
u_int16_t timings[] =
{ 0x00cb, 0x0067, 0x0044, 0x0033, 0x0031, 0x0044, 0x0033,
0x0031, 0x8b31, 0x8731, 0x8531, 0x8431, 0x8231, 0x8131 };
u_int16_t reg = 0x40 + (devno << 1);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, reg, timings[ata_mode2idx(mode)], 2);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
break;
}
case SIS100OLD:
Add a few more older SiS ID's.
2003-04-10 12:24:39 +00:00
case SIS66:
case SIS33: {
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
u_int16_t timings[] =
{ 0x0c0b, 0x0607, 0x0404, 0x0303, 0x0301, 0x0404, 0x0303,
0x0301, 0xf301, 0xd301, 0xb301, 0xa301, 0x9301, 0x8301 };
u_int16_t reg = 0x40 + (devno << 1);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, reg, timings[ata_mode2idx(mode)], 2);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
break;
}
}
atadev->mode = mode;
}
}
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
/* VIA Technologies Inc. chipset support functions */
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
int
ata_via_ident(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
struct ata_chip_id *idx;
static struct ata_chip_id ids[] =
Cleanup the strings for printing chipset names, saves quite some space. Add a couble new nVidia chips now we are here.
2005-12-16 08:12:13 +00:00
{{ ATA_VIA82C586, 0x02, VIA33, 0x00, ATA_UDMA2, "82C586B" },
{ ATA_VIA82C586, 0x00, VIA33, 0x00, ATA_WDMA2, "82C586" },
{ ATA_VIA82C596, 0x12, VIA66, VIACLK, ATA_UDMA4, "82C596B" },
{ ATA_VIA82C596, 0x00, VIA33, 0x00, ATA_UDMA2, "82C596" },
{ ATA_VIA82C686, 0x40, VIA100, VIABUG, ATA_UDMA5, "82C686B"},
{ ATA_VIA82C686, 0x10, VIA66, VIACLK, ATA_UDMA4, "82C686A" },
{ ATA_VIA82C686, 0x00, VIA33, 0x00, ATA_UDMA2, "82C686" },
{ ATA_VIA8231, 0x00, VIA100, VIABUG, ATA_UDMA5, "8231" },
{ ATA_VIA8233, 0x00, VIA100, 0x00, ATA_UDMA5, "8233" },
{ ATA_VIA8233C, 0x00, VIA100, 0x00, ATA_UDMA5, "8233C" },
{ ATA_VIA8233A, 0x00, VIA133, 0x00, ATA_UDMA6, "8233A" },
{ ATA_VIA8235, 0x00, VIA133, 0x00, ATA_UDMA6, "8235" },
{ ATA_VIA8237, 0x00, VIA133, 0x00, ATA_UDMA6, "8237" },
Add support for VIA 8237A.
2006-09-02 12:55:42 +00:00
{ ATA_VIA8237A, 0x00, VIA133, 0x00, ATA_UDMA6, "8237A" },
Cleanup the strings for printing chipset names, saves quite some space. Add a couble new nVidia chips now we are here.
2005-12-16 08:12:13 +00:00
{ ATA_VIA8251, 0x00, VIA133, 0x00, ATA_UDMA6, "8251" },
Add support for the VIA 8237 (both PATA and SATA part). Cleanup the SATA support a bit now we are here anyways.
2003-10-30 13:16:21 +00:00
{ 0, 0, 0, 0, 0, 0 }};
static struct ata_chip_id new_ids[] =
Cleanup the strings for printing chipset names, saves quite some space. Add a couble new nVidia chips now we are here.
2005-12-16 08:12:13 +00:00
{{ ATA_VIA6410, 0x00, 0, 0x00, ATA_UDMA6, "6410" },
{ ATA_VIA6420, 0x00, 7, 0x00, ATA_SA150, "6420" },
{ ATA_VIA6421, 0x00, 6, VIABAR, ATA_SA150, "6421" },
Add support for VIA 8237A.
2006-09-02 12:55:42 +00:00
{ ATA_VIA8237A, 0x00, 0, 0x00, ATA_SA150, "8237A" },
Update AHCI support to be more generic. Add support for AHCI on the VIA VT8251.
2006-06-28 09:59:09 +00:00
{ ATA_VIA8251, 0x00, 0, VIAAHCI, ATA_SA300, "8251" },
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
{ 0, 0, 0, 0, 0, 0 }};
char buffer[64];
Add support for the VIA 8237 (both PATA and SATA part). Cleanup the SATA support a bit now we are here anyways.
2003-10-30 13:16:21 +00:00
if (pci_get_devid(dev) == ATA_VIA82C571) {
Update the support for the VIA 8237 southbridge.
2004-01-20 16:51:02 +00:00
if (!(idx = ata_find_chip(dev, ids, -99)))
Add support for the VIA 8237 (both PATA and SATA part). Cleanup the SATA support a bit now we are here anyways.
2003-10-30 13:16:21 +00:00
return ENXIO;
}
else {
if (!(idx = ata_match_chip(dev, new_ids)))
return ENXIO;
}
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
Cleanup the strings for printing chipset names, saves quite some space. Add a couble new nVidia chips now we are here.
2005-12-16 08:12:13 +00:00
sprintf(buffer, "VIA %s %s controller",
idx->text, ata_mode2str(idx->max_dma));
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
device_set_desc_copy(dev, buffer);
ctlr->chip = idx;
ctlr->chipinit = ata_via_chipinit;
return 0;
}
static int
ata_via_chipinit(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
if (ata_setup_interrupt(dev))
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
return ENXIO;
Add support for the VIA 8237 (both PATA and SATA part). Cleanup the SATA support a bit now we are here anyways.
2003-10-30 13:16:21 +00:00
if (ctlr->chip->max_dma >= ATA_SA150) {
Update AHCI support to be more generic. Add support for AHCI on the VIA VT8251.
2006-06-28 09:59:09 +00:00
if (ctlr->chip->cfg2 == VIAAHCI) {
ctlr->r_type2 = SYS_RES_MEMORY;
ctlr->r_rid2 = PCIR_BAR(5);
if ((ctlr->r_res2 = bus_alloc_resource_any(dev, ctlr->r_type2,
&ctlr->r_rid2,
RF_ACTIVE))) {
return ata_ahci_chipinit(dev);
}
}
Only try to allocate and use the SATA resource if they are enabled by the BIOS. It seems some BIOS's doesn't get this right, and that would result in ATA panic'ing.
2005-04-25 07:50:51 +00:00
ctlr->r_type2 = SYS_RES_IOPORT;
ctlr->r_rid2 = PCIR_BAR(5);
if ((ctlr->r_res2 = bus_alloc_resource_any(dev, ctlr->r_type2,
&ctlr->r_rid2, RF_ACTIVE))) {
ctlr->allocate = ata_via_allocate;
ctlr->reset = ata_via_reset;
Add support for the JMicron JMB360 SATAII controller. Thanks to JMicron for providing needed info. HW donated by: Ralf Folkerts
2006-01-25 23:07:42 +00:00
/* enable PCI interrupt */
pci_write_config(dev, PCIR_COMMAND,
pci_read_config(dev, PCIR_COMMAND, 2) & ~0x0400,2);
Only try to allocate and use the SATA resource if they are enabled by the BIOS. It seems some BIOS's doesn't get this right, and that would result in ATA panic'ing.
2005-04-25 07:50:51 +00:00
}
Add support for the VIA 8237 (both PATA and SATA part). Cleanup the SATA support a bit now we are here anyways.
2003-10-30 13:16:21 +00:00
ctlr->setmode = ata_sata_setmode;
return 0;
}
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
/* prepare for ATA-66 on the 82C686a and 82C596b */
if (ctlr->chip->cfg2 & VIACLK)
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(dev, 0x50, 0x030b030b, 4);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
/* the southbridge might need the data corruption fix */
if (ctlr->chip->cfg2 & VIABUG)
ata_via_southbridge_fixup(dev);
/* set fifo configuration half'n'half */
pci_write_config(dev, 0x43,
(pci_read_config(dev, 0x43, 1) & 0x90) | 0x2a, 1);
/* set status register read retry */
pci_write_config(dev, 0x44, pci_read_config(dev, 0x44, 1) | 0x08, 1);
/* set DMA read & end-of-sector fifo flush */
pci_write_config(dev, 0x46,
(pci_read_config(dev, 0x46, 1) & 0x0c) | 0xf0, 1);
/* set sector size */
pci_write_config(dev, 0x60, DEV_BSIZE, 2);
pci_write_config(dev, 0x68, DEV_BSIZE, 2);
ctlr->setmode = ata_via_family_setmode;
return 0;
}
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
static int
ata_via_allocate(device_t dev)
{
struct ata_pci_controller *ctlr = device_get_softc(device_get_parent(dev));
struct ata_channel *ch = device_get_softc(dev);
Fix support for the VIA 6421 controller.
2005-08-29 09:01:57 +00:00
/* newer SATA chips has resources in one BAR for each channel */
if (ctlr->chip->cfg2 & VIABAR) {
struct resource *r_io;
int i, rid;
rid = PCIR_BAR(ch->unit);
if (!(r_io = bus_alloc_resource_any(device_get_parent(dev),
Whitespace cleanup.
2006-01-18 13:10:17 +00:00
SYS_RES_IOPORT,
Fix support for the VIA 6421 controller.
2005-08-29 09:01:57 +00:00
&rid, RF_ACTIVE)))
return ENXIO;
for (i = ATA_DATA; i <= ATA_COMMAND; i ++) {
ch->r_io[i].res = r_io;
ch->r_io[i].offset = i;
}
ch->r_io[ATA_CONTROL].res = r_io;
ch->r_io[ATA_CONTROL].offset = 2 + ATA_IOSIZE;
ch->r_io[ATA_IDX_ADDR].res = r_io;
ata_default_registers(dev);
for (i = ATA_BMCMD_PORT; i <= ATA_BMDTP_PORT; i++) {
ch->r_io[i].res = ctlr->r_res1;
ch->r_io[i].offset = i - ATA_BMCMD_PORT;
}
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
ata_pci_hw(dev);
Fix support for the VIA 6421 controller.
2005-08-29 09:01:57 +00:00
}
Properly react to allocation failures. Found by: imp@
2005-10-10 19:12:43 +00:00
else {
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
/* setup the usual register normal pci style */
Whitespace cleanup.
2006-01-18 13:10:17 +00:00
if (ata_pci_allocate(dev))
Properly react to allocation failures. Found by: imp@
2005-10-10 19:12:43 +00:00
return ENXIO;
}
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
ch->r_io[ATA_SSTATUS].res = ctlr->r_res2;
ch->r_io[ATA_SSTATUS].offset = (ch->unit << ctlr->chip->cfg1);
ch->r_io[ATA_SERROR].res = ctlr->r_res2;
ch->r_io[ATA_SERROR].offset = 0x04 + (ch->unit << ctlr->chip->cfg1);
ch->r_io[ATA_SCONTROL].res = ctlr->r_res2;
ch->r_io[ATA_SCONTROL].offset = 0x08 + (ch->unit << ctlr->chip->cfg1);
ch->flags |= ATA_NO_SLAVE;
/* XXX SOS PHY hotplug handling missing in VIA chip ?? */
/* XXX SOS unknown how to enable PHY state change interrupt */
return 0;
}
static void
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_via_reset(device_t dev)
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
{
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
struct ata_channel *ch = device_get_softc(dev);
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
ata_sata_phy_enable(ch);
}
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
static void
ata_via_southbridge_fixup(device_t dev)
{
device_t *children;
int nchildren, i;
if (device_get_children(device_get_parent(dev), &children, &nchildren))
return;
for (i = 0; i < nchildren; i++) {
if (pci_get_devid(children[i]) == ATA_VIA8363 ||
pci_get_devid(children[i]) == ATA_VIA8371 ||
pci_get_devid(children[i]) == ATA_VIA8662 ||
pci_get_devid(children[i]) == ATA_VIA8361) {
u_int8_t reg76 = pci_read_config(children[i], 0x76, 1);
if ((reg76 & 0xf0) != 0xd0) {
device_printf(dev,
"Correcting VIA config for southbridge data corruption bug\n");
pci_write_config(children[i], 0x75, 0x80, 1);
pci_write_config(children[i], 0x76, (reg76 & 0x0f) | 0xd0, 1);
}
break;
}
}
free(children, M_TEMP);
}
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
/* common code for VIA, AMD & nVidia */
static void
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_via_family_setmode(device_t dev, int mode)
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
{
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
device_t gparent = GRANDPARENT(dev);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
struct ata_pci_controller *ctlr = device_get_softc(gparent);
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
struct ata_channel *ch = device_get_softc(device_get_parent(dev));
struct ata_device *atadev = device_get_softc(dev);
Fix the register timings for AMD/VIA/nVidia chipsets.
2003-12-18 17:36:41 +00:00
u_int8_t timings[] = { 0xa8, 0x65, 0x42, 0x22, 0x20, 0x42, 0x22, 0x20,
0x20, 0x20, 0x20, 0x20, 0x20, 0x20 };
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
int modes[][7] = {
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
{ 0xc2, 0xc1, 0xc0, 0x00, 0x00, 0x00, 0x00 }, /* VIA ATA33 */
{ 0xee, 0xec, 0xea, 0xe9, 0xe8, 0x00, 0x00 }, /* VIA ATA66 */
{ 0xf7, 0xf6, 0xf4, 0xf2, 0xf1, 0xf0, 0x00 }, /* VIA ATA100 */
{ 0xf7, 0xf7, 0xf6, 0xf4, 0xf2, 0xf1, 0xf0 }, /* VIA ATA133 */
{ 0xc2, 0xc1, 0xc0, 0xc4, 0xc5, 0xc6, 0xc7 }}; /* AMD/nVIDIA */
int devno = (ch->unit << 1) + ATA_DEV(atadev->unit);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
int reg = 0x53 - devno;
int error;
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
mode = ata_limit_mode(dev, mode, ctlr->chip->max_dma);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (ctlr->chip->cfg2 & AMDCABLE) {
Fix cable detection on AMD chips.
2003-11-01 09:30:15 +00:00
if (mode > ATA_UDMA2 &&
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
!(pci_read_config(gparent, 0x42, 1) & (1 << devno))) {
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_print_cable(dev, "controller");
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
mode = ATA_UDMA2;
}
}
else
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
mode = ata_check_80pin(dev, mode);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (ctlr->chip->cfg2 & NVIDIA)
reg += 0x10;
Dont set "address setup timing" on newer VIA chips. closes PR: 69180
2004-07-17 14:48:46 +00:00
if (ctlr->chip->cfg1 != VIA133)
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, reg - 0x08, timings[ata_mode2idx(mode)], 1);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
error = ata_controlcmd(dev, ATA_SETFEATURES, ATA_SF_SETXFER, 0, mode);
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (bootverbose)
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
device_printf(dev, "%ssetting %s on %s chip\n",
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
(error) ? "FAILURE " : "", ata_mode2str(mode),
ctlr->chip->text);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (!error) {
if (mode >= ATA_UDMA0)
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, reg,
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
modes[ctlr->chip->cfg1][mode & ATA_MODE_MASK], 1);
else
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
pci_write_config(gparent, reg, 0x8b, 1);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
atadev->mode = mode;
}
}
Add support for the PHY on nVidia, SiS and VIA chips This code is done blindfolded as I dont have such HW here, so reports of success/failure are most welcome. More SATA PHY cleanups.
2005-04-08 15:33:04 +00:00
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
/* misc functions */
Add support for using DMA on dump, greatly speeds up the dump process. Add dump support in ataraid.
2006-01-18 09:14:55 +00:00
static struct ata_chip_id *
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
ata_match_chip(device_t dev, struct ata_chip_id *index)
{
while (index->chipid != 0) {
if (pci_get_devid(dev) == index->chipid &&
pci_get_revid(dev) >= index->chiprev)
return index;
index++;
}
return NULL;
}
Support the ICH5 SATA part. Fix HPT374 UDMA133 timing. Fix Promise ID. Cosmetics on probe print for Promise & HPT. Approved by: re
2003-05-18 16:45:48 +00:00
static struct ata_chip_id *
ata_find_chip(device_t dev, struct ata_chip_id *index, int slot)
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
{
device_t *children;
Fix the cable detection and properly find the host bridge on older SiS chips.
2003-03-16 16:26:51 +00:00
int nchildren, i;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
if (device_get_children(device_get_parent(dev), &children, &nchildren))
return 0;
Support the ICH5 SATA part. Fix HPT374 UDMA133 timing. Fix Promise ID. Cosmetics on probe print for Promise & HPT. Approved by: re
2003-05-18 16:45:48 +00:00
while (index->chipid != 0) {
for (i = 0; i < nchildren; i++) {
Add support for the SiS964 ATA/SATA southbridge. This could not have been done without the support from kuriyama. Approved by: re@
2003-11-18 15:27:28 +00:00
if (((slot >= 0 && pci_get_slot(children[i]) == slot) ||
(slot < 0 && pci_get_slot(children[i]) <= -slot)) &&
Unbork the chip locating code. Approved by: re@
2003-05-19 13:42:23 +00:00
pci_get_devid(children[i]) == index->chipid &&
pci_get_revid(children[i]) >= index->chiprev) {
Support the ICH5 SATA part. Fix HPT374 UDMA133 timing. Fix Promise ID. Cosmetics on probe print for Promise & HPT. Approved by: re
2003-05-18 16:45:48 +00:00
free(children, M_TEMP);
return index;
}
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
Unbork the chip locating code. Approved by: re@
2003-05-19 13:42:23 +00:00
index++;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
free(children, M_TEMP);
Support the ICH5 SATA part. Fix HPT374 UDMA133 timing. Fix Promise ID. Cosmetics on probe print for Promise & HPT. Approved by: re
2003-05-18 16:45:48 +00:00
return NULL;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
static int
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
ata_setup_interrupt(device_t dev)
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
{
struct ata_pci_controller *ctlr = device_get_softc(dev);
int rid = ATA_IRQ_RID;
Make the test for ATA PCI legacy addressing mode more robust. Add code (currently ifdef'd out) to allow ATA PCI native addressing. Fix the altio offset for ATA PCI devices.
2004-04-21 20:03:26 +00:00
if (!ata_legacy(dev)) {
Convert callers to the new bus_alloc_resource_any(9) API. Submitted by: Mark Santcroos <marks@ripe.net> Reviewed by: imp, dfr, bde
2004-03-17 17:50:55 +00:00
if (!(ctlr->r_irq = bus_alloc_resource_any(dev, SYS_RES_IRQ, &rid,
RF_SHAREABLE | RF_ACTIVE))) {
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
device_printf(dev, "unable to map interrupt\n");
return ENXIO;
}
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
if ((bus_setup_intr(dev, ctlr->r_irq, ATA_INTR_FLAGS,
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
ata_generic_intr, ctlr, &ctlr->handle))) {
device_printf(dev, "unable to setup interrupt\n");
return ENXIO;
}
}
return 0;
}
Fix the PC98 lockups on boot. The interchannel locking for PC98 needed to be updated to match the rest of the locking in ATA.
2004-10-06 19:46:08 +00:00
struct ata_serialize {
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
struct mtx locked_mtx;
int locked_ch;
int restart_ch;
Fix the PC98 lockups on boot. The interchannel locking for PC98 needed to be updated to match the rest of the locking in ATA.
2004-10-06 19:46:08 +00:00
};
static int
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_serialize(device_t dev, int flags)
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
{
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
struct ata_pci_controller *ctlr = device_get_softc(device_get_parent(dev));
struct ata_channel *ch = device_get_softc(dev);
Fix the PC98 lockups on boot. The interchannel locking for PC98 needed to be updated to match the rest of the locking in ATA.
2004-10-06 19:46:08 +00:00
struct ata_serialize *serial;
static int inited = 0;
int res;
if (!inited) {
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
serial = malloc(sizeof(struct ata_serialize),
Fix the PC98 lockups on boot. The interchannel locking for PC98 needed to be updated to match the rest of the locking in ATA.
2004-10-06 19:46:08 +00:00
M_TEMP, M_NOWAIT | M_ZERO);
mtx_init(&serial->locked_mtx, "ATA serialize lock", NULL, MTX_DEF);
serial->locked_ch = -1;
serial->restart_ch = -1;
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
device_set_ivars(ctlr->dev, serial);
Fix the PC98 lockups on boot. The interchannel locking for PC98 needed to be updated to match the rest of the locking in ATA.
2004-10-06 19:46:08 +00:00
inited = 1;
}
else
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
serial = device_get_ivars(ctlr->dev);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
Fix the PC98 lockups on boot. The interchannel locking for PC98 needed to be updated to match the rest of the locking in ATA.
2004-10-06 19:46:08 +00:00
mtx_lock(&serial->locked_mtx);
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
switch (flags) {
case ATA_LF_LOCK:
Fix the PC98 lockups on boot. The interchannel locking for PC98 needed to be updated to match the rest of the locking in ATA.
2004-10-06 19:46:08 +00:00
if (serial->locked_ch == -1)
serial->locked_ch = ch->unit;
if (serial->locked_ch != ch->unit)
serial->restart_ch = ch->unit;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
break;
case ATA_LF_UNLOCK:
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
if (serial->locked_ch == ch->unit) {
serial->locked_ch = -1;
if (serial->restart_ch != -1) {
if ((ch = ctlr->interrupt[serial->restart_ch].argument)) {
serial->restart_ch = -1;
mtx_unlock(&serial->locked_mtx);
ata_start(ch->dev);
return -1;
}
}
}
Fix the PC98 lockups on boot. The interchannel locking for PC98 needed to be updated to match the rest of the locking in ATA.
2004-10-06 19:46:08 +00:00
break;
case ATA_LF_WHICH:
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
break;
}
Fix the PC98 lockups on boot. The interchannel locking for PC98 needed to be updated to match the rest of the locking in ATA.
2004-10-06 19:46:08 +00:00
res = serial->locked_ch;
mtx_unlock(&serial->locked_mtx);
return res;
First round off updates/fixes to the ATA driver. This moves all chipset specific code to a new file 'ata-chipset.c'. Extensive use of tables and pointers to avoid having the same switch on chipset type in several places, and to allow substituting various functions for different HW arch needs. Added PIO mode setup and all DMA modes. Support for all known SiS chipsets. Thanks to Christoph Kukulies for sponsoring a nice ASUS P4S8X SiS648 based board for this work! Tested on: i386, PC98, alpha and sparc64
2003-02-20 20:02:32 +00:00
}
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
Generalise the SATA PHY handling code so it wont be duplicated for each SATA chip. Promise and Silicon Image are the current candidates for this.
2005-04-08 09:37:47 +00:00
static void
ata_print_cable(device_t dev, u_int8_t *who)
{
device_printf(dev,
"DMA limited to UDMA33, %s found non-ATA66 cable\n", who);
}
static int
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_atapi(device_t dev)
Generalise the SATA PHY handling code so it wont be duplicated for each SATA chip. Promise and Silicon Image are the current candidates for this.
2005-04-08 09:37:47 +00:00
{
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
struct ata_channel *ch = device_get_softc(device_get_parent(dev));
struct ata_device *atadev = device_get_softc(dev);
Generalise the SATA PHY handling code so it wont be duplicated for each SATA chip. Promise and Silicon Image are the current candidates for this.
2005-04-08 09:37:47 +00:00
return ((atadev->unit == ATA_MASTER && ch->devices & ATA_ATAPI_MASTER) ||
(atadev->unit == ATA_SLAVE && ch->devices & ATA_ATAPI_SLAVE));
}
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
static int
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_check_80pin(device_t dev, int mode)
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
{
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
struct ata_device *atadev = device_get_softc(dev);
This is the much rumoured ATA mkIII update that I've been working on. o ATA is now fully newbus'd and split into modules. This means that on a modern system you just load "atapci and ata" to get the base support, and then one or more of the device subdrivers "atadisk atapicd atapifd atapist ataraid". All can be loaded/unloaded anytime, but for obvious reasons you dont want to unload atadisk when you have mounted filesystems. o The device identify part of the probe has been rewritten to fix the problems with odd devices the old had, and to try to remove so of the long delays some HW could provoke. Also probing is done without the need for interrupts, making earlier probing possible. o SATA devices can be hot inserted/removed and devices will be created/ removed in /dev accordingly. NOTE: only supported on controllers that has this feature: Promise and Silicon Image for now. On other controllers the usual atacontrol detach/attach dance is still needed. o Support for "atomic" composite ATA requests used for RAID. o ATA RAID support has been rewritten and and now supports these metadata formats: "Adaptec HostRAID" "Highpoint V2 RocketRAID" "Highpoint V3 RocketRAID" "Intel MatrixRAID" "Integrated Technology Express" "LSILogic V2 MegaRAID" "LSILogic V3 MegaRAID" "Promise FastTrak" "Silicon Image Medley" "FreeBSD PseudoRAID" o Update the ioctl API to match new RAID levels etc. o Update atacontrol to know about the new RAID levels etc NOTE: you need to recompile atacontrol with the new sys/ata.h, make world will take care of that. NOTE2: that rebuild is done differently from the old system as the rebuild is now done piggybacked on read requests to the array, so atacontrol simply starts a background "dd" to rebuild the array. o The reinit code has been worked over to be much more robust. o The timeout code has been overhauled for races. o Support of new chipsets. o Lots of fixes for bugs found while doing the modulerization and reviewing the old code. Missing or changed features from current ATA: o atapi-cd no longer has support for ATAPI changers. Todays its much cheaper and alot faster to copy those CD images to disk and serve them from there. Besides they dont seem to be made anymore, maybe for that exact reason. o ATA RAID can only read metadata from all the above metadata formats, not write all of them (Promise and Highpoint V2 so far). This means that arrays can be picked up from the BIOS, but they cannot be created from FreeBSD. There is more to it than just the missing write metadata support, those formats are not unique to a given controller like Promise and Highpoint formats, instead they exist for several types, and even worse, some controllers can have different formats and its impossible to tell which one. The outcome is that we cannot reliably create the metadata of those formats and be sure the controller BIOS will understand it. However write support is needed to update/fail/rebuild the arrays properly so it sits fairly high on the TODO list. o So far atapicam is not supported with these changes. When/if this will change is up to the maintainer of atapi-cam so go there for questions. HW donated by: Webveveriet AS HW donated by: Frode Nordahl HW donated by: Yahoo! HW donated by: Sentex Patience by: Vife and my boys (and even the cats)
2005-03-30 12:03:40 +00:00
if (mode > ATA_UDMA2 && !(atadev->param.hwres & ATA_CABLE_ID)) {
Take newbusification one step further, ie use the device_t more consequently all way through the code down the layers, instead of the mix'n'match that resulted from the conversion done earlier. Sponsored by: pair.com
2005-04-30 16:22:07 +00:00
ata_print_cable(dev, "device");
This is a major rework of the ATA driver (ATAng) Restructure the way ATA/ATAPI commands are processed, use a common ata_request structure for both. This centralises the way requests are handled so locking is much easier to handle. The driver is now layered much more cleanly to seperate the lowlevel HW access so it can be tailored to specific controllers without touching the upper layers. This is needed to support some of the newer semi-intelligent ATA controllers showing up. The top level drivers (disk, ATAPI devices) are more or less still the same with just corrections to use the new interface. Pull ATA out from under Gaint now that locking can be done in a sane way. Add support for a the National Geode SC1100. Thanks to Soekris engineering for sponsoring a Soekris 4801 to make this support. Fixed alot of small bugs in the chipset code for various chips now we are around in that corner anyways.
2003-08-24 09:22:26 +00:00
mode = ATA_UDMA2;
}
return mode;
}
static int
ata_mode2idx(int mode)
{
if ((mode & ATA_DMA_MASK) == ATA_UDMA0)
return (mode & ATA_MODE_MASK) + 8;
if ((mode & ATA_DMA_MASK) == ATA_WDMA0)
return (mode & ATA_MODE_MASK) + 5;
return (mode & ATA_MODE_MASK) - ATA_PIO0;
}
Reference in New Issue Copy Permalink