d/freebsd-skq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
6f01a2296f
freebsd-skq/sys/dev/usb/ohci.c

3258 lines
81 KiB
C
Raw Normal View History

MFNetBSD: ohci.c (1.99), uhci.c (1.133), usb.c (1.49), usb_port.h (1.39), usbdi.c (1.79), usbdi.h (1.47), usbdivar.h (1.62) date: 2001/01/21 02:39:52; author: augustss; Add code to use soft interrupt to handle USB interrupt processing. Don't enable the code since it doesn't work with the kludgy Ethernet drivers.
2002-04-01 21:34:01 +00:00
/* $NetBSD: ohci.c,v 1.99 2001/01/21 02:39:52 augustss Exp $ */
1) Bug fix: Interrupt transfers worked only for the first packet. The second packet was not seen as finished. 2) Textual changes
1999-01-13 00:50:02 +00:00
/* $FreeBSD$ */
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
/*
* Copyright (c) 1998 The NetBSD Foundation, Inc.
* All rights reserved.
*
* This code is derived from software contributed to The NetBSD Foundation
Change Lennart's e-mail address.
2000-05-14 16:43:10 +00:00
* by Lennart Augustsson (lennart@augustsson.net) at
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
* Carlstedt Research & Technology.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by the NetBSD
* Foundation, Inc. and its contributors.
* 4. Neither the name of The NetBSD Foundation nor the names of its
* contributors may be used to endorse or promote products derived
* from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
* ``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 FOUNDATION OR CONTRIBUTORS
* 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.
*/
/*
* USB Open Host Controller driver.
*
MFNetBSD: Update many URLs.
2002-04-01 21:13:22 +00:00
* OHCI spec: http://www.compaq.com/productinfo/development/openhci.html
* USB spec: http://www.usb.org/developers/data/usbspec.zip
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
*/
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/malloc.h>
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
#if defined(__NetBSD__) || defined(__OpenBSD__)
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
#include <sys/kernel.h>
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
#include <sys/device.h>
Remove superfluous header file includes. Pushed-By: phk
1999-11-08 21:06:21 +00:00
#include <sys/select.h>
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
#elif defined(__FreeBSD__)
#include <sys/module.h>
#include <sys/bus.h>
Bring the 'new-bus' to the i386. This extensively changes the way the i386 platform boots, it is no longer ISA-centric, and is fully dynamic. Most old drivers compile and run without modification via 'compatability shims' to enable a smoother transition. eisa, isapnp and pccard* are not yet using the new resource manager. Once fully converted, all drivers will be loadable, including PCI and ISA. (Some other changes appear to have snuck in, including a port of Soren's ATA driver to the Alpha. Soren, back this out if you need to.) This is a checkpoint of work-in-progress, but is quite functional. The bulk of the work was done over the last few years by Doug Rabson and Garrett Wollman. Approved by: core
1999-04-16 21:22:55 +00:00
#include <machine/bus_pio.h>
#include <machine/bus_memio.h>
Synchronisation with NetBSD 1999/11/27 - more req[uest]->xfer changes. - get the corresponding NetBSD Id's right ohci.c - move untimeout above print statement - remove usb_delay that panics the system (tsleep in intr context) when ohcidebug > 5. ugen.c - create the devices for endpoints with make_dev. uhub.c - change from using usbdebug to uhubdebug - add more debugging statements
1999-11-28 21:01:06 +00:00
#if defined(DIAGNOSTIC) && defined(__i386__) && defined(__FreeBSD__)
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
#include <machine/cpu.h>
#endif
Bring the 'new-bus' to the i386. This extensively changes the way the i386 platform boots, it is no longer ISA-centric, and is fully dynamic. Most old drivers compile and run without modification via 'compatability shims' to enable a smoother transition. eisa, isapnp and pccard* are not yet using the new resource manager. Once fully converted, all drivers will be loadable, including PCI and ISA. (Some other changes appear to have snuck in, including a port of Soren's ATA driver to the Alpha. Soren, back this out if you need to.) This is a checkpoint of work-in-progress, but is quite functional. The bulk of the work was done over the last few years by Doug Rabson and Garrett Wollman. Approved by: core
1999-04-16 21:22:55 +00:00
#endif
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
#include <sys/queue.h>
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
#include <machine/bus.h>
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
#include <machine/endian.h>
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
#include <dev/usb/usb.h>
#include <dev/usb/usbdi.h>
#include <dev/usb/usbdivar.h>
#include <dev/usb/usb_mem.h>
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
#include <dev/usb/usb_quirks.h>
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
#include <dev/usb/ohcireg.h>
#include <dev/usb/ohcivar.h>
#if defined(__FreeBSD__)
Undo previous commit. <machine/clock.h> is needed for the prototype for DELAY().
2000-10-16 18:50:00 +00:00
#include <machine/clock.h>
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
#define delay(d) DELAY(d)
1) Make debugging more selective. 2) create function usbd_errstr which turns a usbd_status into a sensible error message 3) Change the printf in DPRINTF to logprintf which is a define for log(KERN_DEBUG, x)
1999-04-11 20:50:33 +00:00
#endif
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
#if defined(__OpenBSD__)
struct cfdriver ohci_cd = {
NULL, "ohci", DV_DULL
};
#endif
1) Make debugging more selective. 2) create function usbd_errstr which turns a usbd_status into a sensible error message 3) Change the printf in DPRINTF to logprintf which is a define for log(KERN_DEBUG, x)
1999-04-11 20:50:33 +00:00
#ifdef OHCI_DEBUG
#define DPRINTF(x) if (ohcidebug) logprintf x
#define DPRINTFN(n,x) if (ohcidebug>(n)) logprintf x
Add the TD's to the hash when they are created not when they are used. This makes it possible to handle the done queue even when the end of tail TD has ended up on the done queue as well (the xfer in this case is NULL but the next field is valid, so we simply skip that element. Make sure we mark all the handled transfers as handled. Stop using powerhooks. They are a NetBSD-ism.
2000-03-15 22:27:24 +00:00
int ohcidebug = 1;
MFNetBSD: ohci.c rev 1.92, uhci.c rev 1.122 (part) date: 2000/08/08 19:51:46; author: tv; state: Exp; lines: +24 -13 %b -> bitmask_snprintf() Because this code is shared, add a macro for bitmask_snprintf() that should expand to the equivalent snprintf() on non-NetBSD systems. This is only used in ?HCI_DEBUG cases anyway.
2002-04-01 19:42:51 +00:00
#ifndef __NetBSD__
#define bitmask_snprintf(q,f,b,l) snprintf((b), (l), "%b", (q), (f))
#endif
1) Make debugging more selective. 2) create function usbd_errstr which turns a usbd_status into a sensible error message 3) Change the printf in DPRINTF to logprintf which is a define for log(KERN_DEBUG, x)
1999-04-11 20:50:33 +00:00
#else
#define DPRINTF(x)
#define DPRINTFN(n,x)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
#endif
struct ohci_pipe;
Remove __P prototypes to reduce diffs between the NetBSD and FreeBSD versions.
2000-07-17 18:41:20 +00:00
Static ohci_soft_ed_t *ohci_alloc_sed(ohci_softc_t *);
Static void ohci_free_sed(ohci_softc_t *, ohci_soft_ed_t *);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
Remove __P prototypes to reduce diffs between the NetBSD and FreeBSD versions.
2000-07-17 18:41:20 +00:00
Static ohci_soft_td_t *ohci_alloc_std(ohci_softc_t *);
Static void ohci_free_std(ohci_softc_t *, ohci_soft_td_t *);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
Remove __P prototypes to reduce diffs between the NetBSD and FreeBSD versions.
2000-07-17 18:41:20 +00:00
Static ohci_soft_itd_t *ohci_alloc_sitd(ohci_softc_t *);
Static void ohci_free_sitd(ohci_softc_t *,ohci_soft_itd_t *);
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
#if 0
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
Static void ohci_free_std_chain(ohci_softc_t *, ohci_soft_td_t *,
ohci_soft_td_t *);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
#endif
Remove __P prototypes to reduce diffs between the NetBSD and FreeBSD versions.
2000-07-17 18:41:20 +00:00
Static usbd_status ohci_alloc_std_chain(struct ohci_pipe *,
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_softc_t *, int, int, u_int16_t, usb_dma_t *,
Remove __P prototypes to reduce diffs between the NetBSD and FreeBSD versions.
2000-07-17 18:41:20 +00:00
ohci_soft_td_t *, ohci_soft_td_t **);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
#if defined(__NetBSD__) || defined(__OpenBSD__)
Remove __P prototypes to reduce diffs between the NetBSD and FreeBSD versions.
2000-07-17 18:41:20 +00:00
Static void ohci_shutdown(void *v);
Static void ohci_power(int, void *);
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
#endif
Remove __P prototypes to reduce diffs between the NetBSD and FreeBSD versions.
2000-07-17 18:41:20 +00:00
Static usbd_status ohci_open(usbd_pipe_handle);
Static void ohci_poll(struct usbd_bus *);
MFNetBSD: ohci.c (1.99), uhci.c (1.133), usb.c (1.49), usb_port.h (1.39), usbdi.c (1.79), usbdi.h (1.47), usbdivar.h (1.62) date: 2001/01/21 02:39:52; author: augustss; Add code to use soft interrupt to handle USB interrupt processing. Don't enable the code since it doesn't work with the kludgy Ethernet drivers.
2002-04-01 21:34:01 +00:00
Static void ohci_softintr (void *);
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
Static void ohci_waitintr(ohci_softc_t *, usbd_xfer_handle);
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
Static void ohci_add_done(ohci_softc_t *, ohci_physaddr_t);
Remove __P prototypes to reduce diffs between the NetBSD and FreeBSD versions.
2000-07-17 18:41:20 +00:00
Static void ohci_rhsc(ohci_softc_t *, usbd_xfer_handle);
Static usbd_status ohci_device_request(usbd_xfer_handle xfer);
Static void ohci_add_ed(ohci_soft_ed_t *, ohci_soft_ed_t *);
Static void ohci_rem_ed(ohci_soft_ed_t *, ohci_soft_ed_t *);
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
Static void ohci_hash_add_td(ohci_softc_t *, ohci_soft_td_t *);
Static void ohci_hash_rem_td(ohci_softc_t *, ohci_soft_td_t *);
Static ohci_soft_td_t *ohci_hash_find_td(ohci_softc_t *, ohci_physaddr_t);
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
Static void ohci_hash_add_itd(ohci_softc_t *, ohci_soft_itd_t *);
Static void ohci_hash_rem_itd(ohci_softc_t *, ohci_soft_itd_t *);
Static ohci_soft_itd_t *ohci_hash_find_itd(ohci_softc_t *, ohci_physaddr_t);
Remove __P prototypes to reduce diffs between the NetBSD and FreeBSD versions.
2000-07-17 18:41:20 +00:00
Static usbd_status ohci_setup_isoc(usbd_pipe_handle pipe);
Static void ohci_device_isoc_enter(usbd_xfer_handle);
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
Static usbd_status ohci_allocm(struct usbd_bus *, usb_dma_t *, u_int32_t);
Remove __P prototypes to reduce diffs between the NetBSD and FreeBSD versions.
2000-07-17 18:41:20 +00:00
Static void ohci_freem(struct usbd_bus *, usb_dma_t *);
Static usbd_xfer_handle ohci_allocx(struct usbd_bus *);
Static void ohci_freex(struct usbd_bus *, usbd_xfer_handle);
Static usbd_status ohci_root_ctrl_transfer(usbd_xfer_handle);
Static usbd_status ohci_root_ctrl_start(usbd_xfer_handle);
Static void ohci_root_ctrl_abort(usbd_xfer_handle);
Static void ohci_root_ctrl_close(usbd_pipe_handle);
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
Static void ohci_root_ctrl_done(usbd_xfer_handle);
Remove __P prototypes to reduce diffs between the NetBSD and FreeBSD versions.
2000-07-17 18:41:20 +00:00
Static usbd_status ohci_root_intr_transfer(usbd_xfer_handle);
Static usbd_status ohci_root_intr_start(usbd_xfer_handle);
Static void ohci_root_intr_abort(usbd_xfer_handle);
Static void ohci_root_intr_close(usbd_pipe_handle);
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
Static void ohci_root_intr_done(usbd_xfer_handle);
Remove __P prototypes to reduce diffs between the NetBSD and FreeBSD versions.
2000-07-17 18:41:20 +00:00
Static usbd_status ohci_device_ctrl_transfer(usbd_xfer_handle);
Static usbd_status ohci_device_ctrl_start(usbd_xfer_handle);
Static void ohci_device_ctrl_abort(usbd_xfer_handle);
Static void ohci_device_ctrl_close(usbd_pipe_handle);
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
Static void ohci_device_ctrl_done(usbd_xfer_handle);
Remove __P prototypes to reduce diffs between the NetBSD and FreeBSD versions.
2000-07-17 18:41:20 +00:00
Static usbd_status ohci_device_bulk_transfer(usbd_xfer_handle);
Static usbd_status ohci_device_bulk_start(usbd_xfer_handle);
Static void ohci_device_bulk_abort(usbd_xfer_handle);
Static void ohci_device_bulk_close(usbd_pipe_handle);
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
Static void ohci_device_bulk_done(usbd_xfer_handle);
Remove __P prototypes to reduce diffs between the NetBSD and FreeBSD versions.
2000-07-17 18:41:20 +00:00
Static usbd_status ohci_device_intr_transfer(usbd_xfer_handle);
Static usbd_status ohci_device_intr_start(usbd_xfer_handle);
Static void ohci_device_intr_abort(usbd_xfer_handle);
Static void ohci_device_intr_close(usbd_pipe_handle);
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
Static void ohci_device_intr_done(usbd_xfer_handle);
Remove __P prototypes to reduce diffs between the NetBSD and FreeBSD versions.
2000-07-17 18:41:20 +00:00
Static usbd_status ohci_device_isoc_transfer(usbd_xfer_handle);
Static usbd_status ohci_device_isoc_start(usbd_xfer_handle);
Static void ohci_device_isoc_abort(usbd_xfer_handle);
Static void ohci_device_isoc_close(usbd_pipe_handle);
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
Static void ohci_device_isoc_done(usbd_xfer_handle);
Remove __P prototypes to reduce diffs between the NetBSD and FreeBSD versions.
2000-07-17 18:41:20 +00:00
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
Static usbd_status ohci_device_setintr(ohci_softc_t *sc,
Remove __P prototypes to reduce diffs between the NetBSD and FreeBSD versions.
2000-07-17 18:41:20 +00:00
struct ohci_pipe *pipe, int ival);
Make the string argument a const.
2001-07-05 10:09:22 +00:00
Static int ohci_str(usb_string_descriptor_t *, int, const char *);
Remove __P prototypes to reduce diffs between the NetBSD and FreeBSD versions.
2000-07-17 18:41:20 +00:00
Static void ohci_timeout(void *);
Static void ohci_rhsc_able(ohci_softc_t *, int);
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
Static void ohci_close_pipe(usbd_pipe_handle, ohci_soft_ed_t *);
Static void ohci_abort_xfer(usbd_xfer_handle, usbd_status);
Remove __P prototypes to reduce diffs between the NetBSD and FreeBSD versions.
2000-07-17 18:41:20 +00:00
Static void ohci_abort_xfer_end(void *);
Static void ohci_device_clear_toggle(usbd_pipe_handle pipe);
Static void ohci_noop(usbd_pipe_handle pipe);
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
UHCI_DEBUG should of course be OHCI_DEBUG Reported by: MAEKAWA Masahide <bishop@rr.iij4u.or.jp>
1999-04-17 21:48:03 +00:00
#ifdef OHCI_DEBUG
Remove __P prototypes to reduce diffs between the NetBSD and FreeBSD versions.
2000-07-17 18:41:20 +00:00
Static void ohci_dumpregs(ohci_softc_t *);
Static void ohci_dump_tds(ohci_soft_td_t *);
Static void ohci_dump_td(ohci_soft_td_t *);
Static void ohci_dump_ed(ohci_soft_ed_t *);
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
Static void ohci_dump_itd(ohci_soft_itd_t *);
Static void ohci_dump_itds(ohci_soft_itd_t *);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
#endif
Merge from NetBSD: ohcivar.h (1.22), uhcivar.h (1.29): ============================================================ date: 2000/04/25 09:20:55; author: augustss; Move the size of the mapped bus_space region into the bus independent softc. ============================================================ ohci.c (1.88), uhci.c (1.112): ============================================================ date: 2000/04/25 14:28:13; author: augustss; Insert (very conservative!) bus_space_barrier() calls at all register accesses. The bus_space(9) man page says you've gotta have them... ============================================================
2002-03-16 12:44:21 +00:00
#define OBARR(sc) bus_space_barrier((sc)->iot, (sc)->ioh, 0, (sc)->sc_size, \
BUS_SPACE_BARRIER_READ|BUS_SPACE_BARRIER_WRITE)
#define OWRITE1(sc, r, x) \
do { OBARR(sc); bus_space_write_1((sc)->iot, (sc)->ioh, (r), (x)); } while (0)
#define OWRITE2(sc, r, x) \
do { OBARR(sc); bus_space_write_2((sc)->iot, (sc)->ioh, (r), (x)); } while (0)
#define OWRITE4(sc, r, x) \
do { OBARR(sc); bus_space_write_4((sc)->iot, (sc)->ioh, (r), (x)); } while (0)
#define OREAD1(sc, r) (OBARR(sc), bus_space_read_1((sc)->iot, (sc)->ioh, (r)))
#define OREAD2(sc, r) (OBARR(sc), bus_space_read_2((sc)->iot, (sc)->ioh, (r)))
#define OREAD4(sc, r) (OBARR(sc), bus_space_read_4((sc)->iot, (sc)->ioh, (r)))
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
/* Reverse the bits in a value 0 .. 31 */
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
Static u_int8_t revbits[OHCI_NO_INTRS] =
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{ 0x00, 0x10, 0x08, 0x18, 0x04, 0x14, 0x0c, 0x1c,
0x02, 0x12, 0x0a, 0x1a, 0x06, 0x16, 0x0e, 0x1e,
0x01, 0x11, 0x09, 0x19, 0x05, 0x15, 0x0d, 0x1d,
0x03, 0x13, 0x0b, 0x1b, 0x07, 0x17, 0x0f, 0x1f };
struct ohci_pipe {
struct usbd_pipe pipe;
ohci_soft_ed_t *sed;
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
union {
ohci_soft_td_t *td;
ohci_soft_itd_t *itd;
} tail;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
/* Info needed for different pipe kinds. */
union {
/* Control pipe */
struct {
usb_dma_t reqdma;
u_int length;
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
ohci_soft_td_t *setup, *data, *stat;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
} ctl;
/* Interrupt pipe */
struct {
int nslots;
int pos;
} intr;
/* Bulk pipe */
struct {
u_int length;
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
int isread;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
} bulk;
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
/* Iso pipe */
struct iso {
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
int next, inuse;
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
} iso;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
} u;
};
#define OHCI_INTR_ENDPT 1
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static struct usbd_bus_methods ohci_bus_methods = {
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
ohci_open,
Merge from NetBSD: ohci.c: revision 1.72 and 1.73 ohcivar.h: revision 1.19 and 1.20 uhci.c: revision 1.85 usbdi.h: a small part of revision 1.40 usbdivar.h: revision 1.47 Relevant commit messages from NetBSD are: date: 2000/02/22 11:30:54; author: augustss; state: Exp; Prepare a little for having USB interrupt processing done outside the hard interrupt level (in a thread or a softintr). No real soft processing done yet. ---------------------------- date: 2000/02/01 05:42:53; author: augustss; state: Exp; Put some #ifdefs around power and shutdown hooks.
2002-01-26 12:04:22 +00:00
ohci_softintr,
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
ohci_poll,
ohci_allocm,
ohci_freem,
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
ohci_allocx,
ohci_freex,
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
};
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
Static struct usbd_pipe_methods ohci_root_ctrl_methods = {
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
ohci_root_ctrl_transfer,
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
ohci_root_ctrl_start,
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
ohci_root_ctrl_abort,
ohci_root_ctrl_close,
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
ohci_noop,
Merge from NetBSD: revision 1.65 date: 2000/01/25 12:06:21; author: augustss; state: Exp; lines: +10 -2 Add done method for root control transfers.
2002-01-21 04:15:39 +00:00
ohci_root_ctrl_done,
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
};
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
Static struct usbd_pipe_methods ohci_root_intr_methods = {
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
ohci_root_intr_transfer,
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
ohci_root_intr_start,
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
ohci_root_intr_abort,
ohci_root_intr_close,
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
ohci_noop,
ohci_root_intr_done,
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
};
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
Static struct usbd_pipe_methods ohci_device_ctrl_methods = {
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
ohci_device_ctrl_transfer,
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
ohci_device_ctrl_start,
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
ohci_device_ctrl_abort,
ohci_device_ctrl_close,
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
ohci_noop,
ohci_device_ctrl_done,
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
};
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
Static struct usbd_pipe_methods ohci_device_intr_methods = {
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
ohci_device_intr_transfer,
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
ohci_device_intr_start,
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
ohci_device_intr_abort,
ohci_device_intr_close,
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
ohci_device_clear_toggle,
ohci_device_intr_done,
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
};
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
Static struct usbd_pipe_methods ohci_device_bulk_methods = {
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
ohci_device_bulk_transfer,
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
ohci_device_bulk_start,
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
ohci_device_bulk_abort,
ohci_device_bulk_close,
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
ohci_device_clear_toggle,
ohci_device_bulk_done,
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
};
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static struct usbd_pipe_methods ohci_device_isoc_methods = {
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
ohci_device_isoc_transfer,
ohci_device_isoc_start,
ohci_device_isoc_abort,
ohci_device_isoc_close,
ohci_noop,
ohci_device_isoc_done,
};
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
#if defined(__NetBSD__) || defined(__OpenBSD__)
int
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_activate(device_ptr_t self, enum devact act)
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
{
struct ohci_softc *sc = (struct ohci_softc *)self;
int rv = 0;
switch (act) {
case DVACT_ACTIVATE:
return (EOPNOTSUPP);
break;
case DVACT_DEACTIVATE:
if (sc->sc_child != NULL)
rv = config_deactivate(sc->sc_child);
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
sc->sc_dying = 1;
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
break;
}
return (rv);
}
int
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_detach(struct ohci_softc *sc, int flags)
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
{
int rv = 0;
if (sc->sc_child != NULL)
rv = config_detach(sc->sc_child, flags);
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
if (rv != 0)
return (rv);
Merge from NetBSD: ohci.c: -r1.69 to 1.71 ohcireg.h: -r1.14 Some of these deltas are based upon patches that we submitted back to NetBSD. They got manifested slightly differently though, so I've brought back those differences to bring our code bases closer together. The logs from the NetBSD version of ohci.c: revision 1.71 date: 2000/02/01 05:42:52; author: augustss; state: Exp; lines: +13 -2 Put some #ifdefs around power and shutdown hooks. ---------------------------- revision 1.70 date: 2000/01/31 22:35:13; author: augustss; state: Exp; lines: +7 -7 Rename TAILMASK to HEADMASK, since it really masks the head pointer. From FreeBSD. ---------------------------- revision 1.69 date: 2000/01/31 22:09:13; author: augustss; state: Exp; lines: +18 -14 Change where the has table for physical-to-virtual address translation is handled. Partly from FreeBSD.
2002-01-21 05:02:21 +00:00
#if defined(__NetBSD__) || defined(__OpenBSD__)
Add the TD's to the hash when they are created not when they are used. This makes it possible to handle the done queue even when the end of tail TD has ended up on the done queue as well (the xfer in this case is NULL but the next field is valid, so we simply skip that element. Make sure we mark all the handled transfers as handled. Stop using powerhooks. They are a NetBSD-ism.
2000-03-15 22:27:24 +00:00
powerhook_disestablish(sc->sc_powerhook);
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
shutdownhook_disestablish(sc->sc_shutdownhook);
#endif
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
/* free data structures XXX */
return (rv);
}
#endif
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
ohci_soft_ed_t *
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_alloc_sed(ohci_softc_t *sc)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
ohci_soft_ed_t *sed;
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
usbd_status err;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
int i, offs;
usb_dma_t dma;
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
if (sc->sc_freeeds == NULL) {
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
DPRINTFN(2, ("ohci_alloc_sed: allocating chunk\n"));
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
err = usb_allocmem(&sc->sc_bus, OHCI_SED_SIZE * OHCI_SED_CHUNK,
OHCI_ED_ALIGN, &dma);
if (err)
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
return (NULL);
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
for(i = 0; i < OHCI_SED_CHUNK; i++) {
offs = i * OHCI_SED_SIZE;
Correctly handle the conversion from virtual to physical addresses. The problem was basically (for offset > 4096): vtophys(addr) + offset != vtophys(addr + offset) Also, use TD's with a maximum size of 4k instead of 8kb for OHCI controllers. This problem occurs in drivers that use large transfer sizes: umass, host2host and ethernet with jumbo frames.
2000-02-10 18:50:19 +00:00
sed = (ohci_soft_ed_t *)((char *)KERNADDR(&dma, offs));
sed->physaddr = DMAADDR(&dma, offs);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
sed->next = sc->sc_freeeds;
sc->sc_freeeds = sed;
}
}
sed = sc->sc_freeeds;
sc->sc_freeeds = sed->next;
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
memset(&sed->ed, 0, sizeof(ohci_ed_t));
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
sed->next = 0;
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
return (sed);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_free_sed(ohci_softc_t *sc, ohci_soft_ed_t *sed)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
sed->next = sc->sc_freeeds;
sc->sc_freeeds = sed;
}
ohci_soft_td_t *
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_alloc_std(ohci_softc_t *sc)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
ohci_soft_td_t *std;
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
usbd_status err;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
int i, offs;
usb_dma_t dma;
Add the TD's to the hash when they are created not when they are used. This makes it possible to handle the done queue even when the end of tail TD has ended up on the done queue as well (the xfer in this case is NULL but the next field is valid, so we simply skip that element. Make sure we mark all the handled transfers as handled. Stop using powerhooks. They are a NetBSD-ism.
2000-03-15 22:27:24 +00:00
int s;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
if (sc->sc_freetds == NULL) {
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
DPRINTFN(2, ("ohci_alloc_std: allocating chunk\n"));
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
err = usb_allocmem(&sc->sc_bus, OHCI_STD_SIZE * OHCI_STD_CHUNK,
OHCI_TD_ALIGN, &dma);
if (err)
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
return (0);
Merge from NetBSD: ohci.c: -r1.69 to 1.71 ohcireg.h: -r1.14 Some of these deltas are based upon patches that we submitted back to NetBSD. They got manifested slightly differently though, so I've brought back those differences to bring our code bases closer together. The logs from the NetBSD version of ohci.c: revision 1.71 date: 2000/02/01 05:42:52; author: augustss; state: Exp; lines: +13 -2 Put some #ifdefs around power and shutdown hooks. ---------------------------- revision 1.70 date: 2000/01/31 22:35:13; author: augustss; state: Exp; lines: +7 -7 Rename TAILMASK to HEADMASK, since it really masks the head pointer. From FreeBSD. ---------------------------- revision 1.69 date: 2000/01/31 22:09:13; author: augustss; state: Exp; lines: +18 -14 Change where the has table for physical-to-virtual address translation is handled. Partly from FreeBSD.
2002-01-21 05:02:21 +00:00
s = splusb();
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
for(i = 0; i < OHCI_STD_CHUNK; i++) {
offs = i * OHCI_STD_SIZE;
Correctly handle the conversion from virtual to physical addresses. The problem was basically (for offset > 4096): vtophys(addr) + offset != vtophys(addr + offset) Also, use TD's with a maximum size of 4k instead of 8kb for OHCI controllers. This problem occurs in drivers that use large transfer sizes: umass, host2host and ethernet with jumbo frames.
2000-02-10 18:50:19 +00:00
std = (ohci_soft_td_t *)((char *)KERNADDR(&dma, offs));
std->physaddr = DMAADDR(&dma, offs);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
std->nexttd = sc->sc_freetds;
sc->sc_freetds = std;
}
Merge from NetBSD: ohci.c: -r1.69 to 1.71 ohcireg.h: -r1.14 Some of these deltas are based upon patches that we submitted back to NetBSD. They got manifested slightly differently though, so I've brought back those differences to bring our code bases closer together. The logs from the NetBSD version of ohci.c: revision 1.71 date: 2000/02/01 05:42:52; author: augustss; state: Exp; lines: +13 -2 Put some #ifdefs around power and shutdown hooks. ---------------------------- revision 1.70 date: 2000/01/31 22:35:13; author: augustss; state: Exp; lines: +7 -7 Rename TAILMASK to HEADMASK, since it really masks the head pointer. From FreeBSD. ---------------------------- revision 1.69 date: 2000/01/31 22:09:13; author: augustss; state: Exp; lines: +18 -14 Change where the has table for physical-to-virtual address translation is handled. Partly from FreeBSD.
2002-01-21 05:02:21 +00:00
splx(s);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
Merge from NetBSD: ohci.c: -r1.69 to 1.71 ohcireg.h: -r1.14 Some of these deltas are based upon patches that we submitted back to NetBSD. They got manifested slightly differently though, so I've brought back those differences to bring our code bases closer together. The logs from the NetBSD version of ohci.c: revision 1.71 date: 2000/02/01 05:42:52; author: augustss; state: Exp; lines: +13 -2 Put some #ifdefs around power and shutdown hooks. ---------------------------- revision 1.70 date: 2000/01/31 22:35:13; author: augustss; state: Exp; lines: +7 -7 Rename TAILMASK to HEADMASK, since it really masks the head pointer. From FreeBSD. ---------------------------- revision 1.69 date: 2000/01/31 22:09:13; author: augustss; state: Exp; lines: +18 -14 Change where the has table for physical-to-virtual address translation is handled. Partly from FreeBSD.
2002-01-21 05:02:21 +00:00
s = splusb();
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
std = sc->sc_freetds;
sc->sc_freetds = std->nexttd;
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
memset(&std->td, 0, sizeof(ohci_td_t));
The toggle carry bit is stored in the headp not the tailp.
2000-01-29 14:18:31 +00:00
std->nexttd = NULL;
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
std->xfer = NULL;
Add the TD's to the hash when they are created not when they are used. This makes it possible to handle the done queue even when the end of tail TD has ended up on the done queue as well (the xfer in this case is NULL but the next field is valid, so we simply skip that element. Make sure we mark all the handled transfers as handled. Stop using powerhooks. They are a NetBSD-ism.
2000-03-15 22:27:24 +00:00
ohci_hash_add_td(sc, std);
splx(s);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
return (std);
}
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_free_std(ohci_softc_t *sc, ohci_soft_td_t *std)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
Add the TD's to the hash when they are created not when they are used. This makes it possible to handle the done queue even when the end of tail TD has ended up on the done queue as well (the xfer in this case is NULL but the next field is valid, so we simply skip that element. Make sure we mark all the handled transfers as handled. Stop using powerhooks. They are a NetBSD-ism.
2000-03-15 22:27:24 +00:00
int s;
s = splusb();
ohci_hash_rem_td(sc, std);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
std->nexttd = sc->sc_freetds;
sc->sc_freetds = std;
Merge from NetBSD: ohci.c: -r1.69 to 1.71 ohcireg.h: -r1.14 Some of these deltas are based upon patches that we submitted back to NetBSD. They got manifested slightly differently though, so I've brought back those differences to bring our code bases closer together. The logs from the NetBSD version of ohci.c: revision 1.71 date: 2000/02/01 05:42:52; author: augustss; state: Exp; lines: +13 -2 Put some #ifdefs around power and shutdown hooks. ---------------------------- revision 1.70 date: 2000/01/31 22:35:13; author: augustss; state: Exp; lines: +7 -7 Rename TAILMASK to HEADMASK, since it really masks the head pointer. From FreeBSD. ---------------------------- revision 1.69 date: 2000/01/31 22:09:13; author: augustss; state: Exp; lines: +18 -14 Change where the has table for physical-to-virtual address translation is handled. Partly from FreeBSD.
2002-01-21 05:02:21 +00:00
splx(s);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
usbd_status
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_alloc_std_chain(struct ohci_pipe *opipe, ohci_softc_t *sc,
int len, int rd, u_int16_t flags, usb_dma_t *dma,
ohci_soft_td_t *std, ohci_soft_td_t **rstd)
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
{
ohci_soft_td_t *next, *cur;
ohci_physaddr_t dataphys, dataphysend;
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
u_int32_t intr, tdflags;
Correctly handle the conversion from virtual to physical addresses. The problem was basically (for offset > 4096): vtophys(addr) + offset != vtophys(addr + offset) Also, use TD's with a maximum size of 4k instead of 8kb for OHCI controllers. This problem occurs in drivers that use large transfer sizes: umass, host2host and ethernet with jumbo frames.
2000-02-10 18:50:19 +00:00
int offset = 0;
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
int curlen;
DPRINTFN(len < 4096,("ohci_alloc_std_chain: start len=%d\n", len));
Correctly handle the conversion from virtual to physical addresses. The problem was basically (for offset > 4096): vtophys(addr) + offset != vtophys(addr + offset) Also, use TD's with a maximum size of 4k instead of 8kb for OHCI controllers. This problem occurs in drivers that use large transfer sizes: umass, host2host and ethernet with jumbo frames.
2000-02-10 18:50:19 +00:00
cur = std;
dataphysend = DMAADDR(dma, len - 1);
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
tdflags =
(rd ? OHCI_TD_IN : OHCI_TD_OUT) |
OHCI_TD_NOCC | OHCI_TD_TOGGLE_CARRY |
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
(flags & USBD_SHORT_XFER_OK ? OHCI_TD_R : 0);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
for (;;) {
next = ohci_alloc_std(sc);
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
if (next == 0)
goto nomem;
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
Correctly handle the conversion from virtual to physical addresses. The problem was basically (for offset > 4096): vtophys(addr) + offset != vtophys(addr + offset) Also, use TD's with a maximum size of 4k instead of 8kb for OHCI controllers. This problem occurs in drivers that use large transfer sizes: umass, host2host and ethernet with jumbo frames.
2000-02-10 18:50:19 +00:00
dataphys = DMAADDR(dma, offset);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
/* The OHCI hardware can handle at most one page crossing. */
Correctly handle the conversion from virtual to physical addresses. The problem was basically (for offset > 4096): vtophys(addr) + offset != vtophys(addr + offset) Also, use TD's with a maximum size of 4k instead of 8kb for OHCI controllers. This problem occurs in drivers that use large transfer sizes: umass, host2host and ethernet with jumbo frames.
2000-02-10 18:50:19 +00:00
#if defined(__NetBSD__) || defined(__OpenBSD__)
if (OHCI_PAGE(dataphys) == OHCI_PAGE(dataphysend) ||
OHCI_PAGE(dataphys) + OHCI_PAGE_SIZE == OHCI_PAGE(dataphysend))
#elif defined(__FreeBSD__)
/* XXX This is pretty broken: Because we do not allocate
* a contiguous buffer (contiguous in physical pages) we
* can only transfer one page in one go.
* So check whether the start and end of the buffer are on
* the same page.
*/
if (OHCI_PAGE(dataphys) == OHCI_PAGE(dataphysend))
#endif
{
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
/* we can handle it in this TD */
curlen = len;
} else {
Correctly handle the conversion from virtual to physical addresses. The problem was basically (for offset > 4096): vtophys(addr) + offset != vtophys(addr + offset) Also, use TD's with a maximum size of 4k instead of 8kb for OHCI controllers. This problem occurs in drivers that use large transfer sizes: umass, host2host and ethernet with jumbo frames.
2000-02-10 18:50:19 +00:00
/* XXX The calculation below is wrong and could
* result in a packet that is not a multiple of the
* MaxPacketSize in the case where the buffer does not
* start on an appropriate address (like for example in
* the case of an mbuf cluster). You'll get an early
* short packet.
*/
#if defined(__NetBSD__) || defined(__OpenBSD__)
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
/* must use multiple TDs, fill as much as possible. */
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
curlen = 2 * OHCI_PAGE_SIZE -
Correctly handle the conversion from virtual to physical addresses. The problem was basically (for offset > 4096): vtophys(addr) + offset != vtophys(addr + offset) Also, use TD's with a maximum size of 4k instead of 8kb for OHCI controllers. This problem occurs in drivers that use large transfer sizes: umass, host2host and ethernet with jumbo frames.
2000-02-10 18:50:19 +00:00
OHCI_PAGE_MASK(dataphys);
#elif defined(__FreeBSD__)
/* See comment above (XXX) */
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
curlen = OHCI_PAGE_SIZE -
Correctly handle the conversion from virtual to physical addresses. The problem was basically (for offset > 4096): vtophys(addr) + offset != vtophys(addr + offset) Also, use TD's with a maximum size of 4k instead of 8kb for OHCI controllers. This problem occurs in drivers that use large transfer sizes: umass, host2host and ethernet with jumbo frames.
2000-02-10 18:50:19 +00:00
OHCI_PAGE_MASK(dataphys);
#endif
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
}
DPRINTFN(4,("ohci_alloc_std_chain: dataphys=0x%08x "
"dataphysend=0x%08x len=%d curlen=%d\n",
dataphys, dataphysend,
len, curlen));
len -= curlen;
intr = len == 0 ? OHCI_TD_SET_DI(1) : OHCI_TD_NOINTR;
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
cur->td.td_flags = htole32(tdflags | intr);
cur->td.td_cbp = htole32(dataphys);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
cur->nexttd = next;
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
cur->td.td_nexttd = htole32(next->physaddr);
cur->td.td_be = htole32(dataphys + curlen - 1);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
cur->len = curlen;
cur->flags = OHCI_ADD_LEN;
DPRINTFN(10,("ohci_alloc_std_chain: cbp=0x%08x be=0x%08x\n",
dataphys, dataphys + curlen - 1));
if (len == 0)
break;
DPRINTFN(10,("ohci_alloc_std_chain: extend chain\n"));
Correctly handle the conversion from virtual to physical addresses. The problem was basically (for offset > 4096): vtophys(addr) + offset != vtophys(addr + offset) Also, use TD's with a maximum size of 4k instead of 8kb for OHCI controllers. This problem occurs in drivers that use large transfer sizes: umass, host2host and ethernet with jumbo frames.
2000-02-10 18:50:19 +00:00
offset += curlen;
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
cur = next;
}
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
if ((flags & USBD_FORCE_SHORT_XFER) &&
len % UGETW(opipe->pipe.endpoint->edesc->wMaxPacketSize) == 0) {
/* Force a 0 length transfer at the end. */
next = ohci_alloc_std(sc);
if (next == 0)
goto nomem;
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
cur->td.td_flags = htole32(tdflags | OHCI_TD_SET_DI(1));
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
cur->td.td_cbp = 0; /* indicate 0 length packet */
cur->nexttd = next;
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
cur->td.td_nexttd = htole32(next->physaddr);
cur->td.td_be = htole32(dataphys - 1);
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
cur->len = 0;
cur->flags = 0;
cur = next;
DPRINTFN(2,("ohci_alloc_std_chain: add 0 xfer\n"));
}
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
cur->flags = OHCI_CALL_DONE | OHCI_ADD_LEN;
Correctly handle the conversion from virtual to physical addresses. The problem was basically (for offset > 4096): vtophys(addr) + offset != vtophys(addr + offset) Also, use TD's with a maximum size of 4k instead of 8kb for OHCI controllers. This problem occurs in drivers that use large transfer sizes: umass, host2host and ethernet with jumbo frames.
2000-02-10 18:50:19 +00:00
*rstd = next;
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
return (USBD_NORMAL_COMPLETION);
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
nomem:
/* XXX free chain */
return (USBD_NOMEM);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
}
#if 0
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_free_std_chain(ohci_softc_t *sc, ohci_soft_td_t *std,
ohci_soft_td_t *stdend)
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
{
ohci_soft_td_t *p;
for (; std != stdend; std = p) {
p = std->nexttd;
ohci_free_std(sc, std);
}
}
#endif
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
ohci_soft_itd_t *
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_alloc_sitd(ohci_softc_t *sc)
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
{
ohci_soft_itd_t *sitd;
usbd_status err;
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
int i, s, offs;
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
usb_dma_t dma;
if (sc->sc_freeitds == NULL) {
DPRINTFN(2, ("ohci_alloc_sitd: allocating chunk\n"));
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
err = usb_allocmem(&sc->sc_bus, OHCI_SITD_SIZE * OHCI_SITD_CHUNK,
OHCI_ITD_ALIGN, &dma);
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
if (err)
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
return (NULL);
for(i = 0; i < OHCI_SITD_CHUNK; i++) {
offs = i * OHCI_SITD_SIZE;
Correctly handle the conversion from virtual to physical addresses. The problem was basically (for offset > 4096): vtophys(addr) + offset != vtophys(addr + offset) Also, use TD's with a maximum size of 4k instead of 8kb for OHCI controllers. This problem occurs in drivers that use large transfer sizes: umass, host2host and ethernet with jumbo frames.
2000-02-10 18:50:19 +00:00
sitd = (ohci_soft_itd_t *)((char*)KERNADDR(&dma, offs));
sitd->physaddr = DMAADDR(&dma, offs);
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
sitd->nextitd = sc->sc_freeitds;
sc->sc_freeitds = sitd;
}
}
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
s = splusb();
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
sitd = sc->sc_freeitds;
sc->sc_freeitds = sitd->nextitd;
memset(&sitd->itd, 0, sizeof(ohci_itd_t));
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
sitd->nextitd = NULL;
sitd->xfer = NULL;
ohci_hash_add_itd(sc, sitd);
splx(s);
#ifdef DIAGNOSTIC
sitd->isdone = 0;
#endif
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
return (sitd);
}
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_free_sitd(ohci_softc_t *sc, ohci_soft_itd_t *sitd)
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
{
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
int s;
DPRINTFN(10,("ohci_free_sitd: sitd=%p\n", sitd));
#ifdef DIAGNOSTIC
if (!sitd->isdone) {
panic("ohci_free_sitd: sitd=%p not done\n", sitd);
return;
}
#endif
s = splusb();
ohci_hash_rem_itd(sc, sitd);
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
sitd->nextitd = sc->sc_freeitds;
sc->sc_freeitds = sitd;
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
splx(s);
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
}
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
usbd_status
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_init(ohci_softc_t *sc)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
ohci_soft_ed_t *sed, *psed;
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
usbd_status err;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
int i;
Merge from NetBSD: ohci.c: -r1.68 ohcireg.h: -r1.13 date: 2000/01/31 20:17:25; author: augustss; state: Exp; Fiddle with over-current protect when turning on port power to make things work for some OHCI controllers.
2002-01-21 04:24:33 +00:00
u_int32_t s, ctl, ival, hcr, fm, per, rev, desca;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
DPRINTF(("ohci_init: start\n"));
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
#if defined(__OpenBSD__)
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
printf(",");
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
#else
Synchronisation with NetBSD 1999/11/27 - more req[uest]->xfer changes. - get the corresponding NetBSD Id's right ohci.c - move untimeout above print statement - remove usb_delay that panics the system (tsleep in intr context) when ohcidebug > 5. ugen.c - create the devices for endpoints with make_dev. uhub.c - change from using usbdebug to uhubdebug - add more debugging statements
1999-11-28 21:01:06 +00:00
printf("%s:", USBDEVNAME(sc->sc_bus.bdev));
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
#endif
Synchronisation with NetBSD 1999/11/27 - more req[uest]->xfer changes. - get the corresponding NetBSD Id's right ohci.c - move untimeout above print statement - remove usb_delay that panics the system (tsleep in intr context) when ohcidebug > 5. ugen.c - create the devices for endpoints with make_dev. uhub.c - change from using usbdebug to uhubdebug - add more debugging statements
1999-11-28 21:01:06 +00:00
rev = OREAD4(sc, OHCI_REVISION);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
printf(" OHCI version %d.%d%s\n", OHCI_REV_HI(rev), OHCI_REV_LO(rev),
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
OHCI_REV_LEGACY(rev) ? ", legacy support" : "");
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
if (OHCI_REV_HI(rev) != 1 || OHCI_REV_LO(rev) != 0) {
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
printf("%s: unsupported OHCI revision\n",
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
USBDEVNAME(sc->sc_bus.bdev));
Synchronisation with NetBSD 1999/11/27 - more req[uest]->xfer changes. - get the corresponding NetBSD Id's right ohci.c - move untimeout above print statement - remove usb_delay that panics the system (tsleep in intr context) when ohcidebug > 5. ugen.c - create the devices for endpoints with make_dev. uhub.c - change from using usbdebug to uhubdebug - add more debugging statements
1999-11-28 21:01:06 +00:00
sc->sc_bus.usbrev = USBREV_UNKNOWN;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
return (USBD_INVAL);
}
Synchronisation with NetBSD 1999/11/27 - more req[uest]->xfer changes. - get the corresponding NetBSD Id's right ohci.c - move untimeout above print statement - remove usb_delay that panics the system (tsleep in intr context) when ohcidebug > 5. ugen.c - create the devices for endpoints with make_dev. uhub.c - change from using usbdebug to uhubdebug - add more debugging statements
1999-11-28 21:01:06 +00:00
sc->sc_bus.usbrev = USBREV_1_0;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
for (i = 0; i < OHCI_HASH_SIZE; i++)
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
LIST_INIT(&sc->sc_hash_itds[i]);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
SIMPLEQ_INIT(&sc->sc_free_xfers);
Merge from NetBSD: ohci.c: revision 1.72 and 1.73 ohcivar.h: revision 1.19 and 1.20 uhci.c: revision 1.85 usbdi.h: a small part of revision 1.40 usbdivar.h: revision 1.47 Relevant commit messages from NetBSD are: date: 2000/02/22 11:30:54; author: augustss; state: Exp; Prepare a little for having USB interrupt processing done outside the hard interrupt level (in a thread or a softintr). No real soft processing done yet. ---------------------------- date: 2000/02/01 05:42:53; author: augustss; state: Exp; Put some #ifdefs around power and shutdown hooks.
2002-01-26 12:04:22 +00:00
/* XXX determine alignment by R/W */
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
/* Allocate the HCCA area. */
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
err = usb_allocmem(&sc->sc_bus, OHCI_HCCA_SIZE,
Synchronisation with NetBSD 1999/11/27 - more req[uest]->xfer changes. - get the corresponding NetBSD Id's right ohci.c - move untimeout above print statement - remove usb_delay that panics the system (tsleep in intr context) when ohcidebug > 5. ugen.c - create the devices for endpoints with make_dev. uhub.c - change from using usbdebug to uhubdebug - add more debugging statements
1999-11-28 21:01:06 +00:00
OHCI_HCCA_ALIGN, &sc->sc_hccadma);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
if (err)
return (err);
Correctly handle the conversion from virtual to physical addresses. The problem was basically (for offset > 4096): vtophys(addr) + offset != vtophys(addr + offset) Also, use TD's with a maximum size of 4k instead of 8kb for OHCI controllers. This problem occurs in drivers that use large transfer sizes: umass, host2host and ethernet with jumbo frames.
2000-02-10 18:50:19 +00:00
sc->sc_hcca = (struct ohci_hcca *)KERNADDR(&sc->sc_hccadma, 0);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
memset(sc->sc_hcca, 0, OHCI_HCCA_SIZE);
sc->sc_eintrs = OHCI_NORMAL_INTRS;
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
/* Allocate dummy ED that starts the control list. */
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
sc->sc_ctrl_head = ohci_alloc_sed(sc);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
if (sc->sc_ctrl_head == NULL) {
err = USBD_NOMEM;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
goto bad1;
}
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
sc->sc_ctrl_head->ed.ed_flags |= htole32(OHCI_ED_SKIP);
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
/* Allocate dummy ED that starts the bulk list. */
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
sc->sc_bulk_head = ohci_alloc_sed(sc);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
if (sc->sc_bulk_head == NULL) {
err = USBD_NOMEM;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
goto bad2;
}
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
sc->sc_bulk_head->ed.ed_flags |= htole32(OHCI_ED_SKIP);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
/* Allocate dummy ED that starts the isochronous list. */
sc->sc_isoc_head = ohci_alloc_sed(sc);
if (sc->sc_isoc_head == NULL) {
err = USBD_NOMEM;
goto bad3;
}
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
sc->sc_isoc_head->ed.ed_flags |= htole32(OHCI_ED_SKIP);
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
/* Allocate all the dummy EDs that make up the interrupt tree. */
for (i = 0; i < OHCI_NO_EDS; i++) {
sed = ohci_alloc_sed(sc);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
if (sed == NULL) {
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
while (--i >= 0)
ohci_free_sed(sc, sc->sc_eds[i]);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
err = USBD_NOMEM;
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
goto bad4;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
/* All ED fields are set to 0. */
sc->sc_eds[i] = sed;
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
sed->ed.ed_flags |= htole32(OHCI_ED_SKIP);
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
if (i != 0)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
psed = sc->sc_eds[(i-1) / 2];
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
else
psed= sc->sc_isoc_head;
sed->next = psed;
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
sed->ed.ed_nexted = htole32(psed->physaddr);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
/*
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
* Fill HCCA interrupt table. The bit reversal is to get
* the tree set up properly to spread the interrupts.
*/
for (i = 0; i < OHCI_NO_INTRS; i++)
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
sc->sc_hcca->hcca_interrupt_table[revbits[i]] =
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
htole32(sc->sc_eds[OHCI_NO_EDS-OHCI_NO_INTRS+i]->physaddr);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Merge from NetBSD: ohci.c: revision 1.72 and 1.73 ohcivar.h: revision 1.19 and 1.20 uhci.c: revision 1.85 usbdi.h: a small part of revision 1.40 usbdivar.h: revision 1.47 Relevant commit messages from NetBSD are: date: 2000/02/22 11:30:54; author: augustss; state: Exp; Prepare a little for having USB interrupt processing done outside the hard interrupt level (in a thread or a softintr). No real soft processing done yet. ---------------------------- date: 2000/02/01 05:42:53; author: augustss; state: Exp; Put some #ifdefs around power and shutdown hooks.
2002-01-26 12:04:22 +00:00
#ifdef OHCI_DEBUG
if (ohcidebug > 15) {
for (i = 0; i < OHCI_NO_EDS; i++) {
printf("ed#%d ", i);
ohci_dump_ed(sc->sc_eds[i]);
}
printf("iso ");
ohci_dump_ed(sc->sc_isoc_head);
}
#endif
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
/* Determine in what context we are running. */
ctl = OREAD4(sc, OHCI_CONTROL);
if (ctl & OHCI_IR) {
/* SMM active, request change */
DPRINTF(("ohci_init: SMM active, request owner change\n"));
s = OREAD4(sc, OHCI_COMMAND_STATUS);
OWRITE4(sc, OHCI_COMMAND_STATUS, s | OHCI_OCR);
for (i = 0; i < 100 && (ctl & OHCI_IR); i++) {
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
usb_delay_ms(&sc->sc_bus, 1);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
ctl = OREAD4(sc, OHCI_CONTROL);
}
if ((ctl & OHCI_IR) == 0) {
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
printf("%s: SMM does not respond, resetting\n",
USBDEVNAME(sc->sc_bus.bdev));
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
OWRITE4(sc, OHCI_CONTROL, OHCI_HCFS_RESET);
goto reset;
}
} else if ((ctl & OHCI_HCFS_MASK) != OHCI_HCFS_RESET) {
/* BIOS started controller. */
DPRINTF(("ohci_init: BIOS active\n"));
if ((ctl & OHCI_HCFS_MASK) != OHCI_HCFS_OPERATIONAL) {
OWRITE4(sc, OHCI_CONTROL, OHCI_HCFS_OPERATIONAL);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
usb_delay_ms(&sc->sc_bus, USB_RESUME_DELAY);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
} else {
DPRINTF(("ohci_init: cold started\n"));
reset:
/* Controller was cold started. */
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
usb_delay_ms(&sc->sc_bus, USB_BUS_RESET_DELAY);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
/*
* This reset should not be necessary according to the OHCI spec, but
* without it some controllers do not start.
*/
DPRINTF(("%s: resetting\n", USBDEVNAME(sc->sc_bus.bdev)));
OWRITE4(sc, OHCI_CONTROL, OHCI_HCFS_RESET);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
usb_delay_ms(&sc->sc_bus, USB_BUS_RESET_DELAY);
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
/* We now own the host controller and the bus has been reset. */
ival = OHCI_GET_IVAL(OREAD4(sc, OHCI_FM_INTERVAL));
OWRITE4(sc, OHCI_COMMAND_STATUS, OHCI_HCR); /* Reset HC */
/* Nominal time for a reset is 10 us. */
for (i = 0; i < 10; i++) {
delay(10);
hcr = OREAD4(sc, OHCI_COMMAND_STATUS) & OHCI_HCR;
if (!hcr)
break;
}
if (hcr) {
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
printf("%s: reset timeout\n", USBDEVNAME(sc->sc_bus.bdev));
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
err = USBD_IOERROR;
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
goto bad5;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
UHCI_DEBUG should of course be OHCI_DEBUG Reported by: MAEKAWA Masahide <bishop@rr.iij4u.or.jp>
1999-04-17 21:48:03 +00:00
#ifdef OHCI_DEBUG
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
if (ohcidebug > 15)
ohci_dumpregs(sc);
#endif
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
/* The controller is now in SUSPEND state, we have 2ms to finish. */
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
/* Set up HC registers. */
Correctly handle the conversion from virtual to physical addresses. The problem was basically (for offset > 4096): vtophys(addr) + offset != vtophys(addr + offset) Also, use TD's with a maximum size of 4k instead of 8kb for OHCI controllers. This problem occurs in drivers that use large transfer sizes: umass, host2host and ethernet with jumbo frames.
2000-02-10 18:50:19 +00:00
OWRITE4(sc, OHCI_HCCA, DMAADDR(&sc->sc_hccadma, 0));
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
OWRITE4(sc, OHCI_CONTROL_HEAD_ED, sc->sc_ctrl_head->physaddr);
OWRITE4(sc, OHCI_BULK_HEAD_ED, sc->sc_bulk_head->physaddr);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
/* disable all interrupts and then switch on all desired interrupts */
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
OWRITE4(sc, OHCI_INTERRUPT_DISABLE, OHCI_ALL_INTRS);
OWRITE4(sc, OHCI_INTERRUPT_ENABLE, sc->sc_eintrs | OHCI_MIE);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
/* switch on desired functional features */
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
ctl = OREAD4(sc, OHCI_CONTROL);
ctl &= ~(OHCI_CBSR_MASK | OHCI_LES | OHCI_HCFS_MASK | OHCI_IR);
ctl |= OHCI_PLE | OHCI_IE | OHCI_CLE | OHCI_BLE |
OHCI_RATIO_1_4 | OHCI_HCFS_OPERATIONAL;
/* And finally start it! */
OWRITE4(sc, OHCI_CONTROL, ctl);
/*
* The controller is now OPERATIONAL. Set a some final
* registers that should be set earlier, but that the
* controller ignores when in the SUSPEND state.
*/
fm = (OREAD4(sc, OHCI_FM_INTERVAL) & OHCI_FIT) ^ OHCI_FIT;
fm |= OHCI_FSMPS(ival) | ival;
OWRITE4(sc, OHCI_FM_INTERVAL, fm);
per = OHCI_PERIODIC(ival); /* 90% periodic */
OWRITE4(sc, OHCI_PERIODIC_START, per);
Merge from NetBSD: ohci.c: -r1.68 ohcireg.h: -r1.13 date: 2000/01/31 20:17:25; author: augustss; state: Exp; Fiddle with over-current protect when turning on port power to make things work for some OHCI controllers.
2002-01-21 04:24:33 +00:00
/* Fiddle the No OverCurrent Protection bit to avoid chip bug. */
desca = OREAD4(sc, OHCI_RH_DESCRIPTOR_A);
OWRITE4(sc, OHCI_RH_DESCRIPTOR_A, desca | OHCI_NOCP);
OWRITE4(sc, OHCI_RH_STATUS, OHCI_LPSC); /* Enable port power */
Merge from NetBSD: ohci.c (1.85), ohcireg.h (1.17): ---------------------------- date: 2000/04/01 09:27:35; author: augustss; Add a delay before reading the number of ports from the controller to avoid getting 0 from it. ----------------------------
2002-04-01 13:21:43 +00:00
usb_delay_ms(&sc->sc_bus, OHCI_ENABLE_POWER_DELAY);
Merge from NetBSD: ohci.c: -r1.68 ohcireg.h: -r1.13 date: 2000/01/31 20:17:25; author: augustss; state: Exp; Fiddle with over-current protect when turning on port power to make things work for some OHCI controllers.
2002-01-21 04:24:33 +00:00
OWRITE4(sc, OHCI_RH_DESCRIPTOR_A, desca);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Merge from NetBSD: ohci.c (1.85), ohcireg.h (1.17): ---------------------------- date: 2000/04/01 09:27:35; author: augustss; Add a delay before reading the number of ports from the controller to avoid getting 0 from it. ----------------------------
2002-04-01 13:21:43 +00:00
/*
* The AMD756 requires a delay before re-reading the register,
* otherwise it will occasionally report 0 ports.
*/
usb_delay_ms(&sc->sc_bus, OHCI_READ_DESC_DELAY);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
sc->sc_noport = OHCI_GET_NDP(OREAD4(sc, OHCI_RH_DESCRIPTOR_A));
UHCI_DEBUG should of course be OHCI_DEBUG Reported by: MAEKAWA Masahide <bishop@rr.iij4u.or.jp>
1999-04-17 21:48:03 +00:00
#ifdef OHCI_DEBUG
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
if (ohcidebug > 5)
ohci_dumpregs(sc);
#endif
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
/* Set up the bus struct. */
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
sc->sc_bus.methods = &ohci_bus_methods;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
sc->sc_bus.pipe_size = sizeof(struct ohci_pipe);
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
Merge from NetBSD: ohci.c: -r1.69 to 1.71 ohcireg.h: -r1.14 Some of these deltas are based upon patches that we submitted back to NetBSD. They got manifested slightly differently though, so I've brought back those differences to bring our code bases closer together. The logs from the NetBSD version of ohci.c: revision 1.71 date: 2000/02/01 05:42:52; author: augustss; state: Exp; lines: +13 -2 Put some #ifdefs around power and shutdown hooks. ---------------------------- revision 1.70 date: 2000/01/31 22:35:13; author: augustss; state: Exp; lines: +7 -7 Rename TAILMASK to HEADMASK, since it really masks the head pointer. From FreeBSD. ---------------------------- revision 1.69 date: 2000/01/31 22:09:13; author: augustss; state: Exp; lines: +18 -14 Change where the has table for physical-to-virtual address translation is handled. Partly from FreeBSD.
2002-01-21 05:02:21 +00:00
#if defined(__NetBSD__) || defined(__OPENBSD__)
Add the TD's to the hash when they are created not when they are used. This makes it possible to handle the done queue even when the end of tail TD has ended up on the done queue as well (the xfer in this case is NULL but the next field is valid, so we simply skip that element. Make sure we mark all the handled transfers as handled. Stop using powerhooks. They are a NetBSD-ism.
2000-03-15 22:27:24 +00:00
sc->sc_powerhook = powerhook_establish(ohci_power, sc);
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
sc->sc_shutdownhook = shutdownhook_establish(ohci_shutdown, sc);
#endif
Move the defines to the right location. Add a DIAGNOSTIC when closing interrupt pipes. Insert splx(s) which were left out with iso pipes (non-functional yet) and in a DIAGNOSTIC.
2000-01-27 23:25:58 +00:00
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
return (USBD_NORMAL_COMPLETION);
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
bad5:
for (i = 0; i < OHCI_NO_EDS; i++)
ohci_free_sed(sc, sc->sc_eds[i]);
bad4:
ohci_free_sed(sc, sc->sc_isoc_head);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
bad3:
ohci_free_sed(sc, sc->sc_ctrl_head);
bad2:
ohci_free_sed(sc, sc->sc_bulk_head);
bad1:
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
usb_freemem(&sc->sc_bus, &sc->sc_hccadma);
return (err);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
usbd_status
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_allocm(struct usbd_bus *bus, usb_dma_t *dma, u_int32_t size)
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
{
#if defined(__NetBSD__) || defined(__OpenBSD__)
struct ohci_softc *sc = (struct ohci_softc *)bus;
#endif
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
return (usb_allocmem(&sc->sc_bus, size, 0, dma));
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
}
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_freem(struct usbd_bus *bus, usb_dma_t *dma)
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
{
#if defined(__NetBSD__) || defined(__OpenBSD__)
struct ohci_softc *sc = (struct ohci_softc *)bus;
#endif
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
usb_freemem(&sc->sc_bus, dma);
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
}
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
usbd_xfer_handle
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_allocx(struct usbd_bus *bus)
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
{
struct ohci_softc *sc = (struct ohci_softc *)bus;
usbd_xfer_handle xfer;
xfer = SIMPLEQ_FIRST(&sc->sc_free_xfers);
if (xfer != NULL)
SIMPLEQ_REMOVE_HEAD(&sc->sc_free_xfers, xfer, next);
else
xfer = malloc(sizeof(*xfer), M_USB, M_NOWAIT);
if (xfer != NULL)
memset(xfer, 0, sizeof *xfer);
return (xfer);
}
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_freex(struct usbd_bus *bus, usbd_xfer_handle xfer)
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
{
struct ohci_softc *sc = (struct ohci_softc *)bus;
SIMPLEQ_INSERT_HEAD(&sc->sc_free_xfers, xfer, next);
}
/*
* Shut down the controller when the system is going down.
*/
#if defined(__NetBSD__) || defined(__OpenBSD__)
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_shutdown(void *v)
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
{
ohci_softc_t *sc = v;
DPRINTF(("ohci_shutdown: stopping the HC\n"));
OWRITE4(sc, OHCI_CONTROL, OHCI_HCFS_RESET);
}
/*
* Handle suspend/resume.
*
* We need to switch to polling mode here, because this routine is
* called from an intterupt context. This is all right since we
* are almost suspended anyway.
*/
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_power(int why, void *v)
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
{
#ifdef OHCI_DEBUG
ohci_softc_t *sc = v;
DPRINTF(("ohci_power: sc=%p, why=%d\n", sc, why));
/* XXX should suspend/resume */
ohci_dumpregs(sc);
#endif
}
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
#endif
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
UHCI_DEBUG should of course be OHCI_DEBUG Reported by: MAEKAWA Masahide <bishop@rr.iij4u.or.jp>
1999-04-17 21:48:03 +00:00
#ifdef OHCI_DEBUG
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_dumpregs(ohci_softc_t *sc)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
DPRINTF(("ohci_dumpregs: rev=0x%08x control=0x%08x command=0x%08x\n",
OREAD4(sc, OHCI_REVISION),
OREAD4(sc, OHCI_CONTROL),
OREAD4(sc, OHCI_COMMAND_STATUS)));
DPRINTF((" intrstat=0x%08x intre=0x%08x intrd=0x%08x\n",
OREAD4(sc, OHCI_INTERRUPT_STATUS),
OREAD4(sc, OHCI_INTERRUPT_ENABLE),
OREAD4(sc, OHCI_INTERRUPT_DISABLE)));
DPRINTF((" hcca=0x%08x percur=0x%08x ctrlhd=0x%08x\n",
OREAD4(sc, OHCI_HCCA),
OREAD4(sc, OHCI_PERIOD_CURRENT_ED),
OREAD4(sc, OHCI_CONTROL_HEAD_ED)));
DPRINTF((" ctrlcur=0x%08x bulkhd=0x%08x bulkcur=0x%08x\n",
OREAD4(sc, OHCI_CONTROL_CURRENT_ED),
OREAD4(sc, OHCI_BULK_HEAD_ED),
OREAD4(sc, OHCI_BULK_CURRENT_ED)));
DPRINTF((" done=0x%08x fmival=0x%08x fmrem=0x%08x\n",
OREAD4(sc, OHCI_DONE_HEAD),
OREAD4(sc, OHCI_FM_INTERVAL),
OREAD4(sc, OHCI_FM_REMAINING)));
DPRINTF((" fmnum=0x%08x perst=0x%08x lsthrs=0x%08x\n",
OREAD4(sc, OHCI_FM_NUMBER),
OREAD4(sc, OHCI_PERIODIC_START),
OREAD4(sc, OHCI_LS_THRESHOLD)));
DPRINTF((" desca=0x%08x descb=0x%08x stat=0x%08x\n",
OREAD4(sc, OHCI_RH_DESCRIPTOR_A),
OREAD4(sc, OHCI_RH_DESCRIPTOR_B),
OREAD4(sc, OHCI_RH_STATUS)));
DPRINTF((" port1=0x%08x port2=0x%08x\n",
OREAD4(sc, OHCI_RH_PORT_STATUS(1)),
OREAD4(sc, OHCI_RH_PORT_STATUS(2))));
DPRINTF((" HCCA: frame_number=0x%04x done_head=0x%08x\n",
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
le32toh(sc->sc_hcca->hcca_frame_number),
le32toh(sc->sc_hcca->hcca_done_head)));
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
#endif
Remove __P prototypes to reduce diffs between the NetBSD and FreeBSD versions.
2000-07-17 18:41:20 +00:00
Static int ohci_intr1(ohci_softc_t *);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
int
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_intr(void *p)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
ohci_softc_t *sc = p;
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
/* If we get an interrupt while polling, then just ignore it. */
Add an extra debugging message.
2000-01-10 22:33:43 +00:00
if (sc->sc_bus.use_polling) {
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
#ifdef DIAGNOSTIC
printf("ohci_intr: ignored interrupt while polling\n");
#endif
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
return (0);
Add an extra debugging message.
2000-01-10 22:33:43 +00:00
}
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
return (ohci_intr1(sc));
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
}
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static int
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_intr1(ohci_softc_t *sc)
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
{
Textual changes
1999-01-22 00:51:12 +00:00
u_int32_t intrs, eintrs;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
ohci_physaddr_t done;
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
/* In case the interrupt occurs before initialization has completed. */
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
if (sc == NULL || sc->sc_hcca == NULL) {
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
#ifdef DIAGNOSTIC
printf("ohci_intr: sc->sc_hcca == NULL\n");
#endif
If we failed to probe/attach somehow, we never have a sc->sc_hcca but the present PCI probe code still thinks we are there as the pci attach can't return an error code. This means we are in the shared interrupt list, but have not been set up. If we are sharing ints with another device, ohci_intr will be called and will coredump on a NULL reference. So just return if it is called when not set up. This fixes the symptom and not the cause. The right answer is to let the PCI system know that the attach failed, or to fail earlier (in the PCI probe). The attach() is a void fn() so it can't return failure..
1998-12-11 06:02:06 +00:00
return (0);
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
}
If we failed to probe/attach somehow, we never have a sc->sc_hcca but the present PCI probe code still thinks we are there as the pci attach can't return an error code. This means we are in the shared interrupt list, but have not been set up. If we are sharing ints with another device, ohci_intr will be called and will coredump on a NULL reference. So just return if it is called when not set up. This fixes the symptom and not the cause. The right answer is to let the PCI system know that the attach failed, or to fail earlier (in the PCI probe). The attach() is a void fn() so it can't return failure..
1998-12-11 06:02:06 +00:00
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
intrs = 0;
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
done = le32toh(sc->sc_hcca->hcca_done_head);
Add comments
2000-01-26 08:46:41 +00:00
/* The LSb of done is used to inform the HC Driver that an interrupt
* condition exists for both the Done list and for another event
* recorded in HcInterruptStatus. On an interrupt from the HC, the HC
* Driver checks the HccaDoneHead Value. If this value is 0, then the
* interrupt was caused by other than the HccaDoneHead update and the
* HcInterruptStatus register needs to be accessed to determine that
* exact interrupt cause. If HccaDoneHead is nonzero, then a Done list
* update interrupt is indicated and if the LSb of done is nonzero,
* then an additional interrupt event is indicated and
* HcInterruptStatus should be checked to determine its cause.
*/
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
if (done != 0) {
1) Bug fix: Interrupt transfers worked only for the first packet. The second packet was not seen as finished. 2) Textual changes
1999-01-13 00:50:02 +00:00
if (done & ~OHCI_DONE_INTRS)
intrs = OHCI_WDH;
Add comments
2000-01-26 08:46:41 +00:00
if (done & OHCI_DONE_INTRS) {
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
intrs |= OREAD4(sc, OHCI_INTERRUPT_STATUS);
Add comments
2000-01-26 08:46:41 +00:00
done &= ~OHCI_DONE_INTRS;
}
} else {
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
intrs = OREAD4(sc, OHCI_INTERRUPT_STATUS);
Add comments
2000-01-26 08:46:41 +00:00
}
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
Do not printf when we get called as a result of a pci shared interrupt, and add a comment to that effect.
2001-08-23 01:16:10 +00:00
if (intrs == 0) /* nothing to be done (PCI shared interrupt) */
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
return (0);
Add comments
2000-01-26 08:46:41 +00:00
intrs &= ~OHCI_MIE; /* mask out Master Interrupt Enable */
/* Acknowledge any interrupts that have happened */
OWRITE4(sc, OHCI_INTERRUPT_STATUS, intrs);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
Add comments
2000-01-26 08:46:41 +00:00
/* Any interrupts we had enabled? */
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
eintrs = intrs & sc->sc_eintrs;
if (!eintrs)
return (0);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
sc->sc_bus.intr_context++;
sc->sc_bus.no_intrs++;
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
DPRINTFN(7, ("ohci_intr: sc=%p intrs=0x%x(0x%x) eintrs=0x%x\n",
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
sc, (u_int)intrs, OREAD4(sc, OHCI_INTERRUPT_STATUS),
(u_int)eintrs));
if (eintrs & OHCI_SO) {
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
printf("%s: scheduling overrun\n",USBDEVNAME(sc->sc_bus.bdev));
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
/* XXX do what */
intrs &= ~OHCI_SO;
}
if (eintrs & OHCI_WDH) {
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
ohci_add_done(sc, done &~ OHCI_DONE_INTRS);
Merge from NetBSD: ohci.c: revision 1.72 and 1.73 ohcivar.h: revision 1.19 and 1.20 uhci.c: revision 1.85 usbdi.h: a small part of revision 1.40 usbdivar.h: revision 1.47 Relevant commit messages from NetBSD are: date: 2000/02/22 11:30:54; author: augustss; state: Exp; Prepare a little for having USB interrupt processing done outside the hard interrupt level (in a thread or a softintr). No real soft processing done yet. ---------------------------- date: 2000/02/01 05:42:53; author: augustss; state: Exp; Put some #ifdefs around power and shutdown hooks.
2002-01-26 12:04:22 +00:00
sc->sc_hcca->hcca_done_head = 0;
usb_schedsoftintr(&sc->sc_bus);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
intrs &= ~OHCI_WDH;
}
if (eintrs & OHCI_RD) {
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
printf("%s: resume detect\n", USBDEVNAME(sc->sc_bus.bdev));
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
/* XXX process resume detect */
}
if (eintrs & OHCI_UE) {
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
printf("%s: unrecoverable error, controller halted\n",
USBDEVNAME(sc->sc_bus.bdev));
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
OWRITE4(sc, OHCI_CONTROL, OHCI_HCFS_RESET);
/* XXX what else */
}
if (eintrs & OHCI_RHSC) {
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
ohci_rhsc(sc, sc->sc_intrxfer);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
intrs &= ~OHCI_RHSC;
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
/*
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
* Disable RHSC interrupt for now, because it will be
* on until the port has been reset.
*/
ohci_rhsc_able(sc, 0);
}
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
sc->sc_bus.intr_context--;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
/* Block unprocessed interrupts. XXX */
OWRITE4(sc, OHCI_INTERRUPT_DISABLE, intrs);
sc->sc_eintrs &= ~intrs;
return (1);
}
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_rhsc_able(ohci_softc_t *sc, int on)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
DPRINTFN(4, ("ohci_rhsc_able: on=%d\n", on));
if (on) {
sc->sc_eintrs |= OHCI_RHSC;
OWRITE4(sc, OHCI_INTERRUPT_ENABLE, OHCI_RHSC);
} else {
sc->sc_eintrs &= ~OHCI_RHSC;
OWRITE4(sc, OHCI_INTERRUPT_DISABLE, OHCI_RHSC);
}
}
UHCI_DEBUG should of course be OHCI_DEBUG Reported by: MAEKAWA Masahide <bishop@rr.iij4u.or.jp>
1999-04-17 21:48:03 +00:00
#ifdef OHCI_DEBUG
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
char *ohci_cc_strs[] = {
"NO_ERROR",
"CRC",
"BIT_STUFFING",
"DATA_TOGGLE_MISMATCH",
"STALL",
"DEVICE_NOT_RESPONDING",
"PID_CHECK_FAILURE",
"UNEXPECTED_PID",
"DATA_OVERRUN",
"DATA_UNDERRUN",
"BUFFER_OVERRUN",
"BUFFER_UNDERRUN",
Correct the list of error messages. It was incomplete.
2000-01-28 13:05:26 +00:00
"reserved",
"reserved",
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
"NOT_ACCESSED",
Correct the list of error messages. It was incomplete.
2000-01-28 13:05:26 +00:00
"NOT_ACCESSED"
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
};
#endif
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
void
ohci_add_done(ohci_softc_t *sc, ohci_physaddr_t done)
{
ohci_soft_itd_t *sitd, *sidone, **ip;
ohci_soft_td_t *std, *sdone, **p;
/* Reverse the done list. */
for (sdone = NULL, sidone = NULL; done != 0; ) {
std = ohci_hash_find_td(sc, done);
if (std != NULL) {
std->dnext = sdone;
done = le32toh(std->td.td_nexttd);
sdone = std;
DPRINTFN(10,("add TD %p\n", std));
continue;
}
sitd = ohci_hash_find_itd(sc, done);
if (sitd != NULL) {
sitd->dnext = sidone;
done = le32toh(sitd->itd.itd_nextitd);
sidone = sitd;
DPRINTFN(5,("add ITD %p\n", sitd));
continue;
}
panic("ohci_add_done: addr 0x%08lx not found\n", (u_long)done);
}
/* sdone & sidone now hold the done lists. */
/* Put them on the already processed lists. */
for (p = &sc->sc_sdone; *p != NULL; p = &(*p)->dnext)
;
*p = sdone;
for (ip = &sc->sc_sidone; *ip != NULL; ip = &(*ip)->dnext)
;
*ip = sidone;
}
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
void
MFNetBSD: ohci.c (1.99), uhci.c (1.133), usb.c (1.49), usb_port.h (1.39), usbdi.c (1.79), usbdi.h (1.47), usbdivar.h (1.62) date: 2001/01/21 02:39:52; author: augustss; Add code to use soft interrupt to handle USB interrupt processing. Don't enable the code since it doesn't work with the kludgy Ethernet drivers.
2002-04-01 21:34:01 +00:00
ohci_softintr(void *v)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
MFNetBSD: ohci.c (1.99), uhci.c (1.133), usb.c (1.49), usb_port.h (1.39), usbdi.c (1.79), usbdi.h (1.47), usbdivar.h (1.62) date: 2001/01/21 02:39:52; author: augustss; Add code to use soft interrupt to handle USB interrupt processing. Don't enable the code since it doesn't work with the kludgy Ethernet drivers.
2002-04-01 21:34:01 +00:00
ohci_softc_t *sc = v;
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
ohci_soft_itd_t *sitd, *sidone, *sitdnext;
ohci_soft_td_t *std, *sdone, *stdnext;
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
usbd_xfer_handle xfer;
Merge from NetBSD: ohci.c: revision 1.72 and 1.73 ohcivar.h: revision 1.19 and 1.20 uhci.c: revision 1.85 usbdi.h: a small part of revision 1.40 usbdivar.h: revision 1.47 Relevant commit messages from NetBSD are: date: 2000/02/22 11:30:54; author: augustss; state: Exp; Prepare a little for having USB interrupt processing done outside the hard interrupt level (in a thread or a softintr). No real soft processing done yet. ---------------------------- date: 2000/02/01 05:42:53; author: augustss; state: Exp; Put some #ifdefs around power and shutdown hooks.
2002-01-26 12:04:22 +00:00
int len, cc, s;
sc->sc_bus.intr_context++;
s = splhardusb();
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
sdone = sc->sc_sdone;
sc->sc_sdone = NULL;
sidone = sc->sc_sidone;
sc->sc_sidone = NULL;
Merge from NetBSD: ohci.c: revision 1.72 and 1.73 ohcivar.h: revision 1.19 and 1.20 uhci.c: revision 1.85 usbdi.h: a small part of revision 1.40 usbdivar.h: revision 1.47 Relevant commit messages from NetBSD are: date: 2000/02/22 11:30:54; author: augustss; state: Exp; Prepare a little for having USB interrupt processing done outside the hard interrupt level (in a thread or a softintr). No real soft processing done yet. ---------------------------- date: 2000/02/01 05:42:53; author: augustss; state: Exp; Put some #ifdefs around power and shutdown hooks.
2002-01-26 12:04:22 +00:00
splx(s);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
DPRINTFN(10,("ohci_process_done: sdone=%p sidone=%p\n", sdone, sidone));
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
UHCI_DEBUG should of course be OHCI_DEBUG Reported by: MAEKAWA Masahide <bishop@rr.iij4u.or.jp>
1999-04-17 21:48:03 +00:00
#ifdef OHCI_DEBUG
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
if (ohcidebug > 10) {
DPRINTF(("ohci_process_done: TD done:\n"));
Fix debugging output. The TDs are linked through the dnext field, not nexttd.
2000-01-29 18:01:15 +00:00
for (std = sdone; std; std = std->dnext)
Add the TD's to the hash when they are created not when they are used. This makes it possible to handle the done queue even when the end of tail TD has ended up on the done queue as well (the xfer in this case is NULL but the next field is valid, so we simply skip that element. Make sure we mark all the handled transfers as handled. Stop using powerhooks. They are a NetBSD-ism.
2000-03-15 22:27:24 +00:00
ohci_dump_td(sdone);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
#endif
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
for (std = sdone; std; std = stdnext) {
Synchronisation with NetBSD 1999/11/27 - more req[uest]->xfer changes. - get the corresponding NetBSD Id's right ohci.c - move untimeout above print statement - remove usb_delay that panics the system (tsleep in intr context) when ohcidebug > 5. ugen.c - create the devices for endpoints with make_dev. uhub.c - change from using usbdebug to uhubdebug - add more debugging statements
1999-11-28 21:01:06 +00:00
xfer = std->xfer;
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
stdnext = std->dnext;
Move the defines to the right location. Add a DIAGNOSTIC when closing interrupt pipes. Insert splx(s) which were left out with iso pipes (non-functional yet) and in a DIAGNOSTIC.
2000-01-27 23:25:58 +00:00
DPRINTFN(5, ("ohci_process_done: std=%p xfer=%p hcpriv=%p\n",
Merge from NetBSD: ohci.c: -r1.69 to 1.71 ohcireg.h: -r1.14 Some of these deltas are based upon patches that we submitted back to NetBSD. They got manifested slightly differently though, so I've brought back those differences to bring our code bases closer together. The logs from the NetBSD version of ohci.c: revision 1.71 date: 2000/02/01 05:42:52; author: augustss; state: Exp; lines: +13 -2 Put some #ifdefs around power and shutdown hooks. ---------------------------- revision 1.70 date: 2000/01/31 22:35:13; author: augustss; state: Exp; lines: +7 -7 Rename TAILMASK to HEADMASK, since it really masks the head pointer. From FreeBSD. ---------------------------- revision 1.69 date: 2000/01/31 22:09:13; author: augustss; state: Exp; lines: +18 -14 Change where the has table for physical-to-virtual address translation is handled. Partly from FreeBSD.
2002-01-21 05:02:21 +00:00
std, xfer, (xfer ? xfer->hcpriv : NULL)));
Add the TD's to the hash when they are created not when they are used. This makes it possible to handle the done queue even when the end of tail TD has ended up on the done queue as well (the xfer in this case is NULL but the next field is valid, so we simply skip that element. Make sure we mark all the handled transfers as handled. Stop using powerhooks. They are a NetBSD-ism.
2000-03-15 22:27:24 +00:00
if (xfer == NULL || (std->flags & OHCI_TD_HANDLED)) {
/* xfer == NULL: There seems to be no xfer associated
* with this TD. It is tailp that happened to end up on
* the done queue.
* flags & OHCI_TD_HANDLED: The TD has already been
* handled by process_done and should not be done again.
*/
continue;
}
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
if (xfer->status == USBD_CANCELLED ||
xfer->status == USBD_TIMEOUT) {
Add comments and debugging info.
2000-01-29 14:41:48 +00:00
DPRINTF(("ohci_process_done: cancel/timeout, xfer=%p\n",
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
xfer));
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
/* Handled by abort routine. */
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
continue;
}
usb_uncallout(xfer->timeout_handle, ohci_timeout, xfer);
cc = OHCI_TD_GET_CC(le32toh(std->td.td_flags));
if (cc == OHCI_CC_NO_ERROR) {
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
len = std->len;
if (std->td.td_cbp != 0)
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
len -= le32toh(std->td.td_be) -
le32toh(std->td.td_cbp) + 1;
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
if (std->flags & OHCI_ADD_LEN)
xfer->actlen += len;
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
if (std->flags & OHCI_CALL_DONE) {
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
xfer->status = USBD_NORMAL_COMPLETION;
usb_transfer_complete(xfer);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
ohci_free_std(sc, std);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
} else {
/*
* Endpoint is halted. First unlink all the TDs
* belonging to the failed transfer, and then restart
* the endpoint.
*/
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
ohci_soft_td_t *p, *n;
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
struct ohci_pipe *opipe =
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
(struct ohci_pipe *)xfer->pipe;
Add comments and debugging info.
2000-01-29 14:41:48 +00:00
DPRINTF(("ohci_process_done: err cc=%d (%s), xfer=%p\n",
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
OHCI_TD_GET_CC(le32toh(std->td.td_flags)),
ohci_cc_strs[OHCI_TD_GET_CC(le32toh(std->td.td_flags))],
Add comments and debugging info.
2000-01-29 14:41:48 +00:00
xfer));
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
Add the TD's to the hash when they are created not when they are used. This makes it possible to handle the done queue even when the end of tail TD has ended up on the done queue as well (the xfer in this case is NULL but the next field is valid, so we simply skip that element. Make sure we mark all the handled transfers as handled. Stop using powerhooks. They are a NetBSD-ism.
2000-03-15 22:27:24 +00:00
/* Mark all the TDs in the done queue for the current
* xfer as handled
*/
for (p = stdnext; p; p = p->dnext) {
if (p->xfer == xfer)
p->flags |= OHCI_TD_HANDLED;
}
/* remove TDs for the current xfer from the ED */
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
for (p = std; p->xfer == xfer; p = n) {
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
n = p->nexttd;
ohci_free_std(sc, p);
}
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
opipe->sed->ed.ed_headp = htole32(p->physaddr);
Add the TD's to the hash when they are created not when they are used. This makes it possible to handle the done queue even when the end of tail TD has ended up on the done queue as well (the xfer in this case is NULL but the next field is valid, so we simply skip that element. Make sure we mark all the handled transfers as handled. Stop using powerhooks. They are a NetBSD-ism.
2000-03-15 22:27:24 +00:00
/* XXX why is this being done? Why not OHCI_BLF too */
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
OWRITE4(sc, OHCI_COMMAND_STATUS, OHCI_CLF);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
if (cc == OHCI_CC_STALL)
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
xfer->status = USBD_STALLED;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
else
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
xfer->status = USBD_IOERROR;
Add comments and debugging info.
2000-01-29 14:41:48 +00:00
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
usb_transfer_complete(xfer);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
}
Merge from NetBSD: ohci.c: revision 1.72 and 1.73 ohcivar.h: revision 1.19 and 1.20 uhci.c: revision 1.85 usbdi.h: a small part of revision 1.40 usbdivar.h: revision 1.47 Relevant commit messages from NetBSD are: date: 2000/02/22 11:30:54; author: augustss; state: Exp; Prepare a little for having USB interrupt processing done outside the hard interrupt level (in a thread or a softintr). No real soft processing done yet. ---------------------------- date: 2000/02/01 05:42:53; author: augustss; state: Exp; Put some #ifdefs around power and shutdown hooks.
2002-01-26 12:04:22 +00:00
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
#ifdef OHCI_DEBUG
if (ohcidebug > 10) {
DPRINTF(("ohci_process_done: ITD done:\n"));
ohci_dump_itds(sidone);
}
#endif
for (sitd = sidone; sitd != NULL; sitd = sitdnext) {
xfer = sitd->xfer;
sitdnext = sitd->dnext;
DPRINTFN(1, ("ohci_process_done: sitd=%p xfer=%p hcpriv=%p\n",
sitd, xfer, xfer ? xfer->hcpriv : 0));
if (xfer == NULL)
continue;
if (xfer->status == USBD_CANCELLED ||
xfer->status == USBD_TIMEOUT) {
DPRINTF(("ohci_process_done: cancel/timeout %p\n",
xfer));
/* Handled by abort routine. */
continue;
}
#ifdef DIAGNOSTIC
if (sitd->isdone)
printf("ohci_softintr: sitd=%p is done\n", sitd);
sitd->isdone = 1;
#endif
cc = OHCI_ITD_GET_CC(le32toh(sitd->itd.itd_flags));
if (cc == OHCI_CC_NO_ERROR) {
/* XXX compute length for input */
struct ohci_pipe *opipe =
(struct ohci_pipe *)xfer->pipe;
if (sitd->flags & OHCI_CALL_DONE) {
opipe->u.iso.inuse -= xfer->nframes;
MFNetBSD: ohci.c rev 1.94, uhci.c rev 1.126 date: 2000/11/10 14:11:49; author: augustss; Update frlengths after a isoc transfer. Suggested by Yuri <yuri@tsoft.com>
2002-04-01 20:01:41 +00:00
/* XXX update frlengths with actual length */
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
/* XXX xfer->actlen = actlen; */
xfer->status = USBD_NORMAL_COMPLETION;
usb_transfer_complete(xfer);
}
} else {
/* XXX Do more */
xfer->status = USBD_IOERROR;
usb_transfer_complete(xfer);
}
}
Merge from NetBSD: ohci.c: revision 1.72 and 1.73 ohcivar.h: revision 1.19 and 1.20 uhci.c: revision 1.85 usbdi.h: a small part of revision 1.40 usbdivar.h: revision 1.47 Relevant commit messages from NetBSD are: date: 2000/02/22 11:30:54; author: augustss; state: Exp; Prepare a little for having USB interrupt processing done outside the hard interrupt level (in a thread or a softintr). No real soft processing done yet. ---------------------------- date: 2000/02/01 05:42:53; author: augustss; state: Exp; Put some #ifdefs around power and shutdown hooks.
2002-01-26 12:04:22 +00:00
sc->sc_bus.intr_context--;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_device_ctrl_done(usbd_xfer_handle xfer)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
DPRINTFN(10,("ohci_ctrl_done: xfer=%p\n", xfer));
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
#ifdef DIAGNOSTIC
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
if (!(xfer->rqflags & URQ_REQUEST)) {
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
panic("ohci_ctrl_done: not a request\n");
}
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
#endif
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
xfer->hcpriv = NULL;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_device_intr_done(usbd_xfer_handle xfer)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
struct ohci_pipe *opipe = (struct ohci_pipe *)xfer->pipe;
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
ohci_softc_t *sc = (ohci_softc_t *)opipe->pipe.device->bus;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
ohci_soft_ed_t *sed = opipe->sed;
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
ohci_soft_td_t *data, *tail;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
DPRINTFN(10,("ohci_intr_done: xfer=%p, actlen=%d\n",
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
xfer, xfer->actlen));
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
xfer->hcpriv = NULL;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
if (xfer->pipe->repeat) {
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
data = opipe->tail.td;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
tail = ohci_alloc_std(sc); /* XXX should reuse TD */
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
if (tail == NULL) {
xfer->status = USBD_NOMEM;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
return;
}
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
tail->xfer = NULL;
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
data->td.td_flags = htole32(
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
OHCI_TD_IN | OHCI_TD_NOCC |
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
OHCI_TD_SET_DI(1) | OHCI_TD_TOGGLE_CARRY);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
if (xfer->flags & USBD_SHORT_XFER_OK)
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
data->td.td_flags |= htole32(OHCI_TD_R);
data->td.td_cbp = htole32(DMAADDR(&xfer->dmabuf, 0));
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
data->nexttd = tail;
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
data->td.td_nexttd = htole32(tail->physaddr);
data->td.td_be = htole32(htole32(data->td.td_cbp) + xfer->length - 1);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
data->len = xfer->length;
data->xfer = xfer;
data->flags = OHCI_CALL_DONE | OHCI_ADD_LEN;
xfer->hcpriv = data;
xfer->actlen = 0;
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
sed->ed.ed_tailp = htole32(tail->physaddr);
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
opipe->tail.td = tail;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
}
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_device_bulk_done(usbd_xfer_handle xfer)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
DPRINTFN(10,("ohci_bulk_done: xfer=%p, actlen=%d\n",
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
xfer, xfer->actlen));
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
xfer->hcpriv = NULL;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_rhsc(ohci_softc_t *sc, usbd_xfer_handle xfer)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
usbd_pipe_handle pipe;
struct ohci_pipe *opipe;
u_char *p;
int i, m;
int hstatus;
hstatus = OREAD4(sc, OHCI_RH_STATUS);
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
DPRINTF(("ohci_rhsc: sc=%p xfer=%p hstatus=0x%08x\n",
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
sc, xfer, hstatus));
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
if (xfer == NULL) {
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
/* Just ignore the change. */
return;
}
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
pipe = xfer->pipe;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
opipe = (struct ohci_pipe *)pipe;
Correctly handle the conversion from virtual to physical addresses. The problem was basically (for offset > 4096): vtophys(addr) + offset != vtophys(addr + offset) Also, use TD's with a maximum size of 4k instead of 8kb for OHCI controllers. This problem occurs in drivers that use large transfer sizes: umass, host2host and ethernet with jumbo frames.
2000-02-10 18:50:19 +00:00
p = KERNADDR(&xfer->dmabuf, 0);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
m = min(sc->sc_noport, xfer->length * 8 - 1);
memset(p, 0, xfer->length);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
for (i = 1; i <= m; i++) {
Add a clarifying comment (from NetBSD). Update the $NetBSD$ ident to reflect reality.
2002-04-01 13:26:27 +00:00
/* Pick out CHANGE bits from the status reg. */
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
if (OREAD4(sc, OHCI_RH_PORT_STATUS(i)) >> 16)
p[i/8] |= 1 << (i%8);
}
DPRINTF(("ohci_rhsc: change=0x%02x\n", *p));
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
xfer->actlen = xfer->length;
xfer->status = USBD_NORMAL_COMPLETION;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
usb_transfer_complete(xfer);
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
}
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_root_intr_done(usbd_xfer_handle xfer)
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
{
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
xfer->hcpriv = NULL;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
Merge from NetBSD: revision 1.65 date: 2000/01/25 12:06:21; author: augustss; state: Exp; lines: +10 -2 Add done method for root control transfers.
2002-01-21 04:15:39 +00:00
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_root_ctrl_done(usbd_xfer_handle xfer)
Merge from NetBSD: revision 1.65 date: 2000/01/25 12:06:21; author: augustss; state: Exp; lines: +10 -2 Add done method for root control transfers.
2002-01-21 04:15:39 +00:00
{
xfer->hcpriv = NULL;
}
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
/*
* Wait here until controller claims to have an interrupt.
* Then call ohci_intr and return. Use timeout to avoid waiting
* too long.
*/
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_waitintr(ohci_softc_t *sc, usbd_xfer_handle xfer)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
int timo = xfer->timeout;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
int usecs;
u_int32_t intrs;
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
xfer->status = USBD_IN_PROGRESS;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
for (usecs = timo * 1000000 / hz; usecs > 0; usecs -= 1000) {
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
usb_delay_ms(&sc->sc_bus, 1);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
intrs = OREAD4(sc, OHCI_INTERRUPT_STATUS) & sc->sc_eintrs;
1) Bug fix: Interrupt transfers worked only for the first packet. The second packet was not seen as finished. 2) Textual changes
1999-01-13 00:50:02 +00:00
DPRINTFN(15,("ohci_waitintr: 0x%04x\n", intrs));
UHCI_DEBUG should of course be OHCI_DEBUG Reported by: MAEKAWA Masahide <bishop@rr.iij4u.or.jp>
1999-04-17 21:48:03 +00:00
#ifdef OHCI_DEBUG
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
if (ohcidebug > 15)
ohci_dumpregs(sc);
#endif
if (intrs) {
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
ohci_intr1(sc);
if (xfer->status != USBD_IN_PROGRESS)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
return;
}
}
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
/* Timeout */
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
DPRINTF(("ohci_waitintr: timeout\n"));
Add the TD's to the hash when they are created not when they are used. This makes it possible to handle the done queue even when the end of tail TD has ended up on the done queue as well (the xfer in this case is NULL but the next field is valid, so we simply skip that element. Make sure we mark all the handled transfers as handled. Stop using powerhooks. They are a NetBSD-ism.
2000-03-15 22:27:24 +00:00
#ifdef OHCI_DEBUG
ohci_dumpregs(sc);
#endif
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
xfer->status = USBD_TIMEOUT;
usb_transfer_complete(xfer);
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
/* XXX should free TD */
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_poll(struct usbd_bus *bus)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
ohci_softc_t *sc = (ohci_softc_t *)bus;
if (OREAD4(sc, OHCI_INTERRUPT_STATUS) & sc->sc_eintrs)
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
ohci_intr1(sc);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
usbd_status
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_device_request(usbd_xfer_handle xfer)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
struct ohci_pipe *opipe = (struct ohci_pipe *)xfer->pipe;
usb_device_request_t *req = &xfer->request;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
usbd_device_handle dev = opipe->pipe.device;
ohci_softc_t *sc = (ohci_softc_t *)dev->bus;
int addr = dev->address;
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
ohci_soft_td_t *setup, *data = 0, *stat, *next, *tail;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
ohci_soft_ed_t *sed;
int isread;
int len;
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
usbd_status err;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
int s;
isread = req->bmRequestType & UT_READ;
len = UGETW(req->wLength);
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
DPRINTFN(3,("ohci_device_control type=0x%02x, request=0x%02x, "
"wValue=0x%04x, wIndex=0x%04x len=%d, addr=%d, endpt=%d\n",
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
req->bmRequestType, req->bRequest, UGETW(req->wValue),
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
UGETW(req->wIndex), len, addr,
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
opipe->pipe.endpoint->edesc->bEndpointAddress));
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
setup = opipe->tail.td;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
stat = ohci_alloc_std(sc);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
if (stat == NULL) {
err = USBD_NOMEM;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
goto bad1;
}
tail = ohci_alloc_std(sc);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
if (tail == NULL) {
err = USBD_NOMEM;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
goto bad2;
}
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
tail->xfer = NULL;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
sed = opipe->sed;
opipe->u.ctl.length = len;
/* Update device address and length since they may have changed. */
/* XXX This only needs to be done once, but it's too early in open. */
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
sed->ed.ed_flags = htole32(
(le32toh(sed->ed.ed_flags) & ~(OHCI_ED_ADDRMASK | OHCI_ED_MAXPMASK)) |
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
OHCI_ED_SET_FA(addr) |
OHCI_ED_SET_MAXP(UGETW(opipe->pipe.endpoint->edesc->wMaxPacketSize)));
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
/* Set up data transaction */
if (len != 0) {
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
data = ohci_alloc_std(sc);
if (data == NULL) {
err = USBD_NOMEM;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
goto bad3;
}
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
data->td.td_flags = htole32(
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
(isread ? OHCI_TD_IN : OHCI_TD_OUT) | OHCI_TD_NOCC |
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
OHCI_TD_TOGGLE_1 | OHCI_TD_NOINTR |
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
(xfer->flags & USBD_SHORT_XFER_OK ? OHCI_TD_R : 0));
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
data->td.td_cbp = htole32(DMAADDR(&xfer->dmabuf, 0));
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
data->nexttd = stat;
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
data->td.td_nexttd = htole32(stat->physaddr);
data->td.td_be = htole32(le32toh(data->td.td_cbp) + len - 1);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
data->len = len;
data->xfer = xfer;
data->flags = OHCI_ADD_LEN;
next = data;
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
stat->flags = OHCI_CALL_DONE;
} else {
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
next = stat;
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
/* XXX ADD_LEN? */
stat->flags = OHCI_CALL_DONE | OHCI_ADD_LEN;
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
}
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Correctly handle the conversion from virtual to physical addresses. The problem was basically (for offset > 4096): vtophys(addr) + offset != vtophys(addr + offset) Also, use TD's with a maximum size of 4k instead of 8kb for OHCI controllers. This problem occurs in drivers that use large transfer sizes: umass, host2host and ethernet with jumbo frames.
2000-02-10 18:50:19 +00:00
memcpy(KERNADDR(&opipe->u.ctl.reqdma, 0), req, sizeof *req);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
setup->td.td_flags = htole32(OHCI_TD_SETUP | OHCI_TD_NOCC |
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
OHCI_TD_TOGGLE_0 | OHCI_TD_NOINTR);
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
setup->td.td_cbp = htole32(DMAADDR(&opipe->u.ctl.reqdma, 0));
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
setup->nexttd = next;
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
setup->td.td_nexttd = htole32(next->physaddr);
setup->td.td_be = htole32(le32toh(setup->td.td_cbp) + sizeof *req - 1);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
setup->len = 0; /* XXX The number of byte we count */
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
setup->xfer = xfer;
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
setup->flags = 0;
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
xfer->hcpriv = setup;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
stat->td.td_flags = htole32(
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
(isread ? OHCI_TD_OUT : OHCI_TD_IN) | OHCI_TD_NOCC |
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
OHCI_TD_TOGGLE_1 | OHCI_TD_SET_DI(1));
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
stat->td.td_cbp = 0;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
stat->nexttd = tail;
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
stat->td.td_nexttd = htole32(tail->physaddr);
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
stat->td.td_be = 0;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
stat->len = 0;
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
stat->xfer = xfer;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Clean up debugging output
1999-05-03 23:11:41 +00:00
#ifdef OHCI_DEBUG
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
if (ohcidebug > 5) {
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
DPRINTF(("ohci_device_request:\n"));
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
ohci_dump_ed(sed);
ohci_dump_tds(setup);
}
#endif
/* Insert ED in schedule */
s = splusb();
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
sed->ed.ed_tailp = htole32(tail->physaddr);
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
opipe->tail.td = tail;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
OWRITE4(sc, OHCI_COMMAND_STATUS, OHCI_CLF);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
if (xfer->timeout && !sc->sc_bus.use_polling) {
Huge merge from NetBSD: usbdi.c (1.61): =================================================================== revision 1.61 date: 2000/01/31 20:13:07; author: augustss; lines: +20 -4 Change the way the HC done method is invoked a little. =================================================================== usbdi.c (1.65): =================================================================== revision 1.65 date: 2000/03/08 15:34:10; author: augustss; lines: +4 -2 Get the status right when a polled transfer times out. =================================================================== ohci.c (1.79), uhci.c (1.89), uhcivar.h (1.24), usb_port.h (1.22), usbdivar.h (1.48): =================================================================== date: 2000/03/23 07:01:46; author: thorpej; New callout mechanism with two major improvements over the old timeout()/untimeout() API: - Clients supply callout handle storage, thus eliminating problems of resource allocation. - Insertion and removal of callouts is constant time, important as this facility is used quite a lot in the kernel. The old timeout()/untimeout() API has been removed from the kernel. =================================================================== uhci.c (1.80), usbdi.c (1.66): =================================================================== date: 2000/03/23 18:59:10; author: thorpej; Shake out some bugs from the callout changes. =================================================================== ohci.c (1.80), uhci.c (1.91), uhcivar.h (1.25), usb_port.h (1.23), usbdi.c (1.67), usbdivar.h (1.49): =================================================================== date: 2000/03/24 22:03:30; author: augustss; Some cleanup and renaming of the callouts used in USB drivers. =================================================================== uhci.c (1.92), uhcivar.h (1.26): =================================================================== date: 2000/03/24 22:57:58; author: augustss; Two major changes: Make each xfer have its own intr_info. This is necessary if we want to queue multiple xfers on an endpoint. This should get rid of the (mostly harmless) DIAGNOSTICs about intr_infos (not) being done. Change (again!) how xfers are aborted. Aborting a TD is a nightmare on the braindead UHCI controller. (Unless you stop the HC, thereby losing isoc traffic.) Hopefully I got it right this time. =================================================================== usbdivar.h (1.50): =================================================================== revision 1.50 date: 2000/03/25 00:10:19; author: augustss; lines: +4 -2 GC an unsued field and add some DIAGNOSTIC in xfer. =================================================================== ums.c: Use the callout functions instead of the timeout ones. uhci.c (1.93): =================================================================== revision 1.93 date: 2000/03/25 00:11:21; author: augustss; lines: +26 -1 Add more DIAGNOSTIC when aborting isoc. =================================================================== uhci.c (1.94), usbdivar.h (1.51): =================================================================== date: 2000/03/25 07:13:05; author: augustss; More DIAGNOSTIC. Initialize a callout handle I forgot. =================================================================== uhci.c (1.95): =================================================================== revision 1.95 date: 2000/03/25 07:23:12; author: augustss; Exp; lines: +24 -7 Improve uhci_dump_ii(). =================================================================== ohci.c (1.81), uhci.c (1.96), uhcivar.h (1.27), usb_subr.c (1.68), usbdi.c (1.68), usbdivar.h (1.52): =================================================================== date: 2000/03/25 18:02:33; author: augustss; Rename and move around callout handles to make it more sane. Add some DIAGNOSTIC. Fix buglet in isoc abort on UHCI. =================================================================== uhci.c (1.98): =================================================================== revision 1.98 date: 2000/03/27 07:39:48; author: augustss; lines: +12 -4 Make it compile without DIAGNOSTIC. =================================================================== uhci.c (1.99): =================================================================== revision 1.99 date: 2000/03/27 08:01:09; author: augustss; lines: +1 -5 Remove some debug nonsense. =================================================================== uhci.c (1.100): =================================================================== revision 1.100 date: 2000/03/27 09:41:36; author: augustss; lines: +13 -3 Don't mess with QH in bulk abort for the moment. =================================================================== uhci.c (1.102): =================================================================== revision 1.102 date: 2000/03/27 22:42:57; author: augustss; lines: +66 -26 Be a little more careful when aborting. Preallocate some TDs for large buffers. =================================================================== uhci.c (1.103): =================================================================== date: 2000/03/28 09:47:10; author: augustss; lines: +11 -1 Another patch for xfer abort... XXX The current xfer queueing and aborting semantics should really XXX be changed. It cannot be implemented in a sane way on UHCI. XXX One day when I have lots of time I'll redesign it... =================================================================== uhci.c (1.104): Correct a debug message. uhci.c (1.105): Be more defensive in a DIAGNOSTIC test. uhci.c (1.106): =================================================================== revision 1.106 date: 2000/03/29 01:49:13; author: augustss; lines: +14 -309 *SIGH* Revert back to the old method of aborting xfers. I had tested the new stuff for two months now, but as soon as I commited it the problems started to appear. Murphy, no doubt... =================================================================== usb_subr.c (1.70), usbdi.c (1.71), usbdivar.h (1.53): =================================================================== revision 1.70 date: 2000/03/29 01:45:20; author: augustss; lines: +2 -1 Do not accept new xfers for queuing while a pipe is aborting. ===================================================================
2002-03-16 12:06:01 +00:00
usb_callout(xfer->timeout_handle, MS_TO_TICKS(xfer->timeout),
ohci_timeout, xfer);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
splx(s);
Clean up debugging output
1999-05-03 23:11:41 +00:00
#ifdef OHCI_DEBUG
Add comments and debugging info.
2000-01-29 14:41:48 +00:00
if (ohcidebug > 25) {
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
usb_delay_ms(&sc->sc_bus, 5);
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
DPRINTF(("ohci_device_request: status=%x\n",
OREAD4(sc, OHCI_COMMAND_STATUS)));
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
ohci_dump_ed(sed);
ohci_dump_tds(setup);
}
#endif
return (USBD_NORMAL_COMPLETION);
bad3:
ohci_free_std(sc, tail);
bad2:
ohci_free_std(sc, stat);
bad1:
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
return (err);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
/*
* Add an ED to the schedule. Called at splusb().
*/
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_add_ed(ohci_soft_ed_t *sed, ohci_soft_ed_t *head)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
SPLUSBCHECK;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
sed->next = head->next;
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
sed->ed.ed_nexted = head->ed.ed_nexted;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
head->next = sed;
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
head->ed.ed_nexted = htole32(sed->physaddr);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
/*
* Remove an ED from the schedule. Called at splusb().
*/
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_rem_ed(ohci_soft_ed_t *sed, ohci_soft_ed_t *head)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_soft_ed_t *p;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
SPLUSBCHECK;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
/* XXX */
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
for (p = head; p == NULL && p->next != sed; p = p->next)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
;
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
if (p == NULL)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
panic("ohci_rem_ed: ED not found\n");
p->next = sed->next;
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
p->ed.ed_nexted = sed->ed.ed_nexted;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
/*
* When a transfer is completed the TD is added to the done queue by
* the host controller. This queue is the processed by software.
* Unfortunately the queue contains the physical address of the TD
* and we have no simple way to translate this back to a kernel address.
* To make the translation possible (and fast) we use a hash table of
* TDs currently in the schedule. The physical address is used as the
* hash value.
*/
#define HASH(a) (((a) >> 4) % OHCI_HASH_SIZE)
/* Called at splusb() */
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_hash_add_td(ohci_softc_t *sc, ohci_soft_td_t *std)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
int h = HASH(std->physaddr);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
SPLUSBCHECK;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
LIST_INSERT_HEAD(&sc->sc_hash_tds[h], std, hnext);
}
/* Called at splusb() */
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_hash_rem_td(ohci_softc_t *sc, ohci_soft_td_t *std)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
SPLUSBCHECK;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
LIST_REMOVE(std, hnext);
}
ohci_soft_td_t *
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_hash_find_td(ohci_softc_t *sc, ohci_physaddr_t a)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
int h = HASH(a);
ohci_soft_td_t *std;
Add comments and debugging info.
2000-01-29 14:41:48 +00:00
/* if these are present they should be masked out at an earlier
* stage.
*/
Merge from NetBSD: ohci.c: -r1.69 to 1.71 ohcireg.h: -r1.14 Some of these deltas are based upon patches that we submitted back to NetBSD. They got manifested slightly differently though, so I've brought back those differences to bring our code bases closer together. The logs from the NetBSD version of ohci.c: revision 1.71 date: 2000/02/01 05:42:52; author: augustss; state: Exp; lines: +13 -2 Put some #ifdefs around power and shutdown hooks. ---------------------------- revision 1.70 date: 2000/01/31 22:35:13; author: augustss; state: Exp; lines: +7 -7 Rename TAILMASK to HEADMASK, since it really masks the head pointer. From FreeBSD. ---------------------------- revision 1.69 date: 2000/01/31 22:09:13; author: augustss; state: Exp; lines: +18 -14 Change where the has table for physical-to-virtual address translation is handled. Partly from FreeBSD.
2002-01-21 05:02:21 +00:00
KASSERT((a&~OHCI_HEADMASK) == 0, ("%s: 0x%b has lower bits set\n",
Tripmine for bad hardware.
2000-01-29 11:50:44 +00:00
USBDEVNAME(sc->sc_bus.bdev),
(int) a, "\20\1HALT\2TOGGLE"));
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
for (std = LIST_FIRST(&sc->sc_hash_tds[h]);
Synchronisation with NetBSD 1999/11/27 - more req[uest]->xfer changes. - get the corresponding NetBSD Id's right ohci.c - move untimeout above print statement - remove usb_delay that panics the system (tsleep in intr context) when ohcidebug > 5. ugen.c - create the devices for endpoints with make_dev. uhub.c - change from using usbdebug to uhubdebug - add more debugging statements
1999-11-28 21:01:06 +00:00
std != NULL;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
std = LIST_NEXT(std, hnext))
if (std->physaddr == a)
return (std);
Move the defines to the right location. Add a DIAGNOSTIC when closing interrupt pipes. Insert splx(s) which were left out with iso pipes (non-functional yet) and in a DIAGNOSTIC.
2000-01-27 23:25:58 +00:00
First part of the fix for ohci_hash_find_td panic. Some controllers submit bogus pointers to the Done queue. ohci_hash_find_td fails to find these in its hash and panics. Instead of panicing we now assume the whole done queue is lost and let the timeout code to clean up the mess after us.
2000-01-29 14:53:47 +00:00
DPRINTF(("%s: ohci_hash_find_td: addr 0x%08lx not found\n",
USBDEVNAME(sc->sc_bus.bdev), (u_long) a));
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
return (NULL);
}
/* Called at splusb() */
void
ohci_hash_add_itd(ohci_softc_t *sc, ohci_soft_itd_t *sitd)
{
int h = HASH(sitd->physaddr);
SPLUSBCHECK;
DPRINTFN(10,("ohci_hash_add_itd: sitd=%p physaddr=0x%08lx\n",
sitd, (u_long)sitd->physaddr));
LIST_INSERT_HEAD(&sc->sc_hash_itds[h], sitd, hnext);
}
/* Called at splusb() */
void
ohci_hash_rem_itd(ohci_softc_t *sc, ohci_soft_itd_t *sitd)
{
SPLUSBCHECK;
DPRINTFN(10,("ohci_hash_rem_itd: sitd=%p physaddr=0x%08lx\n",
sitd, (u_long)sitd->physaddr));
LIST_REMOVE(sitd, hnext);
}
ohci_soft_itd_t *
ohci_hash_find_itd(ohci_softc_t *sc, ohci_physaddr_t a)
{
int h = HASH(a);
ohci_soft_itd_t *sitd;
for (sitd = LIST_FIRST(&sc->sc_hash_itds[h]);
sitd != NULL;
sitd = LIST_NEXT(sitd, hnext))
if (sitd->physaddr == a)
return (sitd);
return (NULL);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_timeout(void *addr)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
usbd_xfer_handle xfer = addr;
int s;
DPRINTF(("ohci_timeout: xfer=%p\n", xfer));
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
s = splusb();
xfer->device->bus->intr_context++;
Synchronisation with NetBSD 1999/11/27 - more req[uest]->xfer changes. - get the corresponding NetBSD Id's right ohci.c - move untimeout above print statement - remove usb_delay that panics the system (tsleep in intr context) when ohcidebug > 5. ugen.c - create the devices for endpoints with make_dev. uhub.c - change from using usbdebug to uhubdebug - add more debugging statements
1999-11-28 21:01:06 +00:00
ohci_abort_xfer(xfer, USBD_TIMEOUT);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
xfer->device->bus->intr_context--;
splx(s);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
UHCI_DEBUG should of course be OHCI_DEBUG Reported by: MAEKAWA Masahide <bishop@rr.iij4u.or.jp>
1999-04-17 21:48:03 +00:00
#ifdef OHCI_DEBUG
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_dump_tds(ohci_soft_td_t *std)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
for (; std; std = std->nexttd)
ohci_dump_td(std);
}
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_dump_td(ohci_soft_td_t *std)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
MFNetBSD: ohci.c rev 1.92, uhci.c rev 1.122 (part) date: 2000/08/08 19:51:46; author: tv; state: Exp; lines: +24 -13 %b -> bitmask_snprintf() Because this code is shared, add a macro for bitmask_snprintf() that should expand to the equivalent snprintf() on non-NetBSD systems. This is only used in ?HCI_DEBUG cases anyway.
2002-04-01 19:42:51 +00:00
char sbuf[128];
bitmask_snprintf((int)le32toh(std->td.td_flags),
"\20\23R\24OUT\25IN\31TOG1\32SETTOGGLE",
sbuf, sizeof(sbuf));
DPRINTF(("TD(%p) at %08lx: %s delay=%d ec=%d cc=%d\ncbp=0x%08lx "
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
"nexttd=0x%08lx be=0x%08lx\n",
MFNetBSD: ohci.c rev 1.92, uhci.c rev 1.122 (part) date: 2000/08/08 19:51:46; author: tv; state: Exp; lines: +24 -13 %b -> bitmask_snprintf() Because this code is shared, add a macro for bitmask_snprintf() that should expand to the equivalent snprintf() on non-NetBSD systems. This is only used in ?HCI_DEBUG cases anyway.
2002-04-01 19:42:51 +00:00
std, (u_long)std->physaddr, sbuf,
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
OHCI_TD_GET_DI(le32toh(std->td.td_flags)),
OHCI_TD_GET_EC(le32toh(std->td.td_flags)),
OHCI_TD_GET_CC(le32toh(std->td.td_flags)),
(u_long)le32toh(std->td.td_cbp),
(u_long)le32toh(std->td.td_nexttd),
(u_long)le32toh(std->td.td_be)));
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
void
ohci_dump_itd(ohci_soft_itd_t *sitd)
{
int i;
DPRINTF(("ITD(%p) at %08lx: sf=%d di=%d fc=%d cc=%d\n"
"bp0=0x%08lx next=0x%08lx be=0x%08lx\n",
sitd, (u_long)sitd->physaddr,
OHCI_ITD_GET_SF(le32toh(sitd->itd.itd_flags)),
OHCI_ITD_GET_DI(le32toh(sitd->itd.itd_flags)),
OHCI_ITD_GET_FC(le32toh(sitd->itd.itd_flags)),
OHCI_ITD_GET_CC(le32toh(sitd->itd.itd_flags)),
(u_long)le32toh(sitd->itd.itd_bp0),
(u_long)le32toh(sitd->itd.itd_nextitd),
(u_long)le32toh(sitd->itd.itd_be)));
for (i = 0; i < OHCI_ITD_NOFFSET; i++)
DPRINTF(("offs[%d]=0x%04x ", i,
(u_int)le16toh(sitd->itd.itd_offset[i])));
DPRINTF(("\n"));
}
void
ohci_dump_itds(ohci_soft_itd_t *sitd)
{
for (; sitd; sitd = sitd->nextitd)
ohci_dump_itd(sitd);
}
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_dump_ed(ohci_soft_ed_t *sed)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
MFNetBSD: ohci.c rev 1.92, uhci.c rev 1.122 (part) date: 2000/08/08 19:51:46; author: tv; state: Exp; lines: +24 -13 %b -> bitmask_snprintf() Because this code is shared, add a macro for bitmask_snprintf() that should expand to the equivalent snprintf() on non-NetBSD systems. This is only used in ?HCI_DEBUG cases anyway.
2002-04-01 19:42:51 +00:00
char sbuf[128], sbuf2[128];
bitmask_snprintf((int)le32toh(sed->ed.ed_flags),
"\20\14OUT\15IN\16LOWSPEED\17SKIP\20ISO",
sbuf, sizeof(sbuf));
bitmask_snprintf((u_long)le32toh(sed->ed.ed_headp),
"\20\1HALT\2CARRY", sbuf2, sizeof(sbuf2));
DPRINTF(("ED(%p) at 0x%08lx: addr=%d endpt=%d maxp=%d %s\ntailp=0x%08lx "
"headflags=%s headp=0x%08lx nexted=0x%08lx\n",
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
sed, (u_long)sed->physaddr,
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
OHCI_ED_GET_FA(le32toh(sed->ed.ed_flags)),
OHCI_ED_GET_EN(le32toh(sed->ed.ed_flags)),
MFNetBSD: ohci.c rev 1.92, uhci.c rev 1.122 (part) date: 2000/08/08 19:51:46; author: tv; state: Exp; lines: +24 -13 %b -> bitmask_snprintf() Because this code is shared, add a macro for bitmask_snprintf() that should expand to the equivalent snprintf() on non-NetBSD systems. This is only used in ?HCI_DEBUG cases anyway.
2002-04-01 19:42:51 +00:00
OHCI_ED_GET_MAXP(le32toh(sed->ed.ed_flags)), sbuf,
(u_long)le32toh(sed->ed.ed_tailp), sbuf2,
Merge from NetBSD: ohci.c: revision 1.72 and 1.73 ohcivar.h: revision 1.19 and 1.20 uhci.c: revision 1.85 usbdi.h: a small part of revision 1.40 usbdivar.h: revision 1.47 Relevant commit messages from NetBSD are: date: 2000/02/22 11:30:54; author: augustss; state: Exp; Prepare a little for having USB interrupt processing done outside the hard interrupt level (in a thread or a softintr). No real soft processing done yet. ---------------------------- date: 2000/02/01 05:42:53; author: augustss; state: Exp; Put some #ifdefs around power and shutdown hooks.
2002-01-26 12:04:22 +00:00
(u_long)le32toh(sed->ed.ed_headp),
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
(u_long)le32toh(sed->ed.ed_nexted)));
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
#endif
usbd_status
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_open(usbd_pipe_handle pipe)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
usbd_device_handle dev = pipe->device;
ohci_softc_t *sc = (ohci_softc_t *)dev->bus;
usb_endpoint_descriptor_t *ed = pipe->endpoint->edesc;
struct ohci_pipe *opipe = (struct ohci_pipe *)pipe;
u_int8_t addr = dev->address;
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
u_int8_t xfertype = ed->bmAttributes & UE_XFERTYPE;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
ohci_soft_ed_t *sed;
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
ohci_soft_td_t *std = NULL;
ohci_soft_itd_t *sitd;
ohci_physaddr_t tdphys;
u_int32_t fmt;
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
usbd_status err;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
int s;
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
int ival;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
DPRINTFN(1, ("ohci_open: pipe=%p, addr=%d, endpt=%d (%d)\n",
pipe, addr, ed->bEndpointAddress, sc->sc_addr));
Huge merge from NetBSD: usbdi.c (1.61): =================================================================== revision 1.61 date: 2000/01/31 20:13:07; author: augustss; lines: +20 -4 Change the way the HC done method is invoked a little. =================================================================== usbdi.c (1.65): =================================================================== revision 1.65 date: 2000/03/08 15:34:10; author: augustss; lines: +4 -2 Get the status right when a polled transfer times out. =================================================================== ohci.c (1.79), uhci.c (1.89), uhcivar.h (1.24), usb_port.h (1.22), usbdivar.h (1.48): =================================================================== date: 2000/03/23 07:01:46; author: thorpej; New callout mechanism with two major improvements over the old timeout()/untimeout() API: - Clients supply callout handle storage, thus eliminating problems of resource allocation. - Insertion and removal of callouts is constant time, important as this facility is used quite a lot in the kernel. The old timeout()/untimeout() API has been removed from the kernel. =================================================================== uhci.c (1.80), usbdi.c (1.66): =================================================================== date: 2000/03/23 18:59:10; author: thorpej; Shake out some bugs from the callout changes. =================================================================== ohci.c (1.80), uhci.c (1.91), uhcivar.h (1.25), usb_port.h (1.23), usbdi.c (1.67), usbdivar.h (1.49): =================================================================== date: 2000/03/24 22:03:30; author: augustss; Some cleanup and renaming of the callouts used in USB drivers. =================================================================== uhci.c (1.92), uhcivar.h (1.26): =================================================================== date: 2000/03/24 22:57:58; author: augustss; Two major changes: Make each xfer have its own intr_info. This is necessary if we want to queue multiple xfers on an endpoint. This should get rid of the (mostly harmless) DIAGNOSTICs about intr_infos (not) being done. Change (again!) how xfers are aborted. Aborting a TD is a nightmare on the braindead UHCI controller. (Unless you stop the HC, thereby losing isoc traffic.) Hopefully I got it right this time. =================================================================== usbdivar.h (1.50): =================================================================== revision 1.50 date: 2000/03/25 00:10:19; author: augustss; lines: +4 -2 GC an unsued field and add some DIAGNOSTIC in xfer. =================================================================== ums.c: Use the callout functions instead of the timeout ones. uhci.c (1.93): =================================================================== revision 1.93 date: 2000/03/25 00:11:21; author: augustss; lines: +26 -1 Add more DIAGNOSTIC when aborting isoc. =================================================================== uhci.c (1.94), usbdivar.h (1.51): =================================================================== date: 2000/03/25 07:13:05; author: augustss; More DIAGNOSTIC. Initialize a callout handle I forgot. =================================================================== uhci.c (1.95): =================================================================== revision 1.95 date: 2000/03/25 07:23:12; author: augustss; Exp; lines: +24 -7 Improve uhci_dump_ii(). =================================================================== ohci.c (1.81), uhci.c (1.96), uhcivar.h (1.27), usb_subr.c (1.68), usbdi.c (1.68), usbdivar.h (1.52): =================================================================== date: 2000/03/25 18:02:33; author: augustss; Rename and move around callout handles to make it more sane. Add some DIAGNOSTIC. Fix buglet in isoc abort on UHCI. =================================================================== uhci.c (1.98): =================================================================== revision 1.98 date: 2000/03/27 07:39:48; author: augustss; lines: +12 -4 Make it compile without DIAGNOSTIC. =================================================================== uhci.c (1.99): =================================================================== revision 1.99 date: 2000/03/27 08:01:09; author: augustss; lines: +1 -5 Remove some debug nonsense. =================================================================== uhci.c (1.100): =================================================================== revision 1.100 date: 2000/03/27 09:41:36; author: augustss; lines: +13 -3 Don't mess with QH in bulk abort for the moment. =================================================================== uhci.c (1.102): =================================================================== revision 1.102 date: 2000/03/27 22:42:57; author: augustss; lines: +66 -26 Be a little more careful when aborting. Preallocate some TDs for large buffers. =================================================================== uhci.c (1.103): =================================================================== date: 2000/03/28 09:47:10; author: augustss; lines: +11 -1 Another patch for xfer abort... XXX The current xfer queueing and aborting semantics should really XXX be changed. It cannot be implemented in a sane way on UHCI. XXX One day when I have lots of time I'll redesign it... =================================================================== uhci.c (1.104): Correct a debug message. uhci.c (1.105): Be more defensive in a DIAGNOSTIC test. uhci.c (1.106): =================================================================== revision 1.106 date: 2000/03/29 01:49:13; author: augustss; lines: +14 -309 *SIGH* Revert back to the old method of aborting xfers. I had tested the new stuff for two months now, but as soon as I commited it the problems started to appear. Murphy, no doubt... =================================================================== usb_subr.c (1.70), usbdi.c (1.71), usbdivar.h (1.53): =================================================================== revision 1.70 date: 2000/03/29 01:45:20; author: augustss; lines: +2 -1 Do not accept new xfers for queuing while a pipe is aborting. ===================================================================
2002-03-16 12:06:01 +00:00
Merge from NetBSD: ---------------------------- revision 1.90 date: 2000/05/08 18:28:46; author: thorpej; lines: +8 -3 Quiet some uninitialized variable warnings that do in fact look legitimate. ----------------------------
2002-04-01 13:28:49 +00:00
std = NULL;
sed = NULL;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
if (addr == sc->sc_addr) {
switch (ed->bEndpointAddress) {
case USB_CONTROL_ENDPOINT:
pipe->methods = &ohci_root_ctrl_methods;
break;
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
case UE_DIR_IN | OHCI_INTR_ENDPT:
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
pipe->methods = &ohci_root_intr_methods;
break;
default:
return (USBD_INVAL);
}
} else {
sed = ohci_alloc_sed(sc);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
if (sed == NULL)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
goto bad0;
opipe->sed = sed;
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
if (xfertype == UE_ISOCHRONOUS) {
sitd = ohci_alloc_sitd(sc);
if (sitd == NULL) {
ohci_free_sitd(sc, sitd);
goto bad1;
}
opipe->tail.itd = sitd;
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
tdphys = sitd->physaddr;
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
fmt = OHCI_ED_FORMAT_ISO;
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
if (UE_GET_DIR(ed->bEndpointAddress) == UE_DIR_IN)
fmt |= OHCI_ED_DIR_IN;
else
fmt |= OHCI_ED_DIR_OUT;
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
} else {
std = ohci_alloc_std(sc);
if (std == NULL) {
ohci_free_std(sc, std);
goto bad1;
}
opipe->tail.td = std;
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
tdphys = std->physaddr;
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
fmt = OHCI_ED_FORMAT_GEN | OHCI_ED_DIR_TD;
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
}
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
sed->ed.ed_flags = htole32(
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
OHCI_ED_SET_FA(addr) |
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
OHCI_ED_SET_EN(ed->bEndpointAddress) |
Merge from NetBSD: Pave the way for USB2, by replacing 'lowspeed' with 'speed', so that it can take the values USB_SPEED_LOW, USB_SPEED_FULL or in time USB_SPEED_HIGH.
2002-02-16 00:51:26 +00:00
(dev->speed == USB_SPEED_LOW ? OHCI_ED_SPEED : 0) |
fmt |
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
OHCI_ED_SET_MAXP(UGETW(ed->wMaxPacketSize)));
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
sed->ed.ed_headp = sed->ed.ed_tailp = htole32(tdphys);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
switch (xfertype) {
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
case UE_CONTROL:
pipe->methods = &ohci_device_ctrl_methods;
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
err = usb_allocmem(&sc->sc_bus,
sizeof(usb_device_request_t),
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
0, &opipe->u.ctl.reqdma);
if (err)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
goto bad;
s = splusb();
ohci_add_ed(sed, sc->sc_ctrl_head);
splx(s);
break;
case UE_INTERRUPT:
pipe->methods = &ohci_device_intr_methods;
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
ival = pipe->interval;
if (ival == USBD_DEFAULT_INTERVAL)
ival = ed->bInterval;
return (ohci_device_setintr(sc, opipe, ival));
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
case UE_ISOCHRONOUS:
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
pipe->methods = &ohci_device_isoc_methods;
return (ohci_setup_isoc(pipe));
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
case UE_BULK:
pipe->methods = &ohci_device_bulk_methods;
s = splusb();
ohci_add_ed(sed, sc->sc_bulk_head);
splx(s);
break;
}
}
return (USBD_NORMAL_COMPLETION);
bad:
Merge from NetBSD: ---------------------------- revision 1.90 date: 2000/05/08 18:28:46; author: thorpej; lines: +8 -3 Quiet some uninitialized variable warnings that do in fact look legitimate. ----------------------------
2002-04-01 13:28:49 +00:00
if (std != NULL)
ohci_free_std(sc, std);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
bad1:
Merge from NetBSD: ---------------------------- revision 1.90 date: 2000/05/08 18:28:46; author: thorpej; lines: +8 -3 Quiet some uninitialized variable warnings that do in fact look legitimate. ----------------------------
2002-04-01 13:28:49 +00:00
if (sed != NULL)
ohci_free_sed(sc, sed);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
bad0:
return (USBD_NOMEM);
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
/*
* Close a reqular pipe.
* Assumes that there are no pending transactions.
*/
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_close_pipe(usbd_pipe_handle pipe, ohci_soft_ed_t *head)
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
{
struct ohci_pipe *opipe = (struct ohci_pipe *)pipe;
ohci_softc_t *sc = (ohci_softc_t *)pipe->device->bus;
ohci_soft_ed_t *sed = opipe->sed;
int s;
s = splusb();
#ifdef DIAGNOSTIC
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
sed->ed.ed_flags |= htole32(OHCI_ED_SKIP);
Merge from NetBSD: ohci.c: -r1.69 to 1.71 ohcireg.h: -r1.14 Some of these deltas are based upon patches that we submitted back to NetBSD. They got manifested slightly differently though, so I've brought back those differences to bring our code bases closer together. The logs from the NetBSD version of ohci.c: revision 1.71 date: 2000/02/01 05:42:52; author: augustss; state: Exp; lines: +13 -2 Put some #ifdefs around power and shutdown hooks. ---------------------------- revision 1.70 date: 2000/01/31 22:35:13; author: augustss; state: Exp; lines: +7 -7 Rename TAILMASK to HEADMASK, since it really masks the head pointer. From FreeBSD. ---------------------------- revision 1.69 date: 2000/01/31 22:09:13; author: augustss; state: Exp; lines: +18 -14 Change where the has table for physical-to-virtual address translation is handled. Partly from FreeBSD.
2002-01-21 05:02:21 +00:00
if ((le32toh(sed->ed.ed_tailp) & OHCI_HEADMASK) !=
(le32toh(sed->ed.ed_headp) & OHCI_HEADMASK)) {
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
ohci_physaddr_t td = sed->ed.ed_headp;
ohci_soft_td_t *std;
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
for (std = LIST_FIRST(&sc->sc_hash_tds[HASH(td)]);
Synchronisation with NetBSD 1999/11/27 - more req[uest]->xfer changes. - get the corresponding NetBSD Id's right ohci.c - move untimeout above print statement - remove usb_delay that panics the system (tsleep in intr context) when ohcidebug > 5. ugen.c - create the devices for endpoints with make_dev. uhub.c - change from using usbdebug to uhubdebug - add more debugging statements
1999-11-28 21:01:06 +00:00
std != NULL;
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
std = LIST_NEXT(std, hnext))
if (std->physaddr == td)
break;
printf("ohci_close_pipe: pipe not empty sed=%p hd=0x%x "
"tl=0x%x pipe=%p, std=%p\n", sed,
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
(int)le32toh(sed->ed.ed_headp),
(int)le32toh(sed->ed.ed_tailp),
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
pipe, std);
usb_delay_ms(&sc->sc_bus, 2);
Merge from NetBSD: ohci.c: -r1.69 to 1.71 ohcireg.h: -r1.14 Some of these deltas are based upon patches that we submitted back to NetBSD. They got manifested slightly differently though, so I've brought back those differences to bring our code bases closer together. The logs from the NetBSD version of ohci.c: revision 1.71 date: 2000/02/01 05:42:52; author: augustss; state: Exp; lines: +13 -2 Put some #ifdefs around power and shutdown hooks. ---------------------------- revision 1.70 date: 2000/01/31 22:35:13; author: augustss; state: Exp; lines: +7 -7 Rename TAILMASK to HEADMASK, since it really masks the head pointer. From FreeBSD. ---------------------------- revision 1.69 date: 2000/01/31 22:09:13; author: augustss; state: Exp; lines: +18 -14 Change where the has table for physical-to-virtual address translation is handled. Partly from FreeBSD.
2002-01-21 05:02:21 +00:00
if ((le32toh(sed->ed.ed_tailp) & OHCI_HEADMASK) !=
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
(le32toh(sed->ed.ed_headp) & OHCI_HEADMASK))
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
printf("ohci_close_pipe: pipe still not empty\n");
}
#endif
ohci_rem_ed(sed, head);
splx(s);
ohci_free_sed(sc, opipe->sed);
}
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
/*
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
* Abort a device request.
* If this routine is called at splusb() it guarantees that the request
* will be removed from the hardware scheduling and that the callback
* for it will be called with USBD_CANCELLED status.
* It's impossible to guarantee that the requested transfer will not
* have happened since the hardware runs concurrently.
* If the transaction has already happened we rely on the ordinary
* interrupt processing to process it.
*/
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_abort_xfer(usbd_xfer_handle xfer, usbd_status status)
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
{
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
struct ohci_pipe *opipe = (struct ohci_pipe *)xfer->pipe;
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
ohci_soft_ed_t *sed;
Synchronisation with NetBSD 1999/11/27 - more req[uest]->xfer changes. - get the corresponding NetBSD Id's right ohci.c - move untimeout above print statement - remove usb_delay that panics the system (tsleep in intr context) when ohcidebug > 5. ugen.c - create the devices for endpoints with make_dev. uhub.c - change from using usbdebug to uhubdebug - add more debugging statements
1999-11-28 21:01:06 +00:00
DPRINTF(("ohci_abort_xfer: xfer=%p pipe=%p\n", xfer, opipe));
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
xfer->status = status;
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
Huge merge from NetBSD: usbdi.c (1.61): =================================================================== revision 1.61 date: 2000/01/31 20:13:07; author: augustss; lines: +20 -4 Change the way the HC done method is invoked a little. =================================================================== usbdi.c (1.65): =================================================================== revision 1.65 date: 2000/03/08 15:34:10; author: augustss; lines: +4 -2 Get the status right when a polled transfer times out. =================================================================== ohci.c (1.79), uhci.c (1.89), uhcivar.h (1.24), usb_port.h (1.22), usbdivar.h (1.48): =================================================================== date: 2000/03/23 07:01:46; author: thorpej; New callout mechanism with two major improvements over the old timeout()/untimeout() API: - Clients supply callout handle storage, thus eliminating problems of resource allocation. - Insertion and removal of callouts is constant time, important as this facility is used quite a lot in the kernel. The old timeout()/untimeout() API has been removed from the kernel. =================================================================== uhci.c (1.80), usbdi.c (1.66): =================================================================== date: 2000/03/23 18:59:10; author: thorpej; Shake out some bugs from the callout changes. =================================================================== ohci.c (1.80), uhci.c (1.91), uhcivar.h (1.25), usb_port.h (1.23), usbdi.c (1.67), usbdivar.h (1.49): =================================================================== date: 2000/03/24 22:03:30; author: augustss; Some cleanup and renaming of the callouts used in USB drivers. =================================================================== uhci.c (1.92), uhcivar.h (1.26): =================================================================== date: 2000/03/24 22:57:58; author: augustss; Two major changes: Make each xfer have its own intr_info. This is necessary if we want to queue multiple xfers on an endpoint. This should get rid of the (mostly harmless) DIAGNOSTICs about intr_infos (not) being done. Change (again!) how xfers are aborted. Aborting a TD is a nightmare on the braindead UHCI controller. (Unless you stop the HC, thereby losing isoc traffic.) Hopefully I got it right this time. =================================================================== usbdivar.h (1.50): =================================================================== revision 1.50 date: 2000/03/25 00:10:19; author: augustss; lines: +4 -2 GC an unsued field and add some DIAGNOSTIC in xfer. =================================================================== ums.c: Use the callout functions instead of the timeout ones. uhci.c (1.93): =================================================================== revision 1.93 date: 2000/03/25 00:11:21; author: augustss; lines: +26 -1 Add more DIAGNOSTIC when aborting isoc. =================================================================== uhci.c (1.94), usbdivar.h (1.51): =================================================================== date: 2000/03/25 07:13:05; author: augustss; More DIAGNOSTIC. Initialize a callout handle I forgot. =================================================================== uhci.c (1.95): =================================================================== revision 1.95 date: 2000/03/25 07:23:12; author: augustss; Exp; lines: +24 -7 Improve uhci_dump_ii(). =================================================================== ohci.c (1.81), uhci.c (1.96), uhcivar.h (1.27), usb_subr.c (1.68), usbdi.c (1.68), usbdivar.h (1.52): =================================================================== date: 2000/03/25 18:02:33; author: augustss; Rename and move around callout handles to make it more sane. Add some DIAGNOSTIC. Fix buglet in isoc abort on UHCI. =================================================================== uhci.c (1.98): =================================================================== revision 1.98 date: 2000/03/27 07:39:48; author: augustss; lines: +12 -4 Make it compile without DIAGNOSTIC. =================================================================== uhci.c (1.99): =================================================================== revision 1.99 date: 2000/03/27 08:01:09; author: augustss; lines: +1 -5 Remove some debug nonsense. =================================================================== uhci.c (1.100): =================================================================== revision 1.100 date: 2000/03/27 09:41:36; author: augustss; lines: +13 -3 Don't mess with QH in bulk abort for the moment. =================================================================== uhci.c (1.102): =================================================================== revision 1.102 date: 2000/03/27 22:42:57; author: augustss; lines: +66 -26 Be a little more careful when aborting. Preallocate some TDs for large buffers. =================================================================== uhci.c (1.103): =================================================================== date: 2000/03/28 09:47:10; author: augustss; lines: +11 -1 Another patch for xfer abort... XXX The current xfer queueing and aborting semantics should really XXX be changed. It cannot be implemented in a sane way on UHCI. XXX One day when I have lots of time I'll redesign it... =================================================================== uhci.c (1.104): Correct a debug message. uhci.c (1.105): Be more defensive in a DIAGNOSTIC test. uhci.c (1.106): =================================================================== revision 1.106 date: 2000/03/29 01:49:13; author: augustss; lines: +14 -309 *SIGH* Revert back to the old method of aborting xfers. I had tested the new stuff for two months now, but as soon as I commited it the problems started to appear. Murphy, no doubt... =================================================================== usb_subr.c (1.70), usbdi.c (1.71), usbdivar.h (1.53): =================================================================== revision 1.70 date: 2000/03/29 01:45:20; author: augustss; lines: +2 -1 Do not accept new xfers for queuing while a pipe is aborting. ===================================================================
2002-03-16 12:06:01 +00:00
usb_uncallout(xfer->timeout_handle, ohci_timeout, xfer);
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
sed = opipe->sed;
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
sed->ed.ed_flags |= htole32(OHCI_ED_SKIP); /* force hardware skip */
Add comments and debugging info.
2000-01-29 14:41:48 +00:00
#ifdef OHCI_DEBUG
DPRINTFN(1,("ohci_abort_xfer: stop ed=%p\n", sed));
ohci_dump_ed(sed);
#endif
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
#if 1
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
if (xfer->device->bus->intr_context) {
/* We have no process context, so we can't use tsleep(). */
Huge merge from NetBSD: usbdi.c (1.61): =================================================================== revision 1.61 date: 2000/01/31 20:13:07; author: augustss; lines: +20 -4 Change the way the HC done method is invoked a little. =================================================================== usbdi.c (1.65): =================================================================== revision 1.65 date: 2000/03/08 15:34:10; author: augustss; lines: +4 -2 Get the status right when a polled transfer times out. =================================================================== ohci.c (1.79), uhci.c (1.89), uhcivar.h (1.24), usb_port.h (1.22), usbdivar.h (1.48): =================================================================== date: 2000/03/23 07:01:46; author: thorpej; New callout mechanism with two major improvements over the old timeout()/untimeout() API: - Clients supply callout handle storage, thus eliminating problems of resource allocation. - Insertion and removal of callouts is constant time, important as this facility is used quite a lot in the kernel. The old timeout()/untimeout() API has been removed from the kernel. =================================================================== uhci.c (1.80), usbdi.c (1.66): =================================================================== date: 2000/03/23 18:59:10; author: thorpej; Shake out some bugs from the callout changes. =================================================================== ohci.c (1.80), uhci.c (1.91), uhcivar.h (1.25), usb_port.h (1.23), usbdi.c (1.67), usbdivar.h (1.49): =================================================================== date: 2000/03/24 22:03:30; author: augustss; Some cleanup and renaming of the callouts used in USB drivers. =================================================================== uhci.c (1.92), uhcivar.h (1.26): =================================================================== date: 2000/03/24 22:57:58; author: augustss; Two major changes: Make each xfer have its own intr_info. This is necessary if we want to queue multiple xfers on an endpoint. This should get rid of the (mostly harmless) DIAGNOSTICs about intr_infos (not) being done. Change (again!) how xfers are aborted. Aborting a TD is a nightmare on the braindead UHCI controller. (Unless you stop the HC, thereby losing isoc traffic.) Hopefully I got it right this time. =================================================================== usbdivar.h (1.50): =================================================================== revision 1.50 date: 2000/03/25 00:10:19; author: augustss; lines: +4 -2 GC an unsued field and add some DIAGNOSTIC in xfer. =================================================================== ums.c: Use the callout functions instead of the timeout ones. uhci.c (1.93): =================================================================== revision 1.93 date: 2000/03/25 00:11:21; author: augustss; lines: +26 -1 Add more DIAGNOSTIC when aborting isoc. =================================================================== uhci.c (1.94), usbdivar.h (1.51): =================================================================== date: 2000/03/25 07:13:05; author: augustss; More DIAGNOSTIC. Initialize a callout handle I forgot. =================================================================== uhci.c (1.95): =================================================================== revision 1.95 date: 2000/03/25 07:23:12; author: augustss; Exp; lines: +24 -7 Improve uhci_dump_ii(). =================================================================== ohci.c (1.81), uhci.c (1.96), uhcivar.h (1.27), usb_subr.c (1.68), usbdi.c (1.68), usbdivar.h (1.52): =================================================================== date: 2000/03/25 18:02:33; author: augustss; Rename and move around callout handles to make it more sane. Add some DIAGNOSTIC. Fix buglet in isoc abort on UHCI. =================================================================== uhci.c (1.98): =================================================================== revision 1.98 date: 2000/03/27 07:39:48; author: augustss; lines: +12 -4 Make it compile without DIAGNOSTIC. =================================================================== uhci.c (1.99): =================================================================== revision 1.99 date: 2000/03/27 08:01:09; author: augustss; lines: +1 -5 Remove some debug nonsense. =================================================================== uhci.c (1.100): =================================================================== revision 1.100 date: 2000/03/27 09:41:36; author: augustss; lines: +13 -3 Don't mess with QH in bulk abort for the moment. =================================================================== uhci.c (1.102): =================================================================== revision 1.102 date: 2000/03/27 22:42:57; author: augustss; lines: +66 -26 Be a little more careful when aborting. Preallocate some TDs for large buffers. =================================================================== uhci.c (1.103): =================================================================== date: 2000/03/28 09:47:10; author: augustss; lines: +11 -1 Another patch for xfer abort... XXX The current xfer queueing and aborting semantics should really XXX be changed. It cannot be implemented in a sane way on UHCI. XXX One day when I have lots of time I'll redesign it... =================================================================== uhci.c (1.104): Correct a debug message. uhci.c (1.105): Be more defensive in a DIAGNOSTIC test. uhci.c (1.106): =================================================================== revision 1.106 date: 2000/03/29 01:49:13; author: augustss; lines: +14 -309 *SIGH* Revert back to the old method of aborting xfers. I had tested the new stuff for two months now, but as soon as I commited it the problems started to appear. Murphy, no doubt... =================================================================== usb_subr.c (1.70), usbdi.c (1.71), usbdivar.h (1.53): =================================================================== revision 1.70 date: 2000/03/29 01:45:20; author: augustss; lines: +2 -1 Do not accept new xfers for queuing while a pipe is aborting. ===================================================================
2002-03-16 12:06:01 +00:00
usb_callout(xfer->pipe->abort_handle,
hz / USB_FRAMES_PER_SECOND, ohci_abort_xfer_end, xfer);
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
} else {
Synchronisation with NetBSD 1999/11/27 - more req[uest]->xfer changes. - get the corresponding NetBSD Id's right ohci.c - move untimeout above print statement - remove usb_delay that panics the system (tsleep in intr context) when ohcidebug > 5. ugen.c - create the devices for endpoints with make_dev. uhub.c - change from using usbdebug to uhubdebug - add more debugging statements
1999-11-28 21:01:06 +00:00
#if defined(DIAGNOSTIC) && defined(__i386__) && defined(__FreeBSD__)
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
KASSERT(curthread->td_intr_nesting_level == 0,
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
("ohci_abort_req in interrupt context"));
#endif
usb_delay_ms(opipe->pipe.device->bus, 1);
Synchronisation with NetBSD 1999/11/27 - more req[uest]->xfer changes. - get the corresponding NetBSD Id's right ohci.c - move untimeout above print statement - remove usb_delay that panics the system (tsleep in intr context) when ohcidebug > 5. ugen.c - create the devices for endpoints with make_dev. uhub.c - change from using usbdebug to uhubdebug - add more debugging statements
1999-11-28 21:01:06 +00:00
ohci_abort_xfer_end(xfer);
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
}
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
#else
delay(1000);
ohci_abort_xfer_end(xfer);
#endif
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
}
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_abort_xfer_end(void *v)
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
{
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
usbd_xfer_handle xfer = v;
struct ohci_pipe *opipe = (struct ohci_pipe *)xfer->pipe;
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
ohci_softc_t *sc = (ohci_softc_t *)opipe->pipe.device->bus;
ohci_soft_ed_t *sed;
ohci_soft_td_t *p, *n;
int s;
s = splusb();
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
p = xfer->hcpriv;
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
#ifdef DIAGNOSTIC
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
if (p == NULL) {
Synchronisation with NetBSD 1999/11/27 - more req[uest]->xfer changes. - get the corresponding NetBSD Id's right ohci.c - move untimeout above print statement - remove usb_delay that panics the system (tsleep in intr context) when ohcidebug > 5. ugen.c - create the devices for endpoints with make_dev. uhub.c - change from using usbdebug to uhubdebug - add more debugging statements
1999-11-28 21:01:06 +00:00
printf("ohci_abort_xfer: hcpriv==0\n");
Move the defines to the right location. Add a DIAGNOSTIC when closing interrupt pipes. Insert splx(s) which were left out with iso pipes (non-functional yet) and in a DIAGNOSTIC.
2000-01-27 23:25:58 +00:00
splx(s);
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
return;
}
#endif
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
for (; p->xfer == xfer; p = n) {
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
n = p->nexttd;
ohci_free_std(sc, p);
}
sed = opipe->sed;
Synchronisation with NetBSD 1999/11/27 - more req[uest]->xfer changes. - get the corresponding NetBSD Id's right ohci.c - move untimeout above print statement - remove usb_delay that panics the system (tsleep in intr context) when ohcidebug > 5. ugen.c - create the devices for endpoints with make_dev. uhub.c - change from using usbdebug to uhubdebug - add more debugging statements
1999-11-28 21:01:06 +00:00
DPRINTFN(2,("ohci_abort_xfer: set hd=%x, tl=%x\n",
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
(int)p->physaddr, (int)le32toh(sed->ed.ed_tailp)));
sed->ed.ed_headp = htole32(p->physaddr); /* unlink TDs */
sed->ed.ed_flags &= htole32(~OHCI_ED_SKIP); /* remove hardware skip */
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
usb_transfer_complete(xfer);
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
splx(s);
}
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
/*
* Data structures and routines to emulate the root hub.
*/
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static usb_device_descriptor_t ohci_devd = {
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
USB_DEVICE_DESCRIPTOR_SIZE,
UDESC_DEVICE, /* type */
{0x00, 0x01}, /* USB version */
Sync with NetBSD: date: 2000/02/29 21:37:01; author: augustss; state: Exp; Distinguish between device and interface classes. (I finally found a document that said that they were different.)
2002-01-02 18:28:45 +00:00
UDCLASS_HUB, /* class */
UDSUBCLASS_HUB, /* subclass */
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
0, /* protocol */
64, /* max packet */
{0},{0},{0x00,0x01}, /* device id */
1,2,0, /* string indicies */
1 /* # of configurations */
};
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static usb_config_descriptor_t ohci_confd = {
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
USB_CONFIG_DESCRIPTOR_SIZE,
UDESC_CONFIG,
{USB_CONFIG_DESCRIPTOR_SIZE +
USB_INTERFACE_DESCRIPTOR_SIZE +
USB_ENDPOINT_DESCRIPTOR_SIZE},
1,
1,
0,
UC_SELF_POWERED,
0 /* max power */
};
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static usb_interface_descriptor_t ohci_ifcd = {
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
USB_INTERFACE_DESCRIPTOR_SIZE,
UDESC_INTERFACE,
0,
0,
1,
Sync with NetBSD: date: 2000/02/29 21:37:01; author: augustss; state: Exp; Distinguish between device and interface classes. (I finally found a document that said that they were different.)
2002-01-02 18:28:45 +00:00
UICLASS_HUB,
UISUBCLASS_HUB,
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
0,
0
};
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static usb_endpoint_descriptor_t ohci_endpd = {
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
USB_ENDPOINT_DESCRIPTOR_SIZE,
UDESC_ENDPOINT,
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
UE_DIR_IN | OHCI_INTR_ENDPT,
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
UE_INTERRUPT,
{8, 0}, /* max packet */
255
};
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static usb_hub_descriptor_t ohci_hubd = {
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
USB_HUB_DESCRIPTOR_SIZE,
UDESC_HUB,
0,
{0,0},
0,
0,
{0},
};
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static int
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_str(usb_string_descriptor_t *p, int l, const char *s)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
int i;
if (l == 0)
return (0);
p->bLength = 2 * strlen(s) + 2;
if (l == 1)
return (1);
p->bDescriptorType = UDESC_STRING;
l -= 2;
for (i = 0; s[i] && l > 1; i++, l -= 2)
USETW2(p->bString[i], 0, s[i]);
return (2*i+2);
}
/*
* Simulate a hardware hub by handling all the necessary requests.
*/
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static usbd_status
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_root_ctrl_transfer(usbd_xfer_handle xfer)
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
{
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
usbd_status err;
/* Insert last in queue. */
err = usb_insert_transfer(xfer);
if (err)
return (err);
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
/* Pipe isn't running, start first */
return (ohci_root_ctrl_start(SIMPLEQ_FIRST(&xfer->pipe->queue)));
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
}
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static usbd_status
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_root_ctrl_start(usbd_xfer_handle xfer)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
ohci_softc_t *sc = (ohci_softc_t *)xfer->pipe->device->bus;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
usb_device_request_t *req;
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
void *buf = NULL;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
int port, i;
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
int s, len, value, index, l, totlen = 0;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
usb_port_status_t ps;
usb_hub_descriptor_t hubd;
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
usbd_status err;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
u_int32_t v;
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
if (sc->sc_dying)
return (USBD_IOERROR);
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
#ifdef DIAGNOSTIC
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
if (!(xfer->rqflags & URQ_REQUEST))
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
/* XXX panic */
return (USBD_INVAL);
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
#endif
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
req = &xfer->request;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
DPRINTFN(4,("ohci_root_ctrl_control type=0x%02x request=%02x\n",
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
req->bmRequestType, req->bRequest));
len = UGETW(req->wLength);
value = UGETW(req->wValue);
index = UGETW(req->wIndex);
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
if (len != 0)
Correctly handle the conversion from virtual to physical addresses. The problem was basically (for offset > 4096): vtophys(addr) + offset != vtophys(addr + offset) Also, use TD's with a maximum size of 4k instead of 8kb for OHCI controllers. This problem occurs in drivers that use large transfer sizes: umass, host2host and ethernet with jumbo frames.
2000-02-10 18:50:19 +00:00
buf = KERNADDR(&xfer->dmabuf, 0);
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
#define C(x,y) ((x) | ((y) << 8))
switch(C(req->bRequest, req->bmRequestType)) {
case C(UR_CLEAR_FEATURE, UT_WRITE_DEVICE):
case C(UR_CLEAR_FEATURE, UT_WRITE_INTERFACE):
case C(UR_CLEAR_FEATURE, UT_WRITE_ENDPOINT):
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
/*
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
* DEVICE_REMOTE_WAKEUP and ENDPOINT_HALT are no-ops
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
* for the integrated root hub.
*/
break;
case C(UR_GET_CONFIG, UT_READ_DEVICE):
if (len > 0) {
*(u_int8_t *)buf = sc->sc_conf;
totlen = 1;
}
break;
case C(UR_GET_DESCRIPTOR, UT_READ_DEVICE):
DPRINTFN(8,("ohci_root_ctrl_control wValue=0x%04x\n", value));
switch(value >> 8) {
case UDESC_DEVICE:
if ((value & 0xff) != 0) {
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
err = USBD_IOERROR;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
goto ret;
}
totlen = l = min(len, USB_DEVICE_DESCRIPTOR_SIZE);
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
USETW(ohci_devd.idVendor, sc->sc_id_vendor);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
memcpy(buf, &ohci_devd, l);
break;
case UDESC_CONFIG:
if ((value & 0xff) != 0) {
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
err = USBD_IOERROR;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
goto ret;
}
totlen = l = min(len, USB_CONFIG_DESCRIPTOR_SIZE);
memcpy(buf, &ohci_confd, l);
buf = (char *)buf + l;
len -= l;
l = min(len, USB_INTERFACE_DESCRIPTOR_SIZE);
totlen += l;
memcpy(buf, &ohci_ifcd, l);
buf = (char *)buf + l;
len -= l;
l = min(len, USB_ENDPOINT_DESCRIPTOR_SIZE);
totlen += l;
memcpy(buf, &ohci_endpd, l);
break;
case UDESC_STRING:
if (len == 0)
break;
*(u_int8_t *)buf = 0;
totlen = 1;
switch (value & 0xff) {
case 1: /* Vendor */
totlen = ohci_str(buf, len, sc->sc_vendor);
break;
case 2: /* Product */
totlen = ohci_str(buf, len, "OHCI root hub");
break;
}
break;
default:
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
err = USBD_IOERROR;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
goto ret;
}
break;
case C(UR_GET_INTERFACE, UT_READ_INTERFACE):
if (len > 0) {
*(u_int8_t *)buf = 0;
totlen = 1;
}
break;
case C(UR_GET_STATUS, UT_READ_DEVICE):
if (len > 1) {
USETW(((usb_status_t *)buf)->wStatus,UDS_SELF_POWERED);
totlen = 2;
}
break;
case C(UR_GET_STATUS, UT_READ_INTERFACE):
case C(UR_GET_STATUS, UT_READ_ENDPOINT):
if (len > 1) {
USETW(((usb_status_t *)buf)->wStatus, 0);
totlen = 2;
}
break;
case C(UR_SET_ADDRESS, UT_WRITE_DEVICE):
if (value >= USB_MAX_DEVICES) {
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
err = USBD_IOERROR;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
goto ret;
}
sc->sc_addr = value;
break;
case C(UR_SET_CONFIG, UT_WRITE_DEVICE):
if (value != 0 && value != 1) {
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
err = USBD_IOERROR;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
goto ret;
}
sc->sc_conf = value;
break;
case C(UR_SET_DESCRIPTOR, UT_WRITE_DEVICE):
break;
case C(UR_SET_FEATURE, UT_WRITE_DEVICE):
case C(UR_SET_FEATURE, UT_WRITE_INTERFACE):
case C(UR_SET_FEATURE, UT_WRITE_ENDPOINT):
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
err = USBD_IOERROR;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
goto ret;
case C(UR_SET_INTERFACE, UT_WRITE_INTERFACE):
break;
case C(UR_SYNCH_FRAME, UT_WRITE_ENDPOINT):
break;
/* Hub requests */
case C(UR_CLEAR_FEATURE, UT_WRITE_CLASS_DEVICE):
break;
case C(UR_CLEAR_FEATURE, UT_WRITE_CLASS_OTHER):
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
DPRINTFN(8, ("ohci_root_ctrl_control: UR_CLEAR_PORT_FEATURE "
"port=%d feature=%d\n",
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
index, value));
if (index < 1 || index > sc->sc_noport) {
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
err = USBD_IOERROR;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
goto ret;
}
port = OHCI_RH_PORT_STATUS(index);
switch(value) {
case UHF_PORT_ENABLE:
OWRITE4(sc, port, UPS_CURRENT_CONNECT_STATUS);
break;
case UHF_PORT_SUSPEND:
OWRITE4(sc, port, UPS_OVERCURRENT_INDICATOR);
break;
case UHF_PORT_POWER:
Merge from NetBSD: Pave the way for USB2, by replacing 'lowspeed' with 'speed', so that it can take the values USB_SPEED_LOW, USB_SPEED_FULL or in time USB_SPEED_HIGH.
2002-02-16 00:51:26 +00:00
/* Yes, writing to the LOW_SPEED bit clears power. */
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
OWRITE4(sc, port, UPS_LOW_SPEED);
break;
case UHF_C_PORT_CONNECTION:
OWRITE4(sc, port, UPS_C_CONNECT_STATUS << 16);
break;
case UHF_C_PORT_ENABLE:
OWRITE4(sc, port, UPS_C_PORT_ENABLED << 16);
break;
case UHF_C_PORT_SUSPEND:
OWRITE4(sc, port, UPS_C_SUSPEND << 16);
break;
case UHF_C_PORT_OVER_CURRENT:
OWRITE4(sc, port, UPS_C_OVERCURRENT_INDICATOR << 16);
break;
case UHF_C_PORT_RESET:
OWRITE4(sc, port, UPS_C_PORT_RESET << 16);
break;
default:
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
err = USBD_IOERROR;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
goto ret;
}
switch(value) {
case UHF_C_PORT_CONNECTION:
case UHF_C_PORT_ENABLE:
case UHF_C_PORT_SUSPEND:
case UHF_C_PORT_OVER_CURRENT:
case UHF_C_PORT_RESET:
/* Enable RHSC interrupt if condition is cleared. */
if ((OREAD4(sc, port) >> 16) == 0)
ohci_rhsc_able(sc, 1);
break;
default:
break;
}
break;
case C(UR_GET_DESCRIPTOR, UT_READ_CLASS_DEVICE):
if (value != 0) {
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
err = USBD_IOERROR;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
goto ret;
}
v = OREAD4(sc, OHCI_RH_DESCRIPTOR_A);
hubd = ohci_hubd;
hubd.bNbrPorts = sc->sc_noport;
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
USETW(hubd.wHubCharacteristics,
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
(v & OHCI_NPS ? UHD_PWR_NO_SWITCH :
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
v & OHCI_PSM ? UHD_PWR_GANGED : UHD_PWR_INDIVIDUAL)
/* XXX overcurrent */
);
hubd.bPwrOn2PwrGood = OHCI_GET_POTPGT(v);
v = OREAD4(sc, OHCI_RH_DESCRIPTOR_B);
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
for (i = 0, l = sc->sc_noport; l > 0; i++, l -= 8, v >>= 8)
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
hubd.DeviceRemovable[i++] = (u_int8_t)v;
hubd.bDescLength = USB_HUB_DESCRIPTOR_SIZE + i;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
l = min(len, hubd.bDescLength);
totlen = l;
memcpy(buf, &hubd, l);
break;
case C(UR_GET_STATUS, UT_READ_CLASS_DEVICE):
if (len != 4) {
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
err = USBD_IOERROR;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
goto ret;
}
memset(buf, 0, len); /* ? XXX */
totlen = len;
break;
case C(UR_GET_STATUS, UT_READ_CLASS_OTHER):
DPRINTFN(8,("ohci_root_ctrl_transfer: get port status i=%d\n",
index));
if (index < 1 || index > sc->sc_noport) {
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
err = USBD_IOERROR;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
goto ret;
}
if (len != 4) {
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
err = USBD_IOERROR;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
goto ret;
}
v = OREAD4(sc, OHCI_RH_PORT_STATUS(index));
DPRINTFN(8,("ohci_root_ctrl_transfer: port status=0x%04x\n",
v));
USETW(ps.wPortStatus, v);
USETW(ps.wPortChange, v >> 16);
l = min(len, sizeof ps);
memcpy(buf, &ps, l);
totlen = l;
break;
case C(UR_SET_DESCRIPTOR, UT_WRITE_CLASS_DEVICE):
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
err = USBD_IOERROR;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
goto ret;
case C(UR_SET_FEATURE, UT_WRITE_CLASS_DEVICE):
break;
case C(UR_SET_FEATURE, UT_WRITE_CLASS_OTHER):
if (index < 1 || index > sc->sc_noport) {
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
err = USBD_IOERROR;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
goto ret;
}
port = OHCI_RH_PORT_STATUS(index);
switch(value) {
case UHF_PORT_ENABLE:
OWRITE4(sc, port, UPS_PORT_ENABLED);
break;
case UHF_PORT_SUSPEND:
OWRITE4(sc, port, UPS_SUSPEND);
break;
case UHF_PORT_RESET:
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
DPRINTFN(5,("ohci_root_ctrl_transfer: reset port %d\n",
index));
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
OWRITE4(sc, port, UPS_RESET);
for (i = 0; i < 10; i++) {
Merge from NetBSD: date: 2001/11/20 16:08:10; author: augustss; state: Exp; Use longer reset for root hubs (as told in the spec).
2002-01-03 00:37:56 +00:00
usb_delay_ms(&sc->sc_bus,
USB_PORT_ROOT_RESET_DELAY);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
if ((OREAD4(sc, port) & UPS_RESET) == 0)
break;
}
DPRINTFN(8,("ohci port %d reset, status = 0x%04x\n",
index, OREAD4(sc, port)));
break;
case UHF_PORT_POWER:
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
DPRINTFN(2,("ohci_root_ctrl_transfer: set port power "
"%d\n", index));
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
OWRITE4(sc, port, UPS_PORT_POWER);
break;
default:
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
err = USBD_IOERROR;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
goto ret;
}
break;
default:
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
err = USBD_IOERROR;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
goto ret;
}
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
xfer->actlen = totlen;
err = USBD_NORMAL_COMPLETION;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
ret:
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
xfer->status = err;
s = splusb();
usb_transfer_complete(xfer);
splx(s);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
return (USBD_IN_PROGRESS);
}
/* Abort a root control request. */
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_root_ctrl_abort(usbd_xfer_handle xfer)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
/* Nothing to do, all transfers are synchronous. */
}
/* Close the root pipe. */
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_root_ctrl_close(usbd_pipe_handle pipe)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
DPRINTF(("ohci_root_ctrl_close\n"));
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
/* Nothing to do. */
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static usbd_status
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_root_intr_transfer(usbd_xfer_handle xfer)
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
{
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
usbd_status err;
/* Insert last in queue. */
err = usb_insert_transfer(xfer);
if (err)
return (err);
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
/* Pipe isn't running, start first */
return (ohci_root_intr_start(SIMPLEQ_FIRST(&xfer->pipe->queue)));
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
}
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static usbd_status
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_root_intr_start(usbd_xfer_handle xfer)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
usbd_pipe_handle pipe = xfer->pipe;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
ohci_softc_t *sc = (ohci_softc_t *)pipe->device->bus;
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
if (sc->sc_dying)
return (USBD_IOERROR);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
sc->sc_intrxfer = xfer;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
return (USBD_IN_PROGRESS);
}
/* Abort a root interrupt request. */
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_root_intr_abort(usbd_xfer_handle xfer)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
int s;
if (xfer->pipe->intrxfer == xfer) {
DPRINTF(("ohci_root_intr_abort: remove\n"));
xfer->pipe->intrxfer = NULL;
}
xfer->status = USBD_CANCELLED;
s = splusb();
usb_transfer_complete(xfer);
splx(s);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
/* Close the root pipe. */
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_root_intr_close(usbd_pipe_handle pipe)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
ohci_softc_t *sc = (ohci_softc_t *)pipe->device->bus;
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
DPRINTF(("ohci_root_intr_close\n"));
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
sc->sc_intrxfer = NULL;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
/************************/
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static usbd_status
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_device_ctrl_transfer(usbd_xfer_handle xfer)
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
{
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
usbd_status err;
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
/* Insert last in queue. */
err = usb_insert_transfer(xfer);
if (err)
return (err);
/* Pipe isn't running, start first */
return (ohci_device_ctrl_start(SIMPLEQ_FIRST(&xfer->pipe->queue)));
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
}
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static usbd_status
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_device_ctrl_start(usbd_xfer_handle xfer)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
ohci_softc_t *sc = (ohci_softc_t *)xfer->pipe->device->bus;
usbd_status err;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
if (sc->sc_dying)
return (USBD_IOERROR);
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
#ifdef DIAGNOSTIC
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
if (!(xfer->rqflags & URQ_REQUEST)) {
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
/* XXX panic */
printf("ohci_device_ctrl_transfer: not a request\n");
return (USBD_INVAL);
}
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
#endif
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
err = ohci_device_request(xfer);
if (err)
return (err);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
if (sc->sc_bus.use_polling)
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
ohci_waitintr(sc, xfer);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
return (USBD_IN_PROGRESS);
}
/* Abort a device control request. */
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_device_ctrl_abort(usbd_xfer_handle xfer)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
DPRINTF(("ohci_device_ctrl_abort: xfer=%p\n", xfer));
Synchronisation with NetBSD 1999/11/27 - more req[uest]->xfer changes. - get the corresponding NetBSD Id's right ohci.c - move untimeout above print statement - remove usb_delay that panics the system (tsleep in intr context) when ohcidebug > 5. ugen.c - create the devices for endpoints with make_dev. uhub.c - change from using usbdebug to uhubdebug - add more debugging statements
1999-11-28 21:01:06 +00:00
ohci_abort_xfer(xfer, USBD_CANCELLED);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
/* Close a device control pipe. */
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_device_ctrl_close(usbd_pipe_handle pipe)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
struct ohci_pipe *opipe = (struct ohci_pipe *)pipe;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
ohci_softc_t *sc = (ohci_softc_t *)pipe->device->bus;
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
DPRINTF(("ohci_device_ctrl_close: pipe=%p\n", pipe));
ohci_close_pipe(pipe, sc->sc_ctrl_head);
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
ohci_free_std(sc, opipe->tail.td);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
/************************/
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_device_clear_toggle(usbd_pipe_handle pipe)
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
{
struct ohci_pipe *opipe = (struct ohci_pipe *)pipe;
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
opipe->sed->ed.ed_headp &= htole32(~OHCI_TOGGLECARRY);
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
}
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_noop(usbd_pipe_handle pipe)
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
{
}
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static usbd_status
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_device_bulk_transfer(usbd_xfer_handle xfer)
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
{
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
usbd_status err;
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
/* Insert last in queue. */
err = usb_insert_transfer(xfer);
if (err)
return (err);
/* Pipe isn't running, start first */
return (ohci_device_bulk_start(SIMPLEQ_FIRST(&xfer->pipe->queue)));
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
}
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static usbd_status
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_device_bulk_start(usbd_xfer_handle xfer)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
struct ohci_pipe *opipe = (struct ohci_pipe *)xfer->pipe;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
usbd_device_handle dev = opipe->pipe.device;
ohci_softc_t *sc = (ohci_softc_t *)dev->bus;
int addr = dev->address;
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
ohci_soft_td_t *data, *tail, *tdp;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
ohci_soft_ed_t *sed;
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
int s, len, isread, endpt;
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
usbd_status err;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
if (sc->sc_dying)
return (USBD_IOERROR);
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
#ifdef DIAGNOSTIC
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
if (xfer->rqflags & URQ_REQUEST) {
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
/* XXX panic */
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
printf("ohci_device_bulk_start: a request\n");
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
return (USBD_INVAL);
}
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
#endif
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
len = xfer->length;
endpt = xfer->pipe->endpoint->edesc->bEndpointAddress;
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
isread = UE_GET_DIR(endpt) == UE_DIR_IN;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
sed = opipe->sed;
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
DPRINTFN(4,("ohci_device_bulk_start: xfer=%p len=%d isread=%d "
"flags=%d endpt=%d\n", xfer, len, isread, xfer->flags,
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
endpt));
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
opipe->u.bulk.isread = isread;
opipe->u.bulk.length = len;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
/* Update device address */
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
sed->ed.ed_flags = htole32(
(le32toh(sed->ed.ed_flags) & ~OHCI_ED_ADDRMASK) |
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
OHCI_ED_SET_FA(addr));
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
/* Allocate a chain of new TDs (including a new tail). */
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
data = opipe->tail.td;
err = ohci_alloc_std_chain(opipe, sc, len, isread, xfer->flags,
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
&xfer->dmabuf, data, &tail);
if (err)
return (err);
tail->xfer = NULL;
xfer->hcpriv = data;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
DPRINTFN(4,("ohci_device_bulk_start: ed_flags=0x%08x td_flags=0x%08x "
"td_cbp=0x%08x td_be=0x%08x\n",
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
(int)le32toh(sed->ed.ed_flags),
(int)le32toh(data->td.td_flags),
(int)le32toh(data->td.td_cbp),
(int)le32toh(data->td.td_be)));
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
#ifdef OHCI_DEBUG
if (ohcidebug > 4) {
ohci_dump_ed(sed);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
ohci_dump_tds(data);
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
}
#endif
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
/* Insert ED in schedule */
s = splusb();
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
for (tdp = data; tdp != tail; tdp = tdp->nexttd) {
tdp->xfer = xfer;
}
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
sed->ed.ed_tailp = htole32(tail->physaddr);
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
opipe->tail.td = tail;
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
sed->ed.ed_flags &= htole32(~OHCI_ED_SKIP);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
OWRITE4(sc, OHCI_COMMAND_STATUS, OHCI_BLF);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
if (xfer->timeout && !sc->sc_bus.use_polling) {
Huge merge from NetBSD: usbdi.c (1.61): =================================================================== revision 1.61 date: 2000/01/31 20:13:07; author: augustss; lines: +20 -4 Change the way the HC done method is invoked a little. =================================================================== usbdi.c (1.65): =================================================================== revision 1.65 date: 2000/03/08 15:34:10; author: augustss; lines: +4 -2 Get the status right when a polled transfer times out. =================================================================== ohci.c (1.79), uhci.c (1.89), uhcivar.h (1.24), usb_port.h (1.22), usbdivar.h (1.48): =================================================================== date: 2000/03/23 07:01:46; author: thorpej; New callout mechanism with two major improvements over the old timeout()/untimeout() API: - Clients supply callout handle storage, thus eliminating problems of resource allocation. - Insertion and removal of callouts is constant time, important as this facility is used quite a lot in the kernel. The old timeout()/untimeout() API has been removed from the kernel. =================================================================== uhci.c (1.80), usbdi.c (1.66): =================================================================== date: 2000/03/23 18:59:10; author: thorpej; Shake out some bugs from the callout changes. =================================================================== ohci.c (1.80), uhci.c (1.91), uhcivar.h (1.25), usb_port.h (1.23), usbdi.c (1.67), usbdivar.h (1.49): =================================================================== date: 2000/03/24 22:03:30; author: augustss; Some cleanup and renaming of the callouts used in USB drivers. =================================================================== uhci.c (1.92), uhcivar.h (1.26): =================================================================== date: 2000/03/24 22:57:58; author: augustss; Two major changes: Make each xfer have its own intr_info. This is necessary if we want to queue multiple xfers on an endpoint. This should get rid of the (mostly harmless) DIAGNOSTICs about intr_infos (not) being done. Change (again!) how xfers are aborted. Aborting a TD is a nightmare on the braindead UHCI controller. (Unless you stop the HC, thereby losing isoc traffic.) Hopefully I got it right this time. =================================================================== usbdivar.h (1.50): =================================================================== revision 1.50 date: 2000/03/25 00:10:19; author: augustss; lines: +4 -2 GC an unsued field and add some DIAGNOSTIC in xfer. =================================================================== ums.c: Use the callout functions instead of the timeout ones. uhci.c (1.93): =================================================================== revision 1.93 date: 2000/03/25 00:11:21; author: augustss; lines: +26 -1 Add more DIAGNOSTIC when aborting isoc. =================================================================== uhci.c (1.94), usbdivar.h (1.51): =================================================================== date: 2000/03/25 07:13:05; author: augustss; More DIAGNOSTIC. Initialize a callout handle I forgot. =================================================================== uhci.c (1.95): =================================================================== revision 1.95 date: 2000/03/25 07:23:12; author: augustss; Exp; lines: +24 -7 Improve uhci_dump_ii(). =================================================================== ohci.c (1.81), uhci.c (1.96), uhcivar.h (1.27), usb_subr.c (1.68), usbdi.c (1.68), usbdivar.h (1.52): =================================================================== date: 2000/03/25 18:02:33; author: augustss; Rename and move around callout handles to make it more sane. Add some DIAGNOSTIC. Fix buglet in isoc abort on UHCI. =================================================================== uhci.c (1.98): =================================================================== revision 1.98 date: 2000/03/27 07:39:48; author: augustss; lines: +12 -4 Make it compile without DIAGNOSTIC. =================================================================== uhci.c (1.99): =================================================================== revision 1.99 date: 2000/03/27 08:01:09; author: augustss; lines: +1 -5 Remove some debug nonsense. =================================================================== uhci.c (1.100): =================================================================== revision 1.100 date: 2000/03/27 09:41:36; author: augustss; lines: +13 -3 Don't mess with QH in bulk abort for the moment. =================================================================== uhci.c (1.102): =================================================================== revision 1.102 date: 2000/03/27 22:42:57; author: augustss; lines: +66 -26 Be a little more careful when aborting. Preallocate some TDs for large buffers. =================================================================== uhci.c (1.103): =================================================================== date: 2000/03/28 09:47:10; author: augustss; lines: +11 -1 Another patch for xfer abort... XXX The current xfer queueing and aborting semantics should really XXX be changed. It cannot be implemented in a sane way on UHCI. XXX One day when I have lots of time I'll redesign it... =================================================================== uhci.c (1.104): Correct a debug message. uhci.c (1.105): Be more defensive in a DIAGNOSTIC test. uhci.c (1.106): =================================================================== revision 1.106 date: 2000/03/29 01:49:13; author: augustss; lines: +14 -309 *SIGH* Revert back to the old method of aborting xfers. I had tested the new stuff for two months now, but as soon as I commited it the problems started to appear. Murphy, no doubt... =================================================================== usb_subr.c (1.70), usbdi.c (1.71), usbdivar.h (1.53): =================================================================== revision 1.70 date: 2000/03/29 01:45:20; author: augustss; lines: +2 -1 Do not accept new xfers for queuing while a pipe is aborting. ===================================================================
2002-03-16 12:06:01 +00:00
usb_callout(xfer->timeout_handle, MS_TO_TICKS(xfer->timeout),
ohci_timeout, xfer);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
#if 0
/* This goes wrong if we are too slow. */
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
if (ohcidebug > 5) {
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
usb_delay_ms(&sc->sc_bus, 5);
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
DPRINTF(("ohci_device_intr_transfer: status=%x\n",
OREAD4(sc, OHCI_COMMAND_STATUS)));
ohci_dump_ed(sed);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
ohci_dump_tds(data);
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
}
#endif
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
splx(s);
return (USBD_IN_PROGRESS);
}
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_device_bulk_abort(usbd_xfer_handle xfer)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
DPRINTF(("ohci_device_bulk_abort: xfer=%p\n", xfer));
Synchronisation with NetBSD 1999/11/27 - more req[uest]->xfer changes. - get the corresponding NetBSD Id's right ohci.c - move untimeout above print statement - remove usb_delay that panics the system (tsleep in intr context) when ohcidebug > 5. ugen.c - create the devices for endpoints with make_dev. uhub.c - change from using usbdebug to uhubdebug - add more debugging statements
1999-11-28 21:01:06 +00:00
ohci_abort_xfer(xfer, USBD_CANCELLED);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
/*
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
* Close a device bulk pipe.
*/
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_device_bulk_close(usbd_pipe_handle pipe)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
struct ohci_pipe *opipe = (struct ohci_pipe *)pipe;
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
ohci_softc_t *sc = (ohci_softc_t *)pipe->device->bus;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
DPRINTF(("ohci_device_bulk_close: pipe=%p\n", pipe));
ohci_close_pipe(pipe, sc->sc_bulk_head);
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
ohci_free_std(sc, opipe->tail.td);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
/************************/
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static usbd_status
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_device_intr_transfer(usbd_xfer_handle xfer)
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
{
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
usbd_status err;
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
/* Insert last in queue. */
err = usb_insert_transfer(xfer);
if (err)
return (err);
/* Pipe isn't running, start first */
return (ohci_device_intr_start(SIMPLEQ_FIRST(&xfer->pipe->queue)));
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
}
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static usbd_status
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_device_intr_start(usbd_xfer_handle xfer)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
struct ohci_pipe *opipe = (struct ohci_pipe *)xfer->pipe;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
usbd_device_handle dev = opipe->pipe.device;
ohci_softc_t *sc = (ohci_softc_t *)dev->bus;
ohci_soft_ed_t *sed = opipe->sed;
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
ohci_soft_td_t *data, *tail;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
int len;
int s;
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
if (sc->sc_dying)
return (USBD_IOERROR);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
DPRINTFN(3, ("ohci_device_intr_transfer: xfer=%p len=%d "
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
"flags=%d priv=%p\n",
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
xfer, xfer->length, xfer->flags, xfer->priv));
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
#ifdef DIAGNOSTIC
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
if (xfer->rqflags & URQ_REQUEST)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
panic("ohci_device_intr_transfer: a request\n");
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
#endif
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
len = xfer->length;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
data = opipe->tail.td;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
tail = ohci_alloc_std(sc);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
if (tail == NULL)
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
return (USBD_NOMEM);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
tail->xfer = NULL;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
data->td.td_flags = htole32(
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
OHCI_TD_IN | OHCI_TD_NOCC |
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
OHCI_TD_SET_DI(1) | OHCI_TD_TOGGLE_CARRY);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
if (xfer->flags & USBD_SHORT_XFER_OK)
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
data->td.td_flags |= htole32(OHCI_TD_R);
data->td.td_cbp = htole32(DMAADDR(&xfer->dmabuf, 0));
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
data->nexttd = tail;
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
data->td.td_nexttd = htole32(tail->physaddr);
data->td.td_be = htole32(le32toh(data->td.td_cbp) + len - 1);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
data->len = len;
data->xfer = xfer;
data->flags = OHCI_CALL_DONE | OHCI_ADD_LEN;
xfer->hcpriv = data;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Clean up debugging output
1999-05-03 23:11:41 +00:00
#ifdef OHCI_DEBUG
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
if (ohcidebug > 5) {
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
DPRINTF(("ohci_device_intr_transfer:\n"));
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
ohci_dump_ed(sed);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
ohci_dump_tds(data);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
#endif
/* Insert ED in schedule */
s = splusb();
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
sed->ed.ed_tailp = htole32(tail->physaddr);
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
opipe->tail.td = tail;
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
sed->ed.ed_flags &= htole32(~OHCI_ED_SKIP);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
Clean up debugging output
1999-05-03 23:11:41 +00:00
#if 0
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
/*
* This goes horribly wrong, printing thousands of descriptors,
* because false references are followed due to the fact that the
* TD is gone.
*/
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
if (ohcidebug > 5) {
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
usb_delay_ms(&sc->sc_bus, 5);
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
DPRINTF(("ohci_device_intr_transfer: status=%x\n",
OREAD4(sc, OHCI_COMMAND_STATUS)));
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
ohci_dump_ed(sed);
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
ohci_dump_tds(data);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
#endif
1) Bug fix: Interrupt transfers worked only for the first packet. The second packet was not seen as finished. 2) Textual changes
1999-01-13 00:50:02 +00:00
splx(s);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
return (USBD_IN_PROGRESS);
}
/* Abort a device control request. */
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_device_intr_abort(usbd_xfer_handle xfer)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
if (xfer->pipe->intrxfer == xfer) {
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
DPRINTF(("ohci_device_intr_abort: remove\n"));
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
xfer->pipe->intrxfer = NULL;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
Synchronisation with NetBSD 1999/11/27 - more req[uest]->xfer changes. - get the corresponding NetBSD Id's right ohci.c - move untimeout above print statement - remove usb_delay that panics the system (tsleep in intr context) when ohcidebug > 5. ugen.c - create the devices for endpoints with make_dev. uhub.c - change from using usbdebug to uhubdebug - add more debugging statements
1999-11-28 21:01:06 +00:00
ohci_abort_xfer(xfer, USBD_CANCELLED);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
}
/* Close a device interrupt pipe. */
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_device_intr_close(usbd_pipe_handle pipe)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
struct ohci_pipe *opipe = (struct ohci_pipe *)pipe;
ohci_softc_t *sc = (ohci_softc_t *)pipe->device->bus;
int nslots = opipe->u.intr.nslots;
int pos = opipe->u.intr.pos;
int j;
ohci_soft_ed_t *p, *sed = opipe->sed;
int s;
DPRINTFN(1,("ohci_device_intr_close: pipe=%p nslots=%d pos=%d\n",
pipe, nslots, pos));
s = splusb();
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
sed->ed.ed_flags |= htole32(OHCI_ED_SKIP);
Merge from NetBSD: ohci.c: -r1.69 to 1.71 ohcireg.h: -r1.14 Some of these deltas are based upon patches that we submitted back to NetBSD. They got manifested slightly differently though, so I've brought back those differences to bring our code bases closer together. The logs from the NetBSD version of ohci.c: revision 1.71 date: 2000/02/01 05:42:52; author: augustss; state: Exp; lines: +13 -2 Put some #ifdefs around power and shutdown hooks. ---------------------------- revision 1.70 date: 2000/01/31 22:35:13; author: augustss; state: Exp; lines: +7 -7 Rename TAILMASK to HEADMASK, since it really masks the head pointer. From FreeBSD. ---------------------------- revision 1.69 date: 2000/01/31 22:09:13; author: augustss; state: Exp; lines: +18 -14 Change where the has table for physical-to-virtual address translation is handled. Partly from FreeBSD.
2002-01-21 05:02:21 +00:00
if ((le32toh(sed->ed.ed_tailp) & OHCI_HEADMASK) !=
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
(le32toh(sed->ed.ed_headp) & OHCI_HEADMASK))
Major synchronisation with NetBSD USB code
1999-01-07 23:07:57 +00:00
usb_delay_ms(&sc->sc_bus, 2);
Move the defines to the right location. Add a DIAGNOSTIC when closing interrupt pipes. Insert splx(s) which were left out with iso pipes (non-functional yet) and in a DIAGNOSTIC.
2000-01-27 23:25:58 +00:00
#ifdef DIAGNOSTIC
Merge from NetBSD: ohci.c: -r1.69 to 1.71 ohcireg.h: -r1.14 Some of these deltas are based upon patches that we submitted back to NetBSD. They got manifested slightly differently though, so I've brought back those differences to bring our code bases closer together. The logs from the NetBSD version of ohci.c: revision 1.71 date: 2000/02/01 05:42:52; author: augustss; state: Exp; lines: +13 -2 Put some #ifdefs around power and shutdown hooks. ---------------------------- revision 1.70 date: 2000/01/31 22:35:13; author: augustss; state: Exp; lines: +7 -7 Rename TAILMASK to HEADMASK, since it really masks the head pointer. From FreeBSD. ---------------------------- revision 1.69 date: 2000/01/31 22:09:13; author: augustss; state: Exp; lines: +18 -14 Change where the has table for physical-to-virtual address translation is handled. Partly from FreeBSD.
2002-01-21 05:02:21 +00:00
if ((le32toh(sed->ed.ed_tailp) & OHCI_HEADMASK) !=
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
(le32toh(sed->ed.ed_headp) & OHCI_HEADMASK))
Move the defines to the right location. Add a DIAGNOSTIC when closing interrupt pipes. Insert splx(s) which were left out with iso pipes (non-functional yet) and in a DIAGNOSTIC.
2000-01-27 23:25:58 +00:00
panic("%s: Intr pipe %p still has TDs queued\n",
USBDEVNAME(sc->sc_bus.bdev), pipe);
#endif
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
for (p = sc->sc_eds[pos]; p && p->next != sed; p = p->next)
;
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
#ifdef DIAGNOSTIC
if (p == NULL)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
panic("ohci_device_intr_close: ED not found\n");
Synchronisation with NetBSD as of 1999/11/16: Cleaning up the code: - Declare many functions static - Change variable names to make them more self explanatory - Change usbd_request_handle -> usbd_xfer_handle - Syntactical changes - Remove some unused code - Other KNF changes Interrupt context handling - Change delay to usbd_delay_ms were possible (takes polling mode into account) - Change detection mechanism for interrupt context Add support for pre-allocation DMA-able memory by device driver Add preliminary support for isochronous to the UHCI driver (not for OHCI yet). usb.c, uhci.c, ohci.c - Initial attempt at detachable USB host controllers - Handle the use_polling flag with a lttle more care and only set it if we are cold booting. usb.c, uhci.c ohci.c, usbdi.c usbdi_util.c usb_subr.c - Make sure an aborted pipe is marked as not running. - Start queued request in the right order. - Insert some more DIAGNOSTIC sanity checks. - Remove (almost) unused definitions USBD_XFER_OUT and USBD_XFER_IN. usb.c, usb_subr.c - Add an event mechanism so that a userland process can watch devices come and go. ohci.c - Handle the case when a USB transfer is so long that it crosses two page (4K) boundaries. OHCI cannot do that with a single TD so we make a chain. ulpt.c - Use a bigger buffer when transferring data. - Pre-allocate the DMA buffer. This makes the driver slightly more efficient. - Comment out the GET_DEVICE_ID code, because for some unknown reason it causes printing to fail sometimes. usb.h - Add a macro to extract the isoc type. - Add a macro to check whether the routine has been entered after splusb and if not, complain. usbdi.c - Fix a glitch in dequeueing and aborting requests on interrupt pipes. - Add a flag in the request to determine if the data copying is done by the driver or the usbdi layer.
1999-11-17 22:33:51 +00:00
#endif
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
p->next = sed->next;
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
p->ed.ed_nexted = sed->ed.ed_nexted;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
splx(s);
for (j = 0; j < nslots; j++)
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
--sc->sc_bws[(pos * nslots + j) % OHCI_NO_INTRS];
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
ohci_free_std(sc, opipe->tail.td);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
ohci_free_sed(sc, opipe->sed);
}
OpenBSD has a broken debugger that does not grok static. Use a #define Static static that the OpenBSD folks can define it to be empty if they like.
2000-04-03 20:58:30 +00:00
Static usbd_status
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_device_setintr(ohci_softc_t *sc, struct ohci_pipe *opipe, int ival)
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
{
int i, j, s, best;
u_int npoll, slow, shigh, nslots;
u_int bestbw, bw;
ohci_soft_ed_t *hsed, *sed = opipe->sed;
DPRINTFN(2, ("ohci_setintr: pipe=%p\n", opipe));
if (ival == 0) {
printf("ohci_setintr: 0 interval\n");
return (USBD_INVAL);
}
npoll = OHCI_NO_INTRS;
while (npoll > ival)
npoll /= 2;
DPRINTFN(2, ("ohci_setintr: ival=%d npoll=%d\n", ival, npoll));
/*
* We now know which level in the tree the ED must go into.
* Figure out which slot has most bandwidth left over.
* Slots to examine:
* npoll
* 1 0
* 2 1 2
* 4 3 4 5 6
* 8 7 8 9 10 11 12 13 14
* N (N-1) .. (N-1+N-1)
*/
slow = npoll-1;
shigh = slow + npoll;
nslots = OHCI_NO_INTRS / npoll;
for (best = i = slow, bestbw = ~0; i < shigh; i++) {
bw = 0;
for (j = 0; j < nslots; j++)
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
bw += sc->sc_bws[(i * nslots + j) % OHCI_NO_INTRS];
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
if (bw < bestbw) {
best = i;
bestbw = bw;
}
}
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
DPRINTFN(2, ("ohci_setintr: best=%d(%d..%d) bestbw=%d\n",
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
best, slow, shigh, bestbw));
s = splusb();
hsed = sc->sc_eds[best];
sed->next = hsed->next;
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
sed->ed.ed_nexted = hsed->ed.ed_nexted;
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
hsed->next = sed;
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
hsed->ed.ed_nexted = htole32(sed->physaddr);
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
splx(s);
for (j = 0; j < nslots; j++)
Major synchronisation with the NetBSD USB stack: - Some cleanup and improvements in the uhci and ohci drivers - Support for plugging and unplugging devices improved - Now available is bulk transport over OHCI controllers - Resume and suspend have been temporarily been disabled again. Proper support for it is available in the uhci.c and ohci.c files but I have not yet spent the brain cycles to use it. - OpenBSD now uses the USB stack as well - Add FreeBSD tags
1999-10-07 19:26:38 +00:00
++sc->sc_bws[(best * nslots + j) % OHCI_NO_INTRS];
Preliminary support for OHCI motherboards
1998-12-09 23:28:43 +00:00
opipe->u.intr.nslots = nslots;
opipe->u.intr.pos = best;
DPRINTFN(5, ("ohci_setintr: returns %p\n", opipe));
return (USBD_NORMAL_COMPLETION);
}
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
/***********************/
usbd_status
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_device_isoc_transfer(usbd_xfer_handle xfer)
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
{
usbd_status err;
DPRINTFN(5,("ohci_device_isoc_transfer: xfer=%p\n", xfer));
/* Put it on our queue, */
err = usb_insert_transfer(xfer);
/* bail out on error, */
if (err && err != USBD_IN_PROGRESS)
return (err);
/* XXX should check inuse here */
/* insert into schedule, */
ohci_device_isoc_enter(xfer);
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
/* and start if the pipe wasn't running */
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
if (!err)
ohci_device_isoc_start(SIMPLEQ_FIRST(&xfer->pipe->queue));
return (err);
}
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_device_isoc_enter(usbd_xfer_handle xfer)
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
{
struct ohci_pipe *opipe = (struct ohci_pipe *)xfer->pipe;
usbd_device_handle dev = opipe->pipe.device;
ohci_softc_t *sc = (ohci_softc_t *)dev->bus;
ohci_soft_ed_t *sed = opipe->sed;
struct iso *iso = &opipe->u.iso;
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_soft_itd_t *sitd, *nsitd;
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
ohci_physaddr_t buf, offs, noffs, bp0;
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
int i, ncur, nframes;
int s;
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
DPRINTFN(1,("ohci_device_isoc_enter: used=%d next=%d xfer=%p "
"nframes=%d\n",
iso->inuse, iso->next, xfer, xfer->nframes));
if (sc->sc_dying)
return;
if (iso->next == -1) {
/* Not in use yet, schedule it a few frames ahead. */
iso->next = le32toh(sc->sc_hcca->hcca_frame_number) + 5;
DPRINTFN(2,("ohci_device_isoc_enter: start next=%d\n",
iso->next));
}
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
sitd = opipe->tail.itd;
Correctly handle the conversion from virtual to physical addresses. The problem was basically (for offset > 4096): vtophys(addr) + offset != vtophys(addr + offset) Also, use TD's with a maximum size of 4k instead of 8kb for OHCI controllers. This problem occurs in drivers that use large transfer sizes: umass, host2host and ethernet with jumbo frames.
2000-02-10 18:50:19 +00:00
buf = DMAADDR(&xfer->dmabuf, 0);
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
bp0 = OHCI_PAGE(buf);
offs = OHCI_PAGE_OFFSET(buf);
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
nframes = xfer->nframes;
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
xfer->hcpriv = sitd;
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
for (i = ncur = 0; i < nframes; i++, ncur++) {
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
noffs = offs + xfer->frlengths[i];
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
if (ncur == OHCI_ITD_NOFFSET || /* all offsets used */
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
OHCI_PAGE(buf + noffs) > bp0 + OHCI_PAGE_SIZE) { /*
too many page crossings */
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
/* Allocate next ITD */
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
nsitd = ohci_alloc_sitd(sc);
if (nsitd == NULL) {
/* XXX what now? */
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
printf("%s: isoc TD alloc failed\n",
USBDEVNAME(sc->sc_bus.bdev));
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
return;
}
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
/* Fill current ITD */
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
sitd->itd.itd_flags = htole32(
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
OHCI_ITD_NOCC |
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
OHCI_ITD_SET_SF(iso->next) |
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
OHCI_ITD_SET_DI(6) | /* delay intr a little */
OHCI_ITD_SET_FC(ncur));
sitd->itd.itd_bp0 = htole32(bp0);
sitd->nextitd = nsitd;
sitd->itd.itd_nextitd = htole32(nsitd->physaddr);
sitd->itd.itd_be = htole32(bp0 + offs - 1);
sitd->xfer = xfer;
sitd->flags = 0;
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
sitd = nsitd;
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
iso->next = iso->next + ncur;
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
bp0 = OHCI_PAGE(buf + offs);
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
ncur = 0;
}
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
sitd->itd.itd_offset[ncur] = htole16(OHCI_ITD_MK_OFFS(offs));
offs = noffs;
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
}
nsitd = ohci_alloc_sitd(sc);
if (nsitd == NULL) {
/* XXX what now? */
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
printf("%s: isoc TD alloc failed\n",
USBDEVNAME(sc->sc_bus.bdev));
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
return;
}
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
/* Fixup last used ITD */
sitd->itd.itd_flags = htole32(
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
OHCI_ITD_NOCC |
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
OHCI_ITD_SET_SF(iso->next) |
OHCI_ITD_SET_DI(0) |
OHCI_ITD_SET_FC(ncur));
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
sitd->itd.itd_bp0 = htole32(bp0);
sitd->nextitd = nsitd;
sitd->itd.itd_nextitd = htole32(nsitd->physaddr);
sitd->itd.itd_be = htole32(bp0 + offs - 1);
sitd->xfer = xfer;
sitd->flags = OHCI_CALL_DONE;
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
iso->next = iso->next + ncur;
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
iso->inuse += nframes;
xfer->actlen = offs; /* XXX pretend we did it all */
xfer->status = USBD_IN_PROGRESS;
#ifdef OHCI_DEBUG
if (ohcidebug > 5) {
DPRINTF(("ohci_device_isoc_enter: frame=%d\n",
le32toh(sc->sc_hcca->hcca_frame_number)));
ohci_dump_itds(xfer->hcpriv);
ohci_dump_ed(sed);
}
#endif
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
s = splusb();
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
opipe->tail.itd = nsitd;
Partial sync with NetBSD: Replace LE() macro with htole32()/le32toh().
2001-12-30 23:02:31 +00:00
sed->ed.ed_tailp = htole32(nsitd->physaddr);
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
splx(s);
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
#ifdef OHCI_DEBUG
if (ohcidebug > 5) {
delay(150000);
DPRINTF(("ohci_device_isoc_enter: after frame=%d\n",
le32toh(sc->sc_hcca->hcca_frame_number)));
ohci_dump_itds(xfer->hcpriv);
ohci_dump_ed(sed);
}
#endif
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
}
usbd_status
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_device_isoc_start(usbd_xfer_handle xfer)
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
{
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
struct ohci_pipe *opipe = (struct ohci_pipe *)xfer->pipe;
ohci_softc_t *sc = (ohci_softc_t *)opipe->pipe.device->bus;
DPRINTFN(5,("ohci_device_isoc_start: xfer=%p\n", xfer));
if (sc->sc_dying)
return (USBD_IOERROR);
#ifdef DIAGNOSTIC
if (xfer->status != USBD_IN_PROGRESS)
printf("uhci_device_isoc_start: not in progress %p\n", xfer);
#endif
/* XXX anything to do? */
return (USBD_IN_PROGRESS);
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
}
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_device_isoc_abort(usbd_xfer_handle xfer)
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
{
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
struct ohci_pipe *opipe = (struct ohci_pipe *)xfer->pipe;
ohci_softc_t *sc = (ohci_softc_t *)opipe->pipe.device->bus;
ohci_soft_ed_t *sed;
ohci_soft_itd_t *sitd;
int s;
s = splusb();
DPRINTFN(1,("ohci_device_isoc_abort: xfer=%p\n", xfer));
/* Transfer is already done. */
if (xfer->status != USBD_NOT_STARTED &&
xfer->status != USBD_IN_PROGRESS) {
splx(s);
printf("ohci_device_isoc_abort: early return\n");
return;
}
/* Give xfer the requested abort code. */
xfer->status = USBD_CANCELLED;
sed = opipe->sed;
sed->ed.ed_flags |= htole32(OHCI_ED_SKIP); /* force hardware skip */
sitd = xfer->hcpriv;
#ifdef DIAGNOSTIC
if (sitd == NULL) {
printf("ohci_device_isoc_abort: hcpriv==0\n");
return;
}
#endif
for (; sitd->xfer == xfer; sitd = sitd->nextitd) {
#ifdef DIAGNOSTIC
DPRINTFN(1,("abort sets done sitd=%p\n", sitd));
sitd->isdone = 1;
#endif
}
splx(s);
usb_delay_ms(&sc->sc_bus, OHCI_ITD_NOFFSET);
s = splusb();
/* Run callback. */
usb_transfer_complete(xfer);
sed->ed.ed_headp = htole32(sitd->physaddr); /* unlink TDs */
sed->ed.ed_flags &= htole32(~OHCI_ED_SKIP); /* remove hardware skip */
splx(s);
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
}
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_device_isoc_done(usbd_xfer_handle xfer)
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
{
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
struct ohci_pipe *opipe = (struct ohci_pipe *)xfer->pipe;
ohci_softc_t *sc = (ohci_softc_t *)opipe->pipe.device->bus;
ohci_soft_itd_t *sitd, *nsitd;
DPRINTFN(1,("ohci_device_isoc_done: xfer=%p\n", xfer));
for (sitd = xfer->hcpriv;
!(sitd->flags & OHCI_CALL_DONE);
sitd = nsitd) {
nsitd = sitd->nextitd;
DPRINTFN(1,("ohci_device_isoc_done: free sitd=%p\n", sitd));
ohci_free_sitd(sc, sitd);
}
ohci_free_sitd(sc, sitd);
xfer->hcpriv = NULL;
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
}
usbd_status
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_setup_isoc(usbd_pipe_handle pipe)
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
{
struct ohci_pipe *opipe = (struct ohci_pipe *)pipe;
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
ohci_softc_t *sc = (ohci_softc_t *)pipe->device->bus;
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
struct iso *iso = &opipe->u.iso;
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
int s;
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
iso->next = -1;
iso->inuse = 0;
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
s = splusb();
ohci_add_ed(opipe->sed, sc->sc_isoc_head);
splx(s);
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
return (USBD_NORMAL_COMPLETION);
}
void
Follow NetBSD and ANSIfy the function definitions. Remove trailing whitespaces (submitted to NetBSD).
2002-02-03 17:03:34 +00:00
ohci_device_isoc_close(usbd_pipe_handle pipe)
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
{
struct ohci_pipe *opipe = (struct ohci_pipe *)pipe;
ohci_softc_t *sc = (ohci_softc_t *)pipe->device->bus;
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
int s;
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
DPRINTF(("ohci_device_isoc_close: pipe=%p\n", pipe));
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
s = splusb();
ohci_rem_ed(opipe->sed, sc->sc_isoc_head);
splx(s);
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
ohci_close_pipe(pipe, sc->sc_isoc_head);
Merge from NetBSD: ohci.c (1.83), ohcireg.h (1.16), ohcivar.h (1.21) =================================================================== date: 2000/03/29 01:46:26; author: augustss; A first stab at support for isochronous transfers. ===================================================================
2002-04-01 13:18:11 +00:00
#ifdef DIAGNOSTIC
opipe->tail.itd->isdone = 1;
#endif
More USB ethernet tweaks: - Sync ohci, uhci and usbdi modules with NetBSD in order to obtain the following improvements: o New USBD_NO_TSLEEP flag can be used in place of UQ_NO_TSLEEP quirk. This allows drivers to specify busy waiting only for certain transfers (namely control transfers for reading/writing registers and stuff). o New USBD_FORCE_SHORT_XFER flag can be used to deal with devices like the ADMtek Pegasus that sense the end of bulk OUT transfers in a special way (if a transfer is exactly a multiple of 64 bytes in size, you need to send an extra empty packet to terminate the transfer). o usbd_open_pipe_intr() now accepts an interval argument which can be used to change the rate at which the interrupt callback routine is invoked. Specifying USBD_DEFAULT_INTERVAL uses the value specified in the device's config data, but drivers can override it if needed. - Change if_aue to use USBD_FORCE_SHORT_XFER for packet transmissions. - Change if_aue, if_kue and if_cue to use USBD_NO_TSLEEP for all control transfers. We no longer force the non-tsleep hack for bulk transfers since these are done asynchronously anyway. - Removed quirk entry fiddling from if_aue and if_kue since we don't need it anymore now that we have the USBD_NO_TSLEEP flag. - Tweak ulpt, uhid, ums and ukbd drivers to use the new arg to usbd_open_pipe_intr(). - Add a flag to the softc struct in the ethernet drivers to indicate when a device has been detached, and use this flag to perform tests to prevent the drivers from trying to do control transfers if this is the case. This is necessary because calling if_detach() with INET6 enabled will eventually result in a call to the driver's ioctl() routine to delete the multicast groups on the interface, which will result in attempts to perform control transfers. (It's possible this also happens even without INET6 support enabled.) This is pointless since we know that if the detach method has been called, the hardware has been unplugged. - Changed watchdog timeout routines to just call the driver init routines to initialize the device states without trying to close and re-open the pipes. This is partly because we don't want to frob things at interrupt context, but also because this doesn't seem to work right and I don't want to panic the system just because a USB device may have stopped responding. - Fix aue_rxeof() to be a little smarter about detecting when a double transfer is needed. Unfortunately, the design of the chip makes it hard to get this exactly right. Hopefully, this will go away once either Nick or Lennart finds the bug in the uhci driver that makes this ugly hack necessary. - Also sync usbdevs with NetBSD.
2000-01-20 07:38:33 +00:00
ohci_free_sitd(sc, opipe->tail.itd);
}
Reference in New Issue Copy Permalink