d/freebsd-skq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
e230c2c3ca
freebsd-skq/sys/dev/pci/pci.c

2156 lines
60 KiB
C
Raw Normal View History

Start each of the license/copyright comments with /*-, minor shuffle of lines
2005-01-06 01:43:34 +00:00
/*-
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
* Copyright (c) 1997, Stefan Esser <se@freebsd.org>
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
* Copyright (c) 2000, Michael Smith <msmith@freebsd.org>
* Copyright (c) 2000, BSDi
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice unmodified, this list of conditions, and the following
* disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
* IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
* NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
* THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
Submitted by: Wolfgang Stanglmeier <wolf@dentaro.GUN.de> + Stefan Esser <se> Directory for PCI autoconfigure and device driver code.
1994-09-01 01:45:19 +00:00
Update the special handling code for ATA devices to allow usage of PCI native addressing. That means that if the HW says that using "real" addresses instead of the hardwired legacy compat ones is allowed, we will use them.
2004-06-29 20:25:43 +00:00
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
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 "opt_bus.h"
Added a bit of missing functionality to make this work correctly on a wider variety of systems. Include the deivers from pci_intel.c in pci_config.c (I hope this is what was intended; my system works ok). Use pmap_mapdev(). Automatically map any large linear frame buffers or whatnot in VGA-style devices which ordinarily would not have their own drivers, and don't call not_supported() for them. (This shuts up complaints about my Matrox card.) Include the beginnings of what could eventually become dynamically-loadable PCI devices. Allow for the possibility of PCI devices simply providing a PCI veneer over an existing ISA device, and shut up about them, too. Make autoconfiguration text conform more to the style of other supported buses.
1994-09-14 01:34:51 +00:00
#include <sys/param.h>
#include <sys/systm.h>
Submitted by: Wolfgang Stanglmeier <wolf@dentaro.GUN.de> Bug fixed, that caused system hang on first interrupt on some motherboards. New version of PCI bus configuration code, now supports dynamic interrupt configuration (using BIOS supplied values). NCR SCSI and DEC Ethernet driver patched to use this feature. *** Remove PCI IRQ specifications from your kernel config file ! ***
1994-10-12 02:33:23 +00:00
#include <sys/malloc.h>
Replace the pcidevice_set linker set based configuration mechanism for old style pci drivers with a simple one-line change to use a module that registers itself under new-bus and should in theory enable just about all of the pci drivers to be loadable (kldload and loader(8)) but without having the impact of converting the APIs yet. This also fixes the problem of having undefined variables when only new-style pci drivers are present.
1999-04-24 19:59:20 +00:00
#include <sys/module.h>
Next phase in the PCI subsystem cleanup. - Move PCI core code to dev/pci. - Split bridge code out into separate modules. - Remove the descriptive strings from the bridge drivers. If you want to know what a device is, use pciconf. Add support for broadly identifying devices based on class/subclass, and for parsing a preloaded device identification database so that if you want to waste the memory, you can identify *anything* we know about. - Remove machine-dependant code from the core PCI code. APIC interrupt mapping is performed by shadowing the intline register in machine- dependant code. - Bring interrupt routing support to the Alpha (although many platforms don't yet support routing or mapping interrupts entirely correctly). This resulted in spamming <sys/bus.h> into more places than it really should have gone. - Put sys/dev on the kernel/modules include path. This avoids having to change *all* the pci*.h includes.
2000-12-08 22:11:23 +00:00
#include <sys/linker.h>
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
#include <sys/fcntl.h>
Add support for ioctl() accesses to PCI config space registers. Garrett Wollman sent me this code a few weeks ago for review, and I made some significant changes, which he in turn accepted ... In order to make use of these changes, a device entry has to added to /dev. Submitted by: wollman
1996-10-22 20:20:14 +00:00
#include <sys/conf.h>
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
#include <sys/kernel.h>
Revive PCIConf. Submitted by: "Kenneth D. Merry" <ken@plutotech.com>
1998-09-15 08:21:13 +00:00
#include <sys/queue.h>
Make PCI_ENABLE_IO_MODES a sysctl hw.pci.enable_io_modes. It can also be set at boot time. It defaults to 1 now since it can be set in the boot loader. If this proves unwise, we can reset it to defaulting to 0.
2002-07-26 07:58:16 +00:00
#include <sys/sysctl.h>
Revive PCIConf. Submitted by: "Kenneth D. Merry" <ken@plutotech.com>
1998-09-15 08:21:13 +00:00
#include <sys/types.h>
Completely new PCI code: 1) Supports PCI to PCI bridge devices (and tries to initialise them, even if the BIOS is brain dead). 2) Supports shared PCI interrupts. Interrupt handlers now MUST return '0' if they found nothing to do, '1' otherwise. New features tested with i486 systems based on the Intel Saturn and a DEC 4channel Ethernet card only, but expected to work on most systems. The option PCI_REMAP has been removed ! Submitted by: Wolfgang Stanglmeier <wolf@kintaro.cologne.de>
1995-03-21 23:01:06 +00:00
Submitted by: Wolfgang Stanglmeier <wolf@dentaro.GUN.de> + Stefan Esser <se> Directory for PCI autoconfigure and device driver code.
1994-09-01 01:45:19 +00:00
#include <vm/vm.h>
Untangled the vm.h include file spaghetti.
1995-12-07 12:48:31 +00:00
#include <vm/pmap.h>
Revive PCIConf. Submitted by: "Kenneth D. Merry" <ken@plutotech.com>
1998-09-15 08:21:13 +00:00
#include <vm/vm_extern.h>
Submitted by: Wolfgang Stanglmeier <wolf@dentaro.GUN.de> + Stefan Esser <se> Directory for PCI autoconfigure and device driver code.
1994-09-01 01:45:19 +00:00
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 <sys/bus.h>
#include <machine/bus.h>
#include <sys/rman.h>
#include <machine/resource.h>
Encapsulate the old PCI compatability support and APIs completely under "options COMPAT_OLDPCI". This option already existed, but now also tidies up the declarations in #include <pci/pci*.h>. It is amazing how much stuff was using the old pre-FreeBSD 3.x names and going silently undetected.
2000-05-28 16:35:57 +00:00
#include <sys/pciio.h>
Use the pci.c code wherever possible, rather than copying all the pci code into cardbus and s/pci/cardbus. This exposes a few pci_* functions that are now static. This work is similar to work Justin posted to the mobile list about a year or two ago, which I have neglected since then. This is a subset of his current work with the multiple inheritance newbus architecutre. When completed, that will eliminate the need for pci/pci_private.h. Similar work is needed for the cardbus_cis and pccard_cis code as well.
2002-02-27 05:09:14 +00:00
#include <dev/pci/pcireg.h>
#include <dev/pci/pcivar.h>
#include <dev/pci/pci_private.h>
Completely new PCI code: 1) Supports PCI to PCI bridge devices (and tries to initialise them, even if the BIOS is brain dead). 2) Supports shared PCI interrupts. Interrupt handlers now MUST return '0' if they found nothing to do, '1' otherwise. New features tested with i486 systems based on the Intel Saturn and a DEC 4channel Ethernet card only, but expected to work on most systems. The option PCI_REMAP has been removed ! Submitted by: Wolfgang Stanglmeier <wolf@kintaro.cologne.de>
1995-03-21 23:01:06 +00:00
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
#include "pcib_if.h"
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
#include "pci_if.h"
Use __HAVE_ACPI and __PCI_REROUTE_INTERRUPT as appropriate rather than the complicated #ifdefs.
2006-01-01 21:04:31 +00:00
#ifdef __HAVE_ACPI
Turn ACPI and PCI devices off or to a lower power state in suspend and back on again in resume. Override the default of D3 with the value the BIOS specifies in _SxD, if present. Skip serial devices (PNP05xx) since they seem to hang when set to D3 and may require special driver support. Also, skip non-type 0 PCI devices (i.e., bridges) since our we don't yet save/restore their config space and that seems to be necessary. If this gives you trouble with suspend/resume, you can disable the new ACPI and PCI power behavior separately with these tunables & sysctls: debug.acpi.do_powerstate hw.pci.do_powerstate Approved by: imp (pci) Tested by: acpi@ (numerous)
2004-12-02 08:07:12 +00:00
#include <contrib/dev/acpica/acpi.h>
#include "acpi_if.h"
Non-x86 platforms cannot use the ACPI includes. This should be fixed but for now, only include the headers for i386, amd64, or ia64. Pointed out by: grehan
2004-12-03 08:13:08 +00:00
#else
#define ACPI_PWR_FOR_SLEEP(x, y, z)
#endif
Turn ACPI and PCI devices off or to a lower power state in suspend and back on again in resume. Override the default of D3 with the value the BIOS specifies in _SxD, if present. Skip serial devices (PNP05xx) since they seem to hang when set to D3 and may require special driver support. Also, skip non-type 0 PCI devices (i.e., bridges) since our we don't yet save/restore their config space and that seems to be necessary. If this gives you trouble with suspend/resume, you can disable the new ACPI and PCI power behavior separately with these tunables & sysctls: debug.acpi.do_powerstate hw.pci.do_powerstate Approved by: imp (pci) Tested by: acpi@ (numerous)
2004-12-02 08:07:12 +00:00
Minor whitespace changes to conform better to stlye(9) and reduce diffs with uncommitted changes I have in p4.
2003-12-24 02:01:22 +00:00
static uint32_t pci_mapbase(unsigned mapreg);
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
static int pci_maptype(unsigned mapreg);
static int pci_mapsize(unsigned testval);
static int pci_maprange(unsigned mapreg);
static void pci_fixancient(pcicfgregs *cfg);
static int pci_porten(device_t pcib, int b, int s, int f);
static int pci_memen(device_t pcib, int b, int s, int f);
- Consolidate duplicated code for assigning interrupts to PCI devices via routing, etc. in a static pci_assign_interrupt() function. - Add a sledgehammer that allows the user to override the interrupt assignment of any PCI device via a tunable (e.g. "hw.pci0.7.INTB=5" would force any functions on the pci device in slot 7 of bus 0 that use B# to use IRQ 5). This should be used with great caution! Generally, if the interrupt routing in use provides specific tunables (such as hard-wiring the IRQ for a given $PIR or ACPI PCI link device), then those should be used instead. One instance where this tunable might be useful is if a box has an MPTable with duplicate entries for the same PCI device with different IRQs. MFC after: 1 week
2005-09-29 15:04:41 +00:00
static void pci_assign_interrupt(device_t bus, device_t dev,
int force_route);
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
static int pci_add_map(device_t pcib, device_t bus, device_t dev,
int b, int s, int f, int reg,
Expose pci_add_resources to the outside world, add a 'force' flag to force allocation of unallocated BARs (cardbus uses this to preallocate everything). Add a prefetchmask to allow for busses that get prefetch hints to set them. Addjust pci_add_map and pci_ata_maps to take a new force flag which pci_add_resources will pass in. Implement 'force' in pci_add_map. Write new value of allocated resource into the bar, if the allocation succeeded (we should have done this before, but with the new force the bug was very obvious).
2005-12-30 19:28:26 +00:00
struct resource_list *rl, int force, int prefetch);
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
static int pci_probe(device_t dev);
Various changes to make it easier to subclass the PCI bus device. - Make the pci devclass a global variable. - Add child devices in pci_attach() instead of pci_probe(). Change pci_probe() to just check for a valid bus number from the associated bridge and return -1000 if successful. This allows subclasses of the PCI bus driver to override the generic driver. - Move the code to load the vendor data into its own public function. Really though, doing this at attach is just plain wrong. This should really be done in the module load routine instead. As a side effect, the 'busno' variable in pci_attach() is now no longer static (minor bug that was harmless so far.) - Change pci_add_children() to take an extra argument that is the size of the device info structure passed to pci_read_device() and make it public so subclasses of the PCI bus can call it in their attach routines. - Move the bits to attach a probed PCI child to a PCI bus into a global pci_add_child() function. This will allow subclasses that can detect a PCI device not found in the normal PCI probe to add those devices in their own attach routine. (I have seen this in the ACPI tree on my laptop for example.) As a side effect, change the static function pci_add_resources() to get the busno, slot, and func from the passed in dinfo structure instead of requiring them as function arguments. Tested on: i386, alpha, ia64, sparc64
2002-08-26 15:23:52 +00:00
static int pci_attach(device_t dev);
- Make pci_load_vendor_data() static and do it during MOD_LOAD instead of when the first PCI bus attaches. - Create /dev/pci during MOD_LOAD as well. - Destroy /dev/pci during MOD_UNLOAD (not that you can kldunload pci, but might as well get the code right)
2002-09-04 03:13:16 +00:00
static void pci_load_vendor_data(void);
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
static int pci_describe_parse_line(char **ptr, int *vendor,
Minor whitespace changes to conform better to stlye(9) and reduce diffs with uncommitted changes I have in p4.
2003-12-24 02:01:22 +00:00
int *device, char **desc);
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
static char *pci_describe_device(device_t dev);
static int pci_modevent(module_t mod, int what, void *arg);
These don't need to be semi-public after all.
2003-02-17 19:47:02 +00:00
static void pci_hdrtypedata(device_t pcib, int b, int s, int f,
Minor whitespace changes to conform better to stlye(9) and reduce diffs with uncommitted changes I have in p4.
2003-12-24 02:01:22 +00:00
pcicfgregs *cfg);
These don't need to be semi-public after all.
2003-02-17 19:47:02 +00:00
static void pci_read_extcap(device_t pcib, pcicfgregs *cfg);
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
static device_method_t pci_methods[] = {
/* Device interface */
DEVMETHOD(device_probe, pci_probe),
Various changes to make it easier to subclass the PCI bus device. - Make the pci devclass a global variable. - Add child devices in pci_attach() instead of pci_probe(). Change pci_probe() to just check for a valid bus number from the associated bridge and return -1000 if successful. This allows subclasses of the PCI bus driver to override the generic driver. - Move the code to load the vendor data into its own public function. Really though, doing this at attach is just plain wrong. This should really be done in the module load routine instead. As a side effect, the 'busno' variable in pci_attach() is now no longer static (minor bug that was harmless so far.) - Change pci_add_children() to take an extra argument that is the size of the device info structure passed to pci_read_device() and make it public so subclasses of the PCI bus can call it in their attach routines. - Move the bits to attach a probed PCI child to a PCI bus into a global pci_add_child() function. This will allow subclasses that can detect a PCI device not found in the normal PCI probe to add those devices in their own attach routine. (I have seen this in the ACPI tree on my laptop for example.) As a side effect, change the static function pci_add_resources() to get the busno, slot, and func from the passed in dinfo structure instead of requiring them as function arguments. Tested on: i386, alpha, ia64, sparc64
2002-08-26 15:23:52 +00:00
DEVMETHOD(device_attach, pci_attach),
Add a detach for pci bridge and pci bus drivers. This allows one to theoretically unload pci bridges or pci drivers. It will also allow detach to work if one needed to detach a subtree. This is inspired by looking at the p4 commits from bms to his 5.4 tree, but I didn't look at the final results.
2005-04-29 06:22:41 +00:00
DEVMETHOD(device_detach, bus_generic_detach),
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
DEVMETHOD(device_shutdown, bus_generic_shutdown),
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
DEVMETHOD(device_suspend, pci_suspend),
Add pci_resume() to reestablish interrupt routing after suspend/resume. Especially after hibernation, interrupt routing went back to initial status on some machines.
2003-09-17 08:32:44 +00:00
DEVMETHOD(device_resume, pci_resume),
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
/* Bus interface */
DEVMETHOD(bus_print_child, pci_print_child),
DEVMETHOD(bus_probe_nomatch, pci_probe_nomatch),
DEVMETHOD(bus_read_ivar, pci_read_ivar),
DEVMETHOD(bus_write_ivar, pci_write_ivar),
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
DEVMETHOD(bus_driver_added, pci_driver_added),
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
DEVMETHOD(bus_setup_intr, bus_generic_setup_intr),
DEVMETHOD(bus_teardown_intr, bus_generic_teardown_intr),
DEVMETHOD(bus_get_resource_list,pci_get_resource_list),
DEVMETHOD(bus_set_resource, bus_generic_rl_set_resource),
DEVMETHOD(bus_get_resource, bus_generic_rl_get_resource),
DEVMETHOD(bus_delete_resource, pci_delete_resource),
DEVMETHOD(bus_alloc_resource, pci_alloc_resource),
DEVMETHOD(bus_release_resource, bus_generic_rl_release_resource),
DEVMETHOD(bus_activate_resource, bus_generic_activate_resource),
DEVMETHOD(bus_deactivate_resource, bus_generic_deactivate_resource),
Move the pnp and location info into the common pci bus. Make all known pci busses implement this. Also minor comment smithing in cardbus. Fix copyright to this year with my name on it since I've been doing a lot to this file. Reviewed by: jhb
2003-02-17 21:20:35 +00:00
DEVMETHOD(bus_child_pnpinfo_str, pci_child_pnpinfo_str_method),
DEVMETHOD(bus_child_location_str, pci_child_location_str_method),
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
/* PCI interface */
DEVMETHOD(pci_read_config, pci_read_config_method),
DEVMETHOD(pci_write_config, pci_write_config_method),
Slightly reimplement some recently added helper functions as methods, so that drivers are not reaching into the internals of the pci bus. There are no driver changes, the public interface is the same.
2001-02-27 23:13:20 +00:00
DEVMETHOD(pci_enable_busmaster, pci_enable_busmaster_method),
DEVMETHOD(pci_disable_busmaster, pci_disable_busmaster_method),
DEVMETHOD(pci_enable_io, pci_enable_io_method),
DEVMETHOD(pci_disable_io, pci_disable_io_method),
DEVMETHOD(pci_get_powerstate, pci_get_powerstate_method),
DEVMETHOD(pci_set_powerstate, pci_set_powerstate_method),
Add a new PCI interface method, assign_interrupt, to determine the interrupt to be used for a device. This is intended solely for internal use of PCI bus implementations, and exists so that PCI bus drivers implementing special interrupt assignment methods which require additional work at the bus level to work right can be easily derived from the generic driver (or any other one) without resorting to hacks. It will be used in the sparc64 ofw_pcibus driver, which will be committed shortly. Make use of this method in the generic implementation, and add it to the method table of bus drivers derived from the PCI one. Reviewed by: imp, -hackers
2003-07-01 14:08:33 +00:00
DEVMETHOD(pci_assign_interrupt, pci_assign_interrupt_method),
Add a new method PCI_FIND_EXTCAP() to the pci bus interface that is used to search for a specific extended capability. If the specified capability is found for the given device, then the function returns success and optionally returns the offset of that capability. If the capability is not found, the function returns an error.
2005-12-20 19:57:47 +00:00
DEVMETHOD(pci_find_extcap, pci_find_extcap_method),
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
{ 0, 0 }
};
Make the cardbus driver a derived class of the pci driver. In theory, this should allow many of the pci methods to be re-staticised.
2003-11-01 12:45:03 +00:00
DEFINE_CLASS_0(pci, pci_driver, pci_methods, 0);
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
Make the 'pci_devclass' pointer variable private (drivers really shouldn't share devclass pointers, a mistake I've encouraged in the past) and move the declaration of the pci_driver kobj class from cardbus.c to pci_private.h so that other drivers can inherit from pci_driver.
2006-01-20 22:00:50 +00:00
static devclass_t pci_devclass;
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
DRIVER_MODULE(pci, pcib, pci_driver, pci_devclass, pci_modevent, 0);
Add MODULE_VERSION.
2002-04-17 00:31:32 +00:00
MODULE_VERSION(pci, 1);
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
Next phase in the PCI subsystem cleanup. - Move PCI core code to dev/pci. - Split bridge code out into separate modules. - Remove the descriptive strings from the bridge drivers. If you want to know what a device is, use pciconf. Add support for broadly identifying devices based on class/subclass, and for parsing a preloaded device identification database so that if you want to waste the memory, you can identify *anything* we know about. - Remove machine-dependant code from the core PCI code. APIC interrupt mapping is performed by shadowing the intline register in machine- dependant code. - Bring interrupt routing support to the Alpha (although many platforms don't yet support routing or mapping interrupts entirely correctly). This resulted in spamming <sys/bus.h> into more places than it really should have gone. - Put sys/dev on the kernel/modules include path. This avoids having to change *all* the pci*.h includes.
2000-12-08 22:11:23 +00:00
static char *pci_vendordata;
static size_t pci_vendordata_size;
Modified to use renamed get_pci_apic_irq() -> pci_apic_pin() function.
1997-06-25 20:56:29 +00:00
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
Fix some resource allocation peculiarities of the intpm device.
1999-10-28 08:06:59 +00:00
struct pci_quirk {
Prefer the uintXX_t to the u_intXX_t names.
2003-08-22 03:11:53 +00:00
uint32_t devid; /* Vendor/device of the card */
Fix some resource allocation peculiarities of the intpm device.
1999-10-28 08:06:59 +00:00
int type;
Fix typo: wierd -> weird. There is no such thing as wierd in the english language.
2001-02-06 09:25:10 +00:00
#define PCI_QUIRK_MAP_REG 1 /* PCI map register in weird place */
Fix some resource allocation peculiarities of the intpm device.
1999-10-28 08:06:59 +00:00
int arg1;
int arg2;
};
struct pci_quirk pci_quirks[] = {
Add support for the Intel 82443MX chipset PR: kern/33032 MFC after: 1 month
2001-12-21 01:28:59 +00:00
/* The Intel 82371AB and 82443MX has a map register at offset 0x90. */
Fix some resource allocation peculiarities of the intpm device.
1999-10-28 08:06:59 +00:00
{ 0x71138086, PCI_QUIRK_MAP_REG, 0x90, 0 },
Add support for the Intel 82443MX chipset PR: kern/33032 MFC after: 1 month
2001-12-21 01:28:59 +00:00
{ 0x719b8086, PCI_QUIRK_MAP_REG, 0x90, 0 },
The serverworks OSB4 pci->isa bridge has the same mapping register at offset 0x90 for the SMBus device as the PIIX4.
2001-03-15 06:51:45 +00:00
/* As does the Serverworks OSB4 (the SMBus mapping register) */
{ 0x02001166, PCI_QUIRK_MAP_REG, 0x90, 0 },
Fix some resource allocation peculiarities of the intpm device.
1999-10-28 08:06:59 +00:00
{ 0 }
};
* Implement bus_set/get/delete_resource for pci. * Change the hack used on the alpha for mapping devices into DENSE or BWX memory spaces to a simpler one. Its still a hack and should be a seperate api to explicitly map the resource. * Add $FreeBSD$ as necessary.
1999-10-14 21:38:33 +00:00
/* map register information */
#define PCI_MAPMEM 0x01 /* memory map */
#define PCI_MAPMEMP 0x02 /* prefetchable memory map */
#define PCI_MAPPORT 0x04 /* port map */
Don't put variable declarations in header files, put prototypes. 'pci_devq' provides useful information now.
2001-12-19 08:49:11 +00:00
struct devlist pci_devq;
Prefer the uintXX_t to the u_intXX_t names.
2003-08-22 03:11:53 +00:00
uint32_t pci_generation;
uint32_t pci_numdevs = 0;
Revive PCIConf. Submitted by: "Kenneth D. Merry" <ken@plutotech.com>
1998-09-15 08:21:13 +00:00
Make PCI_ENABLE_IO_MODES a sysctl hw.pci.enable_io_modes. It can also be set at boot time. It defaults to 1 now since it can be set in the boot loader. If this proves unwise, we can reset it to defaulting to 0.
2002-07-26 07:58:16 +00:00
/* sysctl vars */
SYSCTL_NODE(_hw, OID_AUTO, pci, CTLFLAG_RD, 0, "PCI bus tuning parameters");
ooops. I disabled pci_enable_io_modes not pci_do_powerstate in the last commit. That was in error. Noticed by: sos
2004-04-16 15:01:54 +00:00
static int pci_enable_io_modes = 1;
Turn ACPI and PCI devices off or to a lower power state in suspend and back on again in resume. Override the default of D3 with the value the BIOS specifies in _SxD, if present. Skip serial devices (PNP05xx) since they seem to hang when set to D3 and may require special driver support. Also, skip non-type 0 PCI devices (i.e., bridges) since our we don't yet save/restore their config space and that seems to be necessary. If this gives you trouble with suspend/resume, you can disable the new ACPI and PCI power behavior separately with these tunables & sysctls: debug.acpi.do_powerstate hw.pci.do_powerstate Approved by: imp (pci) Tested by: acpi@ (numerous)
2004-12-02 08:07:12 +00:00
TUNABLE_INT("hw.pci.enable_io_modes", &pci_enable_io_modes);
Make PCI_ENABLE_IO_MODES a sysctl hw.pci.enable_io_modes. It can also be set at boot time. It defaults to 1 now since it can be set in the boot loader. If this proves unwise, we can reset it to defaulting to 0.
2002-07-26 07:58:16 +00:00
SYSCTL_INT(_hw_pci, OID_AUTO, enable_io_modes, CTLFLAG_RW,
ooops. I disabled pci_enable_io_modes not pci_do_powerstate in the last commit. That was in error. Noticed by: sos
2004-04-16 15:01:54 +00:00
&pci_enable_io_modes, 1,
Make PCI_ENABLE_IO_MODES a sysctl hw.pci.enable_io_modes. It can also be set at boot time. It defaults to 1 now since it can be set in the boot loader. If this proves unwise, we can reset it to defaulting to 0.
2002-07-26 07:58:16 +00:00
"Enable I/O and memory bits in the config register. Some BIOSes do not\n\
enable these bits correctly. We'd like to do this all the time, but there\n\
are some peripherals that this causes problems with.");
Split power state control into two variables. hw.pci.do_powerstate has been removed. It has been replaced by hw.pci.do_power_nodriver and hw.pci.do_power_resume. The former defaults to 0 while the latter defaults to 1. When do_powerstate was set to 0, it broke suspend/resume for a lot of people as an unintended consequence. This change will only affect the areas that were intended to affect. This change will have no effect on servers, but will help laptops quite a bit. MFC After: 3 days.
2005-09-21 19:47:00 +00:00
static int pci_do_power_nodriver = 0;
TUNABLE_INT("hw.pci.do_power_nodriver", &pci_do_power_nodriver);
SYSCTL_INT(_hw_pci, OID_AUTO, do_power_nodriver, CTLFLAG_RW,
&pci_do_power_nodriver, 0,
"Place a function into D3 state when no driver attaches to it. 0 means\n\
disable. 1 means conservatively place devices into D3 state. 2 means\n\
agressively place devices into D3 state. 3 means put absolutely everything\n\
in D3 state.");
static int pci_do_power_resume = 1;
TUNABLE_INT("hw.pci.do_power_resume", &pci_do_power_resume);
SYSCTL_INT(_hw_pci, OID_AUTO, do_power_resume, CTLFLAG_RW,
&pci_do_power_resume, 1,
"Transition from D3 -> D0 on resume.");
Add system tunable to turn off power state changes. Default to off until we get the resource allocation stuff hammered out. Fix and off by one error that caused unnecessary filtering of valid BARs for only 4 bytes than ICH3 and other PCI IDE controllers have. Andrew Gallatin submitted this, although it doesn't solve the problems ICH3 controllers have with the new code, it does restore the former resource list on the probe line.
2004-04-11 07:02:49 +00:00
Implement 2 small helper functions: pci_find_bsf() - Find a device_t by bus/slot/function. pci_find_device() - Find a device_t by vendor/device ID.
2002-01-10 00:56:02 +00:00
/* Find a device_t by bus/slot/function */
device_t
Prefer the uintXX_t to the u_intXX_t names.
2003-08-22 03:11:53 +00:00
pci_find_bsf(uint8_t bus, uint8_t slot, uint8_t func)
Implement 2 small helper functions: pci_find_bsf() - Find a device_t by bus/slot/function. pci_find_device() - Find a device_t by vendor/device ID.
2002-01-10 00:56:02 +00:00
{
struct pci_devinfo *dinfo;
STAILQ_FOREACH(dinfo, &pci_devq, pci_links) {
if ((dinfo->cfg.bus == bus) &&
(dinfo->cfg.slot == slot) &&
(dinfo->cfg.func == func)) {
return (dinfo->cfg.dev);
}
}
return (NULL);
}
/* Find a device_t by vendor/device ID */
device_t
Prefer the uintXX_t to the u_intXX_t names.
2003-08-22 03:11:53 +00:00
pci_find_device(uint16_t vendor, uint16_t device)
Implement 2 small helper functions: pci_find_bsf() - Find a device_t by bus/slot/function. pci_find_device() - Find a device_t by vendor/device ID.
2002-01-10 00:56:02 +00:00
{
struct pci_devinfo *dinfo;
STAILQ_FOREACH(dinfo, &pci_devq, pci_links) {
if ((dinfo->cfg.vendor == vendor) &&
(dinfo->cfg.device == device)) {
return (dinfo->cfg.dev);
}
}
return (NULL);
}
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
/* return base address of memory or port map */
Submitted by: Wolfgang Stanglmeier <wolf@dentaro.GUN.de> Bug fixed, that caused system hang on first interrupt on some motherboards. New version of PCI bus configuration code, now supports dynamic interrupt configuration (using BIOS supplied values). NCR SCSI and DEC Ethernet driver patched to use this feature. *** Remove PCI IRQ specifications from your kernel config file ! ***
1994-10-12 02:33:23 +00:00
Prefer the uintXX_t to the u_intXX_t names.
2003-08-22 03:11:53 +00:00
static uint32_t
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
pci_mapbase(unsigned mapreg)
Completely new PCI code: 1) Supports PCI to PCI bridge devices (and tries to initialise them, even if the BIOS is brain dead). 2) Supports shared PCI interrupts. Interrupt handlers now MUST return '0' if they found nothing to do, '1' otherwise. New features tested with i486 systems based on the Intel Saturn and a DEC 4channel Ethernet card only, but expected to work on most systems. The option PCI_REMAP has been removed ! Submitted by: Wolfgang Stanglmeier <wolf@kintaro.cologne.de>
1995-03-21 23:01:06 +00:00
{
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
int mask = 0x03;
if ((mapreg & 0x01) == 0)
mask = 0x0f;
return (mapreg & ~mask);
Completely new PCI code: 1) Supports PCI to PCI bridge devices (and tries to initialise them, even if the BIOS is brain dead). 2) Supports shared PCI interrupts. Interrupt handlers now MUST return '0' if they found nothing to do, '1' otherwise. New features tested with i486 systems based on the Intel Saturn and a DEC 4channel Ethernet card only, but expected to work on most systems. The option PCI_REMAP has been removed ! Submitted by: Wolfgang Stanglmeier <wolf@kintaro.cologne.de>
1995-03-21 23:01:06 +00:00
}
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
/* return map type of memory or port map */
Update PCI bus code from my current sources: - always use pci_conf_read() and pci_conf_write(). (This is required to simulate non-existant devices in my system for PCI bridge code tests.) - reorder some functions (put the main functions at the end). - correct off by one bug in the code dealing with unitialized PCI to PCI bridge chips. (Bug found by ASAMI Satoshi.) - print function number for multi-function devices.
1996-04-14 20:14:36 +00:00
static int
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
pci_maptype(unsigned mapreg)
{
Prefer the uintXX_t to the u_intXX_t names.
2003-08-22 03:11:53 +00:00
static uint8_t maptype[0x10] = {
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
PCI_MAPMEM, PCI_MAPPORT,
PCI_MAPMEM, 0,
PCI_MAPMEM, PCI_MAPPORT,
0, 0,
PCI_MAPMEM|PCI_MAPMEMP, PCI_MAPPORT,
PCI_MAPMEM|PCI_MAPMEMP, 0,
PCI_MAPMEM|PCI_MAPMEMP, PCI_MAPPORT,
0, 0,
Add PCI LKM support.
1997-01-21 23:23:40 +00:00
};
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
return maptype[mapreg & 0x0f];
Add PCI LKM support.
1997-01-21 23:23:40 +00:00
}
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
/* return log2 of map size decoded for memory or port map */
Completely new PCI code: 1) Supports PCI to PCI bridge devices (and tries to initialise them, even if the BIOS is brain dead). 2) Supports shared PCI interrupts. Interrupt handlers now MUST return '0' if they found nothing to do, '1' otherwise. New features tested with i486 systems based on the Intel Saturn and a DEC 4channel Ethernet card only, but expected to work on most systems. The option PCI_REMAP has been removed ! Submitted by: Wolfgang Stanglmeier <wolf@kintaro.cologne.de>
1995-03-21 23:01:06 +00:00
Add heuristic to detect multi-function devices that don't announce this feature in the header type register, though it is required by the PCI spec. This should correctly probe both functions of the Intel 82371FB chip, without the need for a special case based on the device ID.
1996-01-30 01:14:29 +00:00
static int
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
pci_mapsize(unsigned testval)
Add heuristic to detect multi-function devices that don't announce this feature in the header type register, though it is required by the PCI spec. This should correctly probe both functions of the Intel 82371FB chip, without the need for a special case based on the device ID.
1996-01-30 01:14:29 +00:00
{
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
int ln2size;
Add heuristic to detect multi-function devices that don't announce this feature in the header type register, though it is required by the PCI spec. This should correctly probe both functions of the Intel 82371FB chip, without the need for a special case based on the device ID.
1996-01-30 01:14:29 +00:00
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
testval = pci_mapbase(testval);
Two minor changes to the code that builds the pci map array: 1) Stop at the first map register that contains a zero value. 2) When testing for the map size work up from low values, since this works around a bug in some BusLogic SCSI card, which has the 16 upper port base address bits hardwired to zero. The config register dump printed in the bootverbose case has been slightly rearranged.
1997-05-28 10:01:03 +00:00
ln2size = 0;
if (testval != 0) {
while ((testval & 1) == 0)
{
ln2size++;
testval >>= 1;
}
Make mapping messages depend on bootverbose flag. Add PCI subclass to unknown device message.
1995-09-07 15:20:53 +00:00
}
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
return (ln2size);
Completely new PCI code: 1) Supports PCI to PCI bridge devices (and tries to initialise them, even if the BIOS is brain dead). 2) Supports shared PCI interrupts. Interrupt handlers now MUST return '0' if they found nothing to do, '1' otherwise. New features tested with i486 systems based on the Intel Saturn and a DEC 4channel Ethernet card only, but expected to work on most systems. The option PCI_REMAP has been removed ! Submitted by: Wolfgang Stanglmeier <wolf@kintaro.cologne.de>
1995-03-21 23:01:06 +00:00
}
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
/* return log2 of address range supported by map register */
Completely new PCI code: 1) Supports PCI to PCI bridge devices (and tries to initialise them, even if the BIOS is brain dead). 2) Supports shared PCI interrupts. Interrupt handlers now MUST return '0' if they found nothing to do, '1' otherwise. New features tested with i486 systems based on the Intel Saturn and a DEC 4channel Ethernet card only, but expected to work on most systems. The option PCI_REMAP has been removed ! Submitted by: Wolfgang Stanglmeier <wolf@kintaro.cologne.de>
1995-03-21 23:01:06 +00:00
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
static int
pci_maprange(unsigned mapreg)
{
int ln2range = 0;
switch (mapreg & 0x07) {
case 0x00:
case 0x01:
case 0x05:
ln2range = 32;
break;
case 0x02:
ln2range = 20;
break;
case 0x04:
ln2range = 64;
break;
Update PCI bus code from my current sources: - always use pci_conf_read() and pci_conf_write(). (This is required to simulate non-existant devices in my system for PCI bridge code tests.) - reorder some functions (put the main functions at the end). - correct off by one bug in the code dealing with unitialized PCI to PCI bridge chips. (Bug found by ASAMI Satoshi.) - print function number for multi-function devices.
1996-04-14 20:14:36 +00:00
}
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
return (ln2range);
Reapplied all of Stefan's changes. What a mess - the files were modified and moved at the same time. This made it *very* difficult to fix the revision log lossage that happend when the files were moved. SIGH.
1995-02-02 13:12:18 +00:00
}
Completely new PCI code: 1) Supports PCI to PCI bridge devices (and tries to initialise them, even if the BIOS is brain dead). 2) Supports shared PCI interrupts. Interrupt handlers now MUST return '0' if they found nothing to do, '1' otherwise. New features tested with i486 systems based on the Intel Saturn and a DEC 4channel Ethernet card only, but expected to work on most systems. The option PCI_REMAP has been removed ! Submitted by: Wolfgang Stanglmeier <wolf@kintaro.cologne.de>
1995-03-21 23:01:06 +00:00
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
/* adjust some values from PCI 1.0 devices to match 2.0 standards ... */
Add PCI LKM support.
1997-01-21 23:23:40 +00:00
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
static void
pci_fixancient(pcicfgregs *cfg)
Add PCI LKM support.
1997-01-21 23:23:40 +00:00
{
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
if (cfg->hdrtype != 0)
return;
Add PCI LKM support.
1997-01-21 23:23:40 +00:00
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
/* PCI to PCI bridges use header type 1 */
Use "baseclass" instead of "class" for storing the contents of PCI register 0xB so that C++ programs can use the PCI conf interface.
1998-08-13 19:12:20 +00:00
if (cfg->baseclass == PCIC_BRIDGE && cfg->subclass == PCIS_BRIDGE_PCI)
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
cfg->hdrtype = 1;
Add PCI LKM support.
1997-01-21 23:23:40 +00:00
}
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
/* extract header type specific config data */
Completely new PCI code: 1) Supports PCI to PCI bridge devices (and tries to initialise them, even if the BIOS is brain dead). 2) Supports shared PCI interrupts. Interrupt handlers now MUST return '0' if they found nothing to do, '1' otherwise. New features tested with i486 systems based on the Intel Saturn and a DEC 4channel Ethernet card only, but expected to work on most systems. The option PCI_REMAP has been removed ! Submitted by: Wolfgang Stanglmeier <wolf@kintaro.cologne.de>
1995-03-21 23:01:06 +00:00
These don't need to be semi-public after all.
2003-02-17 19:47:02 +00:00
static void
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
pci_hdrtypedata(device_t pcib, int b, int s, int f, pcicfgregs *cfg)
Completely new PCI code: 1) Supports PCI to PCI bridge devices (and tries to initialise them, even if the BIOS is brain dead). 2) Supports shared PCI interrupts. Interrupt handlers now MUST return '0' if they found nothing to do, '1' otherwise. New features tested with i486 systems based on the Intel Saturn and a DEC 4channel Ethernet card only, but expected to work on most systems. The option PCI_REMAP has been removed ! Submitted by: Wolfgang Stanglmeier <wolf@kintaro.cologne.de>
1995-03-21 23:01:06 +00:00
{
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
#define REG(n, w) PCIB_READ_CONFIG(pcib, b, s, f, n, w)
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
switch (cfg->hdrtype) {
case 0:
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
cfg->subvendor = REG(PCIR_SUBVEND_0, 2);
cfg->subdevice = REG(PCIR_SUBDEV_0, 2);
* Implement bus_set/get/delete_resource for pci. * Change the hack used on the alpha for mapping devices into DENSE or BWX memory spaces to a simpler one. Its still a hack and should be a seperate api to explicitly map the resource. * Add $FreeBSD$ as necessary.
1999-10-14 21:38:33 +00:00
cfg->nummaps = PCI_MAXMAPS_0;
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
break;
case 1:
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
cfg->subvendor = REG(PCIR_SUBVEND_1, 2);
cfg->subdevice = REG(PCIR_SUBDEV_1, 2);
* Implement bus_set/get/delete_resource for pci. * Change the hack used on the alpha for mapping devices into DENSE or BWX memory spaces to a simpler one. Its still a hack and should be a seperate api to explicitly map the resource. * Add $FreeBSD$ as necessary.
1999-10-14 21:38:33 +00:00
cfg->nummaps = PCI_MAXMAPS_1;
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
break;
case 2:
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
cfg->subvendor = REG(PCIR_SUBVEND_2, 2);
cfg->subdevice = REG(PCIR_SUBDEV_2, 2);
* Implement bus_set/get/delete_resource for pci. * Change the hack used on the alpha for mapping devices into DENSE or BWX memory spaces to a simpler one. Its still a hack and should be a seperate api to explicitly map the resource. * Add $FreeBSD$ as necessary.
1999-10-14 21:38:33 +00:00
cfg->nummaps = PCI_MAXMAPS_2;
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
break;
Make mapping messages depend on bootverbose flag. Add PCI subclass to unknown device message.
1995-09-07 15:20:53 +00:00
}
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
#undef REG
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
}
Completely new PCI code: 1) Supports PCI to PCI bridge devices (and tries to initialise them, even if the BIOS is brain dead). 2) Supports shared PCI interrupts. Interrupt handlers now MUST return '0' if they found nothing to do, '1' otherwise. New features tested with i486 systems based on the Intel Saturn and a DEC 4channel Ethernet card only, but expected to work on most systems. The option PCI_REMAP has been removed ! Submitted by: Wolfgang Stanglmeier <wolf@kintaro.cologne.de>
1995-03-21 23:01:06 +00:00
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
/* read configuration header into pcicfgregs structure */
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
Use the pci.c code wherever possible, rather than copying all the pci code into cardbus and s/pci/cardbus. This exposes a few pci_* functions that are now static. This work is similar to work Justin posted to the mobile list about a year or two ago, which I have neglected since then. This is a subset of his current work with the multiple inheritance newbus architecutre. When completed, that will eliminate the need for pci/pci_private.h. Similar work is needed for the cardbus_cis and pccard_cis code as well.
2002-02-27 05:09:14 +00:00
struct pci_devinfo *
Add 5th parameter to pci_read_device specifying the size of the object to create.
2002-03-13 16:32:11 +00:00
pci_read_device(device_t pcib, int b, int s, int f, size_t size)
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
{
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
#define REG(n, w) PCIB_READ_CONFIG(pcib, b, s, f, n, w)
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
pcicfgregs *cfg = NULL;
Revive PCIConf. Submitted by: "Kenneth D. Merry" <ken@plutotech.com>
1998-09-15 08:21:13 +00:00
struct pci_devinfo *devlist_entry;
struct devlist *devlist_head;
devlist_head = &pci_devq;
devlist_entry = NULL;
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
use a REG macro that was already defined. Reorder how the pci probing in handled. before adding devices, check to see if the slot is a multi-function device to see if we should probe all the functions. Original idea by: imp
2003-06-22 02:26:17 +00:00
if (REG(PCIR_DEVVENDOR, 4) != -1) {
Back out M_* changes, per decision of the TRB. Approved by: trb
2003-02-19 05:47:46 +00:00
devlist_entry = malloc(size, M_DEVBUF, M_WAITOK | M_ZERO);
Revive PCIConf. Submitted by: "Kenneth D. Merry" <ken@plutotech.com>
1998-09-15 08:21:13 +00:00
if (devlist_entry == NULL)
return (NULL);
cfg = &devlist_entry->cfg;
Add support for multiple PCI "hoses" used on various alpha platforms. The specific intent of this commit is to pave the way for importing Compaq XP1000 support. These changes should not affect the i386 port. Reviewed by: Doug Rabson <dfr@nlsystems.com> (actually, he walked me through most of it & deserves more than reviewd-by credit )
1999-05-20 15:33:33 +00:00
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
cfg->bus = b;
cfg->slot = s;
cfg->func = f;
cfg->vendor = REG(PCIR_VENDOR, 2);
cfg->device = REG(PCIR_DEVICE, 2);
cfg->cmdreg = REG(PCIR_COMMAND, 2);
cfg->statreg = REG(PCIR_STATUS, 2);
cfg->baseclass = REG(PCIR_CLASS, 1);
cfg->subclass = REG(PCIR_SUBCLASS, 1);
cfg->progif = REG(PCIR_PROGIF, 1);
cfg->revid = REG(PCIR_REVID, 1);
- Rename PCIx_HEADERTYPE* to PCIx_HDRTYPE* so the constants aren't so long. - Add a new PCIM_HDRTYPE constant for the field in PCIR_HDRTYPE that holds the header type. - Replace several magic numbers with appropriate constants for the header type register and a couple of PCI_FUNCMAX. - Merge to amd64 the fix to the i386 bridge code to skip devices with unknown header types. Requested by: imp (1, 2)
2003-08-28 21:22:25 +00:00
cfg->hdrtype = REG(PCIR_HDRTYPE, 1);
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
cfg->cachelnsz = REG(PCIR_CACHELNSZ, 1);
cfg->lattimer = REG(PCIR_LATTIMER, 1);
cfg->intpin = REG(PCIR_INTPIN, 1);
cfg->intline = REG(PCIR_INTLINE, 1);
Add support for APIC_IO to pci IRQ configuration. The support for APIC_IO was lost in the new set of pci modules. This patch restores the ability to build SMP/APIC_IO kernels.
1997-05-27 04:09:01 +00:00
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
cfg->mingnt = REG(PCIR_MINGNT, 1);
cfg->maxlat = REG(PCIR_MAXLAT, 1);
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
cfg->mfdev = (cfg->hdrtype & PCIM_MFDEV) != 0;
cfg->hdrtype &= ~PCIM_MFDEV;
pci_fixancient(cfg);
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
pci_hdrtypedata(pcib, b, s, f, cfg);
Revive PCIConf. Submitted by: "Kenneth D. Merry" <ken@plutotech.com>
1998-09-15 08:21:13 +00:00
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
if (REG(PCIR_STATUS, 2) & PCIM_STATUS_CAPPRESENT)
pci_read_extcap(pcib, cfg);
Revive PCIConf. Submitted by: "Kenneth D. Merry" <ken@plutotech.com>
1998-09-15 08:21:13 +00:00
STAILQ_INSERT_TAIL(devlist_head, devlist_entry, pci_links);
devlist_entry->conf.pc_sel.pc_bus = cfg->bus;
devlist_entry->conf.pc_sel.pc_dev = cfg->slot;
devlist_entry->conf.pc_sel.pc_func = cfg->func;
devlist_entry->conf.pc_hdr = cfg->hdrtype;
devlist_entry->conf.pc_subvendor = cfg->subvendor;
devlist_entry->conf.pc_subdevice = cfg->subdevice;
devlist_entry->conf.pc_vendor = cfg->vendor;
devlist_entry->conf.pc_device = cfg->device;
devlist_entry->conf.pc_class = cfg->baseclass;
devlist_entry->conf.pc_subclass = cfg->subclass;
devlist_entry->conf.pc_progif = cfg->progif;
devlist_entry->conf.pc_revid = cfg->revid;
pci_numdevs++;
pci_generation++;
Make mapping messages depend on bootverbose flag. Add PCI subclass to unknown device message.
1995-09-07 15:20:53 +00:00
}
Revive PCIConf. Submitted by: "Kenneth D. Merry" <ken@plutotech.com>
1998-09-15 08:21:13 +00:00
return (devlist_entry);
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
#undef REG
Completely new PCI code: 1) Supports PCI to PCI bridge devices (and tries to initialise them, even if the BIOS is brain dead). 2) Supports shared PCI interrupts. Interrupt handlers now MUST return '0' if they found nothing to do, '1' otherwise. New features tested with i486 systems based on the Intel Saturn and a DEC 4channel Ethernet card only, but expected to work on most systems. The option PCI_REMAP has been removed ! Submitted by: Wolfgang Stanglmeier <wolf@kintaro.cologne.de>
1995-03-21 23:01:06 +00:00
}
These don't need to be semi-public after all.
2003-02-17 19:47:02 +00:00
static void
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
pci_read_extcap(device_t pcib, pcicfgregs *cfg)
{
#define REG(n, w) PCIB_READ_CONFIG(pcib, cfg->bus, cfg->slot, cfg->func, n, w)
int ptr, nextptr, ptrptr;
Remove most of the magic constants from the extcap parsing code.
2003-09-14 06:23:19 +00:00
switch (cfg->hdrtype & PCIM_HDRTYPE) {
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
case 0:
Remove most of the magic constants from the extcap parsing code.
2003-09-14 06:23:19 +00:00
ptrptr = PCIR_CAP_PTR;
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
break;
case 2:
Use symbolic name rather thanhard coding the cap pointer offset for type two devices.
2005-10-29 05:49:06 +00:00
ptrptr = PCIR_CAP_PTR_2;
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
break;
default:
return; /* no extended capabilities support */
}
nextptr = REG(ptrptr, 1); /* sanity check? */
/*
* Read capability entries.
*/
while (nextptr != 0) {
Sanity check ptr for legal values so it is less likely (but not impossible) to get stuck in an infinite loop. Obtained from: msmith@freebsd.org
2001-01-01 16:49:31 +00:00
/* Sanity check */
if (nextptr > 255) {
printf("illegal PCI extended capability offset %d\n",
nextptr);
return;
}
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
/* Find the next entry */
ptr = nextptr;
Add a new method PCI_FIND_EXTCAP() to the pci bus interface that is used to search for a specific extended capability. If the specified capability is found for the given device, then the function returns success and optionally returns the offset of that capability. If the capability is not found, the function returns an error.
2005-12-20 19:57:47 +00:00
nextptr = REG(ptr + PCICAP_NEXTPTR, 1);
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
/* Process this entry */
Add a new method PCI_FIND_EXTCAP() to the pci bus interface that is used to search for a specific extended capability. If the specified capability is found for the given device, then the function returns success and optionally returns the offset of that capability. If the capability is not found, the function returns an error.
2005-12-20 19:57:47 +00:00
switch (REG(ptr + PCICAP_ID, 1)) {
Remove most of the magic constants from the extcap parsing code.
2003-09-14 06:23:19 +00:00
case PCIY_PMG: /* PCI power management */
Teach the PCI code to parse MSI extended capabilities. Re-arrange the pcicfg struct a bit to hold extcap structures instead of structure members.
2003-09-14 19:30:00 +00:00
if (cfg->pp.pp_cap == 0) {
cfg->pp.pp_cap = REG(ptr + PCIR_POWER_CAP, 2);
cfg->pp.pp_status = ptr + PCIR_POWER_STATUS;
cfg->pp.pp_pmcsr = ptr + PCIR_POWER_PMCSR;
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
if ((nextptr - ptr) > PCIR_POWER_DATA)
Teach the PCI code to parse MSI extended capabilities. Re-arrange the pcicfg struct a bit to hold extcap structures instead of structure members.
2003-09-14 19:30:00 +00:00
cfg->pp.pp_data = ptr + PCIR_POWER_DATA;
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
}
break;
Teach the PCI code to parse MSI extended capabilities. Re-arrange the pcicfg struct a bit to hold extcap structures instead of structure members.
2003-09-14 19:30:00 +00:00
case PCIY_MSI: /* PCI MSI */
cfg->msi.msi_ctrl = REG(ptr + PCIR_MSI_CTRL, 2);
if (cfg->msi.msi_ctrl & PCIM_MSICTRL_64BIT)
cfg->msi.msi_data = PCIR_MSI_DATA_64BIT;
else
cfg->msi.msi_data = PCIR_MSI_DATA;
cfg->msi.msi_msgnum = 1 << ((cfg->msi.msi_ctrl &
PCIM_MSICTRL_MMC_MASK)>>1);
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
default:
break;
}
}
#undef REG
}
Add a new method PCI_FIND_EXTCAP() to the pci bus interface that is used to search for a specific extended capability. If the specified capability is found for the given device, then the function returns success and optionally returns the offset of that capability. If the capability is not found, the function returns an error.
2005-12-20 19:57:47 +00:00
/*
* Return the offset in configuration space of the requested extended
* capability entry or 0 if the specified capability was not found.
*/
int
pci_find_extcap_method(device_t dev, device_t child, int capability,
int *capreg)
{
struct pci_devinfo *dinfo = device_get_ivars(child);
pcicfgregs *cfg = &dinfo->cfg;
u_int32_t status;
u_int8_t ptr;
/*
* Check the CAP_LIST bit of the PCI status register first.
*/
status = pci_read_config(child, PCIR_STATUS, 2);
if (!(status & PCIM_STATUS_CAPPRESENT))
return (ENXIO);
/*
* Determine the start pointer of the capabilities list.
*/
switch (cfg->hdrtype & PCIM_HDRTYPE) {
case 0:
ptr = PCIR_CAP_PTR;
break;
case 2:
ptr = PCIR_CAP_PTR_2;
break;
default:
/* XXX: panic? */
return (ENXIO); /* no extended capabilities support */
}
ptr = pci_read_config(child, ptr, 1);
/*
* Traverse the capabilities list.
*/
while (ptr != 0) {
if (pci_read_config(child, ptr + PCICAP_ID, 1) == capability) {
if (capreg != NULL)
*capreg = ptr;
return (0);
}
ptr = pci_read_config(child, ptr + PCICAP_NEXTPTR, 1);
}
return (ENOENT);
}
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
/* free pcicfgregs structure and all depending data structures */
Completely new PCI code: 1) Supports PCI to PCI bridge devices (and tries to initialise them, even if the BIOS is brain dead). 2) Supports shared PCI interrupts. Interrupt handlers now MUST return '0' if they found nothing to do, '1' otherwise. New features tested with i486 systems based on the Intel Saturn and a DEC 4channel Ethernet card only, but expected to work on most systems. The option PCI_REMAP has been removed ! Submitted by: Wolfgang Stanglmeier <wolf@kintaro.cologne.de>
1995-03-21 23:01:06 +00:00
Use the pci.c code wherever possible, rather than copying all the pci code into cardbus and s/pci/cardbus. This exposes a few pci_* functions that are now static. This work is similar to work Justin posted to the mobile list about a year or two ago, which I have neglected since then. This is a subset of his current work with the multiple inheritance newbus architecutre. When completed, that will eliminate the need for pci/pci_private.h. Similar work is needed for the cardbus_cis and pccard_cis code as well.
2002-02-27 05:09:14 +00:00
int
Revive PCIConf. Submitted by: "Kenneth D. Merry" <ken@plutotech.com>
1998-09-15 08:21:13 +00:00
pci_freecfg(struct pci_devinfo *dinfo)
Completely new PCI code: 1) Supports PCI to PCI bridge devices (and tries to initialise them, even if the BIOS is brain dead). 2) Supports shared PCI interrupts. Interrupt handlers now MUST return '0' if they found nothing to do, '1' otherwise. New features tested with i486 systems based on the Intel Saturn and a DEC 4channel Ethernet card only, but expected to work on most systems. The option PCI_REMAP has been removed ! Submitted by: Wolfgang Stanglmeier <wolf@kintaro.cologne.de>
1995-03-21 23:01:06 +00:00
{
Revive PCIConf. Submitted by: "Kenneth D. Merry" <ken@plutotech.com>
1998-09-15 08:21:13 +00:00
struct devlist *devlist_head;
devlist_head = &pci_devq;
Back out the previous change to the queue(3) interface. It was not discussed and should probably not happen. Requested by: msmith and others
2000-05-26 02:09:24 +00:00
STAILQ_REMOVE(devlist_head, dinfo, pci_devinfo, pci_links);
Revive PCIConf. Submitted by: "Kenneth D. Merry" <ken@plutotech.com>
1998-09-15 08:21:13 +00:00
free(dinfo, M_DEVBUF);
/* increment the generation count */
pci_generation++;
/* we're losing one device */
pci_numdevs--;
Completely replace the PCI bus driver code to make it better reflect reality. There will be a new call interface, but for now the file pci_compat.c (which is to be deleted, after all drivers are converted) provides an emulation of the old PCI bus driver functions. The only change that might be visible to drivers is, that the type pcici_t (which had been meant to be just a handle, whose exact definition should not be relied on), has been converted into a pcicfgregs* . The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t and has been converted to just call the PCI drivers functions to access configuration space register, instead of inventing its own ... This code is by no means complete, but assumed to be fully operational, and brings the official code base more in line with my development code. A new generic device descriptor data type has to be agreed on. The PCI code will then use that data type to provide new functionality: 1) userconfig support 2) "wired" PCI devices 3) conflicts checking against ISA/EISA 4) maps will depend on the command register enable bits 5) PCI to Anything bridges can be defined as devices, and are probed like any "standard" PCI device. The following features are currently missing, but will be added back, soon: 1) unknown device probe message 2) suppression of "mirrored" devices caused by ancient, broken chip-sets This code relies on generic shared interrupt support just commited to kern_intr.c (plus the modifications of isa.c and isa_device.h).
1997-05-26 15:08:43 +00:00
return (0);
Completely new PCI code: 1) Supports PCI to PCI bridge devices (and tries to initialise them, even if the BIOS is brain dead). 2) Supports shared PCI interrupts. Interrupt handlers now MUST return '0' if they found nothing to do, '1' otherwise. New features tested with i486 systems based on the Intel Saturn and a DEC 4channel Ethernet card only, but expected to work on most systems. The option PCI_REMAP has been removed ! Submitted by: Wolfgang Stanglmeier <wolf@kintaro.cologne.de>
1995-03-21 23:01:06 +00:00
}
Add support for ioctl() accesses to PCI config space registers. Garrett Wollman sent me this code a few weeks ago for review, and I made some significant changes, which he in turn accepted ... In order to make use of these changes, a device entry has to added to /dev. Submitted by: wollman
1996-10-22 20:20:14 +00:00
/*
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
* PCI power manangement
Add support for ioctl() accesses to PCI config space registers. Garrett Wollman sent me this code a few weeks ago for review, and I made some significant changes, which he in turn accepted ... In order to make use of these changes, a device entry has to added to /dev. Submitted by: wollman
1996-10-22 20:20:14 +00:00
*/
Use the pci.c code wherever possible, rather than copying all the pci code into cardbus and s/pci/cardbus. This exposes a few pci_* functions that are now static. This work is similar to work Justin posted to the mobile list about a year or two ago, which I have neglected since then. This is a subset of his current work with the multiple inheritance newbus architecutre. When completed, that will eliminate the need for pci/pci_private.h. Similar work is needed for the cardbus_cis and pccard_cis code as well.
2002-02-27 05:09:14 +00:00
int
Slightly reimplement some recently added helper functions as methods, so that drivers are not reaching into the internals of the pci bus. There are no driver changes, the public interface is the same.
2001-02-27 23:13:20 +00:00
pci_set_powerstate_method(device_t dev, device_t child, int state)
Add support for ioctl() accesses to PCI config space registers. Garrett Wollman sent me this code a few weeks ago for review, and I made some significant changes, which he in turn accepted ... In order to make use of these changes, a device entry has to added to /dev. Submitted by: wollman
1996-10-22 20:20:14 +00:00
{
Slightly reimplement some recently added helper functions as methods, so that drivers are not reaching into the internals of the pci bus. There are no driver changes, the public interface is the same.
2001-02-27 23:13:20 +00:00
struct pci_devinfo *dinfo = device_get_ivars(child);
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
pcicfgregs *cfg = &dinfo->cfg;
Prefer the uintXX_t to the u_intXX_t names.
2003-08-22 03:11:53 +00:00
uint16_t status;
Implement mimimum system software delays, per PCI PM 1.1 spec, as suggested by Peter Edwards. This seems to fix my fxp problems and likely will fix his as well. Use DELAY rather than *sleep because we can be called from any context.
2004-12-31 20:43:46 +00:00
int result, oldstate, highest, delay;
if (cfg->pp.pp_cap == 0)
return (EOPNOTSUPP);
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
Add system tunable to turn off power state changes. Default to off until we get the resource allocation stuff hammered out. Fix and off by one error that caused unnecessary filtering of valid BARs for only 4 bytes than ICH3 and other PCI IDE controllers have. Andrew Gallatin submitted this, although it doesn't solve the problems ICH3 controllers have with the new code, it does restore the former resource list on the probe line.
2004-04-11 07:02:49 +00:00
/*
Implement mimimum system software delays, per PCI PM 1.1 spec, as suggested by Peter Edwards. This seems to fix my fxp problems and likely will fix his as well. Use DELAY rather than *sleep because we can be called from any context.
2004-12-31 20:43:46 +00:00
* Optimize a no state change request away. While it would be OK to
* write to the hardware in theory, some devices have shown odd
* behavior when going from D3 -> D3.
Add system tunable to turn off power state changes. Default to off until we get the resource allocation stuff hammered out. Fix and off by one error that caused unnecessary filtering of valid BARs for only 4 bytes than ICH3 and other PCI IDE controllers have. Andrew Gallatin submitted this, although it doesn't solve the problems ICH3 controllers have with the new code, it does restore the former resource list on the probe line.
2004-04-11 07:02:49 +00:00
*/
Implement mimimum system software delays, per PCI PM 1.1 spec, as suggested by Peter Edwards. This seems to fix my fxp problems and likely will fix his as well. Use DELAY rather than *sleep because we can be called from any context.
2004-12-31 20:43:46 +00:00
oldstate = pci_get_powerstate(child);
if (oldstate == state)
Add system tunable to turn off power state changes. Default to off until we get the resource allocation stuff hammered out. Fix and off by one error that caused unnecessary filtering of valid BARs for only 4 bytes than ICH3 and other PCI IDE controllers have. Andrew Gallatin submitted this, although it doesn't solve the problems ICH3 controllers have with the new code, it does restore the former resource list on the probe line.
2004-04-11 07:02:49 +00:00
return (0);
Implement mimimum system software delays, per PCI PM 1.1 spec, as suggested by Peter Edwards. This seems to fix my fxp problems and likely will fix his as well. Use DELAY rather than *sleep because we can be called from any context.
2004-12-31 20:43:46 +00:00
/*
* The PCI power management specification states that after a state
* transition between PCI power states, system software must
* guarantee a minimal delay before the function accesses the device.
* Compute the worst case delay that we need to guarantee before we
* access the device. Many devices will be responsive much more
* quickly than this delay, but there are some that don't respond
* instantly to state changes. Transitions to/from D3 state require
* 10ms, while D2 requires 200us, and D0/1 require none. The delay
* is done below with DELAY rather than a sleeper function because
* this function can be called from contexts where we cannot sleep.
*/
highest = (oldstate > state) ? oldstate : state;
Fix last second typos that crept in :-(.
2004-12-31 23:59:24 +00:00
if (highest == PCI_POWERSTATE_D3)
Implement mimimum system software delays, per PCI PM 1.1 spec, as suggested by Peter Edwards. This seems to fix my fxp problems and likely will fix his as well. Use DELAY rather than *sleep because we can be called from any context.
2004-12-31 20:43:46 +00:00
delay = 10000;
Fix last second typos that crept in :-(.
2004-12-31 23:59:24 +00:00
else if (highest == PCI_POWERSTATE_D2)
Implement mimimum system software delays, per PCI PM 1.1 spec, as suggested by Peter Edwards. This seems to fix my fxp problems and likely will fix his as well. Use DELAY rather than *sleep because we can be called from any context.
2004-12-31 20:43:46 +00:00
delay = 200;
else
delay = 0;
status = PCI_READ_CONFIG(dev, child, cfg->pp.pp_status, 2)
& ~PCIM_PSTAT_DMASK;
result = 0;
switch (state) {
case PCI_POWERSTATE_D0:
status |= PCIM_PSTAT_D0;
break;
case PCI_POWERSTATE_D1:
if ((cfg->pp.pp_cap & PCIM_PCAP_D1SUPP) == 0)
return (EOPNOTSUPP);
status |= PCIM_PSTAT_D1;
break;
case PCI_POWERSTATE_D2:
if ((cfg->pp.pp_cap & PCIM_PCAP_D2SUPP) == 0)
return (EOPNOTSUPP);
status |= PCIM_PSTAT_D2;
break;
case PCI_POWERSTATE_D3:
status |= PCIM_PSTAT_D3;
break;
default:
return (EINVAL);
On the new Meteor cards, the Philips SAA 7116 is connected to the PCI bus via an IBM PCI-PCI bridge (82351 or 82352 or 82353) The driver must identify if it is on a secondary PCI bus, which is created via the IBM PCI-PCI bridge. If it is, then it must initialise the IBM PCI-PCI bridge correctly. To do this, the following new functions are added. Because they use the pcici_t tag, they are considered 2.2 compatibility APIs pcici_t * pci_get_parent_from_tag(pcici_t tag); int pci_get_bus_from_tag(pcici_t tag); (The _from_tag suffix is used to prevent clashes with similarly named newbus PCI API functions) Submitted by: Anton Berezin <tobez@plab.ku.dk> Reviewed by: Doug Rabson <dfr@nlsystems.com> Reworked by: Me (roger)
1999-05-31 22:13:37 +00:00
}
move the statement about switching power states to just before we do it, so we don't print a false statement if the destination powerstate is unsupported...
2005-04-01 16:22:50 +00:00
if (bootverbose)
printf(
"pci%d:%d:%d: Transition from D%d to D%d\n",
dinfo->cfg.bus, dinfo->cfg.slot, dinfo->cfg.func,
oldstate, state);
Implement mimimum system software delays, per PCI PM 1.1 spec, as suggested by Peter Edwards. This seems to fix my fxp problems and likely will fix his as well. Use DELAY rather than *sleep because we can be called from any context.
2004-12-31 20:43:46 +00:00
PCI_WRITE_CONFIG(dev, child, cfg->pp.pp_status, status, 2);
if (delay)
DELAY(delay);
return (0);
On the new Meteor cards, the Philips SAA 7116 is connected to the PCI bus via an IBM PCI-PCI bridge (82351 or 82352 or 82353) The driver must identify if it is on a secondary PCI bus, which is created via the IBM PCI-PCI bridge. If it is, then it must initialise the IBM PCI-PCI bridge correctly. To do this, the following new functions are added. Because they use the pcici_t tag, they are considered 2.2 compatibility APIs pcici_t * pci_get_parent_from_tag(pcici_t tag); int pci_get_bus_from_tag(pcici_t tag); (The _from_tag suffix is used to prevent clashes with similarly named newbus PCI API functions) Submitted by: Anton Berezin <tobez@plab.ku.dk> Reviewed by: Doug Rabson <dfr@nlsystems.com> Reworked by: Me (roger)
1999-05-31 22:13:37 +00:00
}
Use the pci.c code wherever possible, rather than copying all the pci code into cardbus and s/pci/cardbus. This exposes a few pci_* functions that are now static. This work is similar to work Justin posted to the mobile list about a year or two ago, which I have neglected since then. This is a subset of his current work with the multiple inheritance newbus architecutre. When completed, that will eliminate the need for pci/pci_private.h. Similar work is needed for the cardbus_cis and pccard_cis code as well.
2002-02-27 05:09:14 +00:00
int
Slightly reimplement some recently added helper functions as methods, so that drivers are not reaching into the internals of the pci bus. There are no driver changes, the public interface is the same.
2001-02-27 23:13:20 +00:00
pci_get_powerstate_method(device_t dev, device_t child)
Add support for ioctl() accesses to PCI config space registers. Garrett Wollman sent me this code a few weeks ago for review, and I made some significant changes, which he in turn accepted ... In order to make use of these changes, a device entry has to added to /dev. Submitted by: wollman
1996-10-22 20:20:14 +00:00
{
Slightly reimplement some recently added helper functions as methods, so that drivers are not reaching into the internals of the pci bus. There are no driver changes, the public interface is the same.
2001-02-27 23:13:20 +00:00
struct pci_devinfo *dinfo = device_get_ivars(child);
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
pcicfgregs *cfg = &dinfo->cfg;
Prefer the uintXX_t to the u_intXX_t names.
2003-08-22 03:11:53 +00:00
uint16_t status;
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
int result;
Teach the PCI code to parse MSI extended capabilities. Re-arrange the pcicfg struct a bit to hold extcap structures instead of structure members.
2003-09-14 19:30:00 +00:00
if (cfg->pp.pp_cap != 0) {
status = PCI_READ_CONFIG(dev, child, cfg->pp.pp_status, 2);
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
switch (status & PCIM_PSTAT_DMASK) {
case PCIM_PSTAT_D0:
result = PCI_POWERSTATE_D0;
Revive PCIConf. Submitted by: "Kenneth D. Merry" <ken@plutotech.com>
1998-09-15 08:21:13 +00:00
break;
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
case PCIM_PSTAT_D1:
result = PCI_POWERSTATE_D1;
Revive PCIConf. Submitted by: "Kenneth D. Merry" <ken@plutotech.com>
1998-09-15 08:21:13 +00:00
break;
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
case PCIM_PSTAT_D2:
result = PCI_POWERSTATE_D2;
Revive PCIConf. Submitted by: "Kenneth D. Merry" <ken@plutotech.com>
1998-09-15 08:21:13 +00:00
break;
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
case PCIM_PSTAT_D3:
result = PCI_POWERSTATE_D3;
Revive PCIConf. Submitted by: "Kenneth D. Merry" <ken@plutotech.com>
1998-09-15 08:21:13 +00:00
break;
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
default:
result = PCI_POWERSTATE_UNKNOWN;
Revive PCIConf. Submitted by: "Kenneth D. Merry" <ken@plutotech.com>
1998-09-15 08:21:13 +00:00
break;
}
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
} else {
/* No support, device is always at D0 */
result = PCI_POWERSTATE_D0;
}
Implement mimimum system software delays, per PCI PM 1.1 spec, as suggested by Peter Edwards. This seems to fix my fxp problems and likely will fix his as well. Use DELAY rather than *sleep because we can be called from any context.
2004-12-31 20:43:46 +00:00
return (result);
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
}
Revive PCIConf. Submitted by: "Kenneth D. Merry" <ken@plutotech.com>
1998-09-15 08:21:13 +00:00
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
/*
* Some convenience functions for PCI device drivers.
*/
Revive PCIConf. Submitted by: "Kenneth D. Merry" <ken@plutotech.com>
1998-09-15 08:21:13 +00:00
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
static __inline void
Prefer the uintXX_t to the u_intXX_t names.
2003-08-22 03:11:53 +00:00
pci_set_command_bit(device_t dev, device_t child, uint16_t bit)
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
{
Prefer the uintXX_t to the u_intXX_t names.
2003-08-22 03:11:53 +00:00
uint16_t command;
Revive PCIConf. Submitted by: "Kenneth D. Merry" <ken@plutotech.com>
1998-09-15 08:21:13 +00:00
Make this file mostly conform to style(9). Approved by: msmith in principle before walkabout
2002-06-01 03:41:02 +00:00
command = PCI_READ_CONFIG(dev, child, PCIR_COMMAND, 2);
command |= bit;
PCI_WRITE_CONFIG(dev, child, PCIR_COMMAND, command, 2);
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
}
Revive PCIConf. Submitted by: "Kenneth D. Merry" <ken@plutotech.com>
1998-09-15 08:21:13 +00:00
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
static __inline void
Prefer the uintXX_t to the u_intXX_t names.
2003-08-22 03:11:53 +00:00
pci_clear_command_bit(device_t dev, device_t child, uint16_t bit)
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
{
Prefer the uintXX_t to the u_intXX_t names.
2003-08-22 03:11:53 +00:00
uint16_t command;
Add support for ioctl() accesses to PCI config space registers. Garrett Wollman sent me this code a few weeks ago for review, and I made some significant changes, which he in turn accepted ... In order to make use of these changes, a device entry has to added to /dev. Submitted by: wollman
1996-10-22 20:20:14 +00:00
Make this file mostly conform to style(9). Approved by: msmith in principle before walkabout
2002-06-01 03:41:02 +00:00
command = PCI_READ_CONFIG(dev, child, PCIR_COMMAND, 2);
command &= ~bit;
PCI_WRITE_CONFIG(dev, child, PCIR_COMMAND, command, 2);
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
}
Add PCI LKM support.
1997-01-21 23:23:40 +00:00
Return status for PCI methods '{enable,disable}_{io,busmaster}'. Reviewed by: imp
2003-04-16 03:15:08 +00:00
int
Slightly reimplement some recently added helper functions as methods, so that drivers are not reaching into the internals of the pci bus. There are no driver changes, the public interface is the same.
2001-02-27 23:13:20 +00:00
pci_enable_busmaster_method(device_t dev, device_t child)
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
{
Make this file mostly conform to style(9). Approved by: msmith in principle before walkabout
2002-06-01 03:41:02 +00:00
pci_set_command_bit(dev, child, PCIM_CMD_BUSMASTEREN);
Return status for PCI methods '{enable,disable}_{io,busmaster}'. Reviewed by: imp
2003-04-16 03:15:08 +00:00
return (0);
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
}
Add support for ioctl() accesses to PCI config space registers. Garrett Wollman sent me this code a few weeks ago for review, and I made some significant changes, which he in turn accepted ... In order to make use of these changes, a device entry has to added to /dev. Submitted by: wollman
1996-10-22 20:20:14 +00:00
Return status for PCI methods '{enable,disable}_{io,busmaster}'. Reviewed by: imp
2003-04-16 03:15:08 +00:00
int
Slightly reimplement some recently added helper functions as methods, so that drivers are not reaching into the internals of the pci bus. There are no driver changes, the public interface is the same.
2001-02-27 23:13:20 +00:00
pci_disable_busmaster_method(device_t dev, device_t child)
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
{
Make this file mostly conform to style(9). Approved by: msmith in principle before walkabout
2002-06-01 03:41:02 +00:00
pci_clear_command_bit(dev, child, PCIM_CMD_BUSMASTEREN);
Return status for PCI methods '{enable,disable}_{io,busmaster}'. Reviewed by: imp
2003-04-16 03:15:08 +00:00
return (0);
Add support for ioctl() accesses to PCI config space registers. Garrett Wollman sent me this code a few weeks ago for review, and I made some significant changes, which he in turn accepted ... In order to make use of these changes, a device entry has to added to /dev. Submitted by: wollman
1996-10-22 20:20:14 +00:00
}
Return status for PCI methods '{enable,disable}_{io,busmaster}'. Reviewed by: imp
2003-04-16 03:15:08 +00:00
int
Slightly reimplement some recently added helper functions as methods, so that drivers are not reaching into the internals of the pci bus. There are no driver changes, the public interface is the same.
2001-02-27 23:13:20 +00:00
pci_enable_io_method(device_t dev, device_t child, int space)
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
{
Prefer the uintXX_t to the u_intXX_t names.
2003-08-22 03:11:53 +00:00
uint16_t command;
uint16_t bit;
Return status for PCI methods '{enable,disable}_{io,busmaster}'. Reviewed by: imp
2003-04-16 03:15:08 +00:00
char *error;
bit = 0;
error = NULL;
Make this file mostly conform to style(9). Approved by: msmith in principle before walkabout
2002-06-01 03:41:02 +00:00
switch(space) {
case SYS_RES_IOPORT:
Return status for PCI methods '{enable,disable}_{io,busmaster}'. Reviewed by: imp
2003-04-16 03:15:08 +00:00
bit = PCIM_CMD_PORTEN;
error = "port";
Make this file mostly conform to style(9). Approved by: msmith in principle before walkabout
2002-06-01 03:41:02 +00:00
break;
case SYS_RES_MEMORY:
Return status for PCI methods '{enable,disable}_{io,busmaster}'. Reviewed by: imp
2003-04-16 03:15:08 +00:00
bit = PCIM_CMD_MEMEN;
error = "memory";
break;
default:
return (EINVAL);
Make this file mostly conform to style(9). Approved by: msmith in principle before walkabout
2002-06-01 03:41:02 +00:00
}
Return status for PCI methods '{enable,disable}_{io,busmaster}'. Reviewed by: imp
2003-04-16 03:15:08 +00:00
pci_set_command_bit(dev, child, bit);
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
/* Some devices seem to need a brief stall here, what do to? */
Return status for PCI methods '{enable,disable}_{io,busmaster}'. Reviewed by: imp
2003-04-16 03:15:08 +00:00
command = PCI_READ_CONFIG(dev, child, PCIR_COMMAND, 2);
if (command & bit)
return (0);
device_printf(child, "failed to enable %s mapping!\n", error);
return (ENXIO);
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
}
Add support for ioctl() accesses to PCI config space registers. Garrett Wollman sent me this code a few weeks ago for review, and I made some significant changes, which he in turn accepted ... In order to make use of these changes, a device entry has to added to /dev. Submitted by: wollman
1996-10-22 20:20:14 +00:00
Return status for PCI methods '{enable,disable}_{io,busmaster}'. Reviewed by: imp
2003-04-16 03:15:08 +00:00
int
Slightly reimplement some recently added helper functions as methods, so that drivers are not reaching into the internals of the pci bus. There are no driver changes, the public interface is the same.
2001-02-27 23:13:20 +00:00
pci_disable_io_method(device_t dev, device_t child, int space)
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
{
Prefer the uintXX_t to the u_intXX_t names.
2003-08-22 03:11:53 +00:00
uint16_t command;
uint16_t bit;
Return status for PCI methods '{enable,disable}_{io,busmaster}'. Reviewed by: imp
2003-04-16 03:15:08 +00:00
char *error;
bit = 0;
error = NULL;
Make this file mostly conform to style(9). Approved by: msmith in principle before walkabout
2002-06-01 03:41:02 +00:00
switch(space) {
case SYS_RES_IOPORT:
Return status for PCI methods '{enable,disable}_{io,busmaster}'. Reviewed by: imp
2003-04-16 03:15:08 +00:00
bit = PCIM_CMD_PORTEN;
error = "port";
Make this file mostly conform to style(9). Approved by: msmith in principle before walkabout
2002-06-01 03:41:02 +00:00
break;
case SYS_RES_MEMORY:
Return status for PCI methods '{enable,disable}_{io,busmaster}'. Reviewed by: imp
2003-04-16 03:15:08 +00:00
bit = PCIM_CMD_MEMEN;
error = "memory";
break;
default:
return (EINVAL);
Make this file mostly conform to style(9). Approved by: msmith in principle before walkabout
2002-06-01 03:41:02 +00:00
}
Return status for PCI methods '{enable,disable}_{io,busmaster}'. Reviewed by: imp
2003-04-16 03:15:08 +00:00
pci_clear_command_bit(dev, child, bit);
command = PCI_READ_CONFIG(dev, child, PCIR_COMMAND, 2);
if (command & bit) {
device_printf(child, "failed to disable %s mapping!\n", error);
return (ENXIO);
}
return (0);
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
}
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
/*
* New style pci driver. Parent device is either a pci-host-bridge or a
* pci-pci-bridge. Both kinds are represented by instances of pcib.
*/
Use the pci.c code wherever possible, rather than copying all the pci code into cardbus and s/pci/cardbus. This exposes a few pci_* functions that are now static. This work is similar to work Justin posted to the mobile list about a year or two ago, which I have neglected since then. This is a subset of his current work with the multiple inheritance newbus architecutre. When completed, that will eliminate the need for pci/pci_private.h. Similar work is needed for the cardbus_cis and pccard_cis code as well.
2002-02-27 05:09:14 +00:00
void
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
pci_print_verbose(struct pci_devinfo *dinfo)
{
if (bootverbose) {
pcicfgregs *cfg = &dinfo->cfg;
printf("found->\tvendor=0x%04x, dev=0x%04x, revid=0x%02x\n",
Make this file mostly conform to style(9). Approved by: msmith in principle before walkabout
2002-06-01 03:41:02 +00:00
cfg->vendor, cfg->device, cfg->revid);
When dumping the 'found devices' list in verbose mode, actually show the bus/slot/function numbers. The old PCI code used other markers or something, but without it here under the new pci code it is very hard to tell which device is which (this only affects bootverbose mode).
2000-09-05 00:32:19 +00:00
printf("\tbus=%d, slot=%d, func=%d\n",
Make this file mostly conform to style(9). Approved by: msmith in principle before walkabout
2002-06-01 03:41:02 +00:00
cfg->bus, cfg->slot, cfg->func);
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
printf("\tclass=%02x-%02x-%02x, hdrtype=0x%02x, mfdev=%d\n",
Make this file mostly conform to style(9). Approved by: msmith in principle before walkabout
2002-06-01 03:41:02 +00:00
cfg->baseclass, cfg->subclass, cfg->progif, cfg->hdrtype,
cfg->mfdev);
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
printf("\tcmdreg=0x%04x, statreg=0x%04x, cachelnsz=%d (dwords)\n",
Make this file mostly conform to style(9). Approved by: msmith in principle before walkabout
2002-06-01 03:41:02 +00:00
cfg->cmdreg, cfg->statreg, cfg->cachelnsz);
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
printf("\tlattimer=0x%02x (%d ns), mingnt=0x%02x (%d ns), maxlat=0x%02x (%d ns)\n",
Make this file mostly conform to style(9). Approved by: msmith in principle before walkabout
2002-06-01 03:41:02 +00:00
cfg->lattimer, cfg->lattimer * 30, cfg->mingnt,
cfg->mingnt * 250, cfg->maxlat, cfg->maxlat * 250);
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
if (cfg->intpin > 0)
Make this file mostly conform to style(9). Approved by: msmith in principle before walkabout
2002-06-01 03:41:02 +00:00
printf("\tintpin=%c, irq=%d\n",
cfg->intpin +'a' -1, cfg->intline);
Teach the PCI code to parse MSI extended capabilities. Re-arrange the pcicfg struct a bit to hold extcap structures instead of structure members.
2003-09-14 19:30:00 +00:00
if (cfg->pp.pp_cap) {
Prefer the uintXX_t to the u_intXX_t names.
2003-08-22 03:11:53 +00:00
uint16_t status;
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
Teach the PCI code to parse MSI extended capabilities. Re-arrange the pcicfg struct a bit to hold extcap structures instead of structure members.
2003-09-14 19:30:00 +00:00
status = pci_read_config(cfg->dev, cfg->pp.pp_status, 2);
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
printf("\tpowerspec %d supports D0%s%s D3 current D%d\n",
Teach the PCI code to parse MSI extended capabilities. Re-arrange the pcicfg struct a bit to hold extcap structures instead of structure members.
2003-09-14 19:30:00 +00:00
cfg->pp.pp_cap & PCIM_PCAP_SPEC,
cfg->pp.pp_cap & PCIM_PCAP_D1SUPP ? " D1" : "",
cfg->pp.pp_cap & PCIM_PCAP_D2SUPP ? " D2" : "",
Make this file mostly conform to style(9). Approved by: msmith in principle before walkabout
2002-06-01 03:41:02 +00:00
status & PCIM_PSTAT_DMASK);
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
}
Teach the PCI code to parse MSI extended capabilities. Re-arrange the pcicfg struct a bit to hold extcap structures instead of structure members.
2003-09-14 19:30:00 +00:00
if (cfg->msi.msi_data) {
int ctrl;
ctrl = cfg->msi.msi_ctrl;
printf("\tMSI supports %d message%s%s%s\n",
cfg->msi.msi_msgnum,
(cfg->msi.msi_msgnum == 1) ? "" : "s",
(ctrl & PCIM_MSICTRL_64BIT) ? ", 64 bit" : "",
(ctrl & PCIM_MSICTRL_VECTOR) ? ", vector masks":"");
}
* Implement bus_set/get/delete_resource for pci. * Change the hack used on the alpha for mapping devices into DENSE or BWX memory spaces to a simpler one. Its still a hack and should be a seperate api to explicitly map the resource. * Add $FreeBSD$ as necessary.
1999-10-14 21:38:33 +00:00
}
}
static int
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
pci_porten(device_t pcib, int b, int s, int f)
* Implement bus_set/get/delete_resource for pci. * Change the hack used on the alpha for mapping devices into DENSE or BWX memory spaces to a simpler one. Its still a hack and should be a seperate api to explicitly map the resource. * Add $FreeBSD$ as necessary.
1999-10-14 21:38:33 +00:00
{
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
return (PCIB_READ_CONFIG(pcib, b, s, f, PCIR_COMMAND, 2)
& PCIM_CMD_PORTEN) != 0;
* Implement bus_set/get/delete_resource for pci. * Change the hack used on the alpha for mapping devices into DENSE or BWX memory spaces to a simpler one. Its still a hack and should be a seperate api to explicitly map the resource. * Add $FreeBSD$ as necessary.
1999-10-14 21:38:33 +00:00
}
static int
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
pci_memen(device_t pcib, int b, int s, int f)
* Implement bus_set/get/delete_resource for pci. * Change the hack used on the alpha for mapping devices into DENSE or BWX memory spaces to a simpler one. Its still a hack and should be a seperate api to explicitly map the resource. * Add $FreeBSD$ as necessary.
1999-10-14 21:38:33 +00:00
{
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
return (PCIB_READ_CONFIG(pcib, b, s, f, PCIR_COMMAND, 2)
& PCIM_CMD_MEMEN) != 0;
* Implement bus_set/get/delete_resource for pci. * Change the hack used on the alpha for mapping devices into DENSE or BWX memory spaces to a simpler one. Its still a hack and should be a seperate api to explicitly map the resource. * Add $FreeBSD$ as necessary.
1999-10-14 21:38:33 +00:00
}
Fix some resource allocation peculiarities of the intpm device.
1999-10-28 08:06:59 +00:00
/*
* Add a resource based on a pci map register. Return 1 if the map
* register is a 32bit map register or 2 if it is a 64bit register.
*/
static int
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
pci_add_map(device_t pcib, device_t bus, device_t dev,
Expose pci_add_resources to the outside world, add a 'force' flag to force allocation of unallocated BARs (cardbus uses this to preallocate everything). Add a prefetchmask to allow for busses that get prefetch hints to set them. Addjust pci_add_map and pci_ata_maps to take a new force flag which pci_add_resources will pass in. Implement 'force' in pci_add_map. Write new value of allocated resource into the bar, if the allocation succeeded (we should have done this before, but with the new force the bug was very obvious).
2005-12-30 19:28:26 +00:00
int b, int s, int f, int reg, struct resource_list *rl, int force,
int prefetch)
* Implement bus_set/get/delete_resource for pci. * Change the hack used on the alpha for mapping devices into DENSE or BWX memory spaces to a simpler one. Its still a hack and should be a seperate api to explicitly map the resource. * Add $FreeBSD$ as necessary.
1999-10-14 21:38:33 +00:00
{
Prefer the uintXX_t to the u_intXX_t names.
2003-08-22 03:11:53 +00:00
uint32_t map;
uint64_t base;
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
uint64_t start, end, count;
Prefer the uintXX_t to the u_intXX_t names.
2003-08-22 03:11:53 +00:00
uint8_t ln2size;
uint8_t ln2range;
uint32_t testval;
uint16_t cmd;
Fix some resource allocation peculiarities of the intpm device.
1999-10-28 08:06:59 +00:00
int type;
More consistantly return the correct BAR size. Before, we'd only return the correct bar size if we encountered a 64-bit BAR that had its resources already assigned. If the resources weren't yet assigned, we'd bogusly assume it was a 32-bit bar and return 1.
2005-09-03 23:15:46 +00:00
int barlen;
Expose pci_add_resources to the outside world, add a 'force' flag to force allocation of unallocated BARs (cardbus uses this to preallocate everything). Add a prefetchmask to allow for busses that get prefetch hints to set them. Addjust pci_add_map and pci_ata_maps to take a new force flag which pci_add_resources will pass in. Implement 'force' in pci_add_map. Write new value of allocated resource into the bar, if the allocation succeeded (we should have done this before, but with the new force the bug was very obvious).
2005-12-30 19:28:26 +00:00
struct resource *res;
Fix some resource allocation peculiarities of the intpm device.
1999-10-28 08:06:59 +00:00
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
map = PCIB_READ_CONFIG(pcib, b, s, f, reg, 4);
PCIB_WRITE_CONFIG(pcib, b, s, f, reg, 0xffffffff, 4);
testval = PCIB_READ_CONFIG(pcib, b, s, f, reg, 4);
PCIB_WRITE_CONFIG(pcib, b, s, f, reg, map, 4);
Fix some resource allocation peculiarities of the intpm device.
1999-10-28 08:06:59 +00:00
if (pci_maptype(map) & PCI_MAPMEM)
type = SYS_RES_MEMORY;
else
type = SYS_RES_IOPORT;
ln2size = pci_mapsize(testval);
ln2range = pci_maprange(testval);
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
base = pci_mapbase(map);
More consistantly return the correct BAR size. Before, we'd only return the correct bar size if we encountered a 64-bit BAR that had its resources already assigned. If the resources weren't yet assigned, we'd bogusly assume it was a 32-bit bar and return 1.
2005-09-03 23:15:46 +00:00
barlen = ln2range == 64 ? 2 : 1;
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
/*
* For I/O registers, if bottom bit is set, and the next bit up
* isn't clear, we know we have a BAR that doesn't conform to the
* spec, so ignore it. Also, sanity check the size of the data
Add system tunable to turn off power state changes. Default to off until we get the resource allocation stuff hammered out. Fix and off by one error that caused unnecessary filtering of valid BARs for only 4 bytes than ICH3 and other PCI IDE controllers have. Andrew Gallatin submitted this, although it doesn't solve the problems ICH3 controllers have with the new code, it does restore the former resource list on the probe line.
2004-04-11 07:02:49 +00:00
* areas to the type of memory involved. Memory must be at least
The size of I/O ranges can be anything from 16 bytes to 2G bytes. Lower the minimum for memory mapped I/O from 32 bytes to 16 bytes. This fixes bus enumeration on ia64 now that the Diva auxiliary serial port is attached to.
2006-04-27 04:53:18 +00:00
* 16 bytes in size, while I/O ranges must be at least 4.
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
*/
if ((testval & 0x1) == 0x1 &&
(testval & 0x2) != 0)
More consistantly return the correct BAR size. Before, we'd only return the correct bar size if we encountered a 64-bit BAR that had its resources already assigned. If the resources weren't yet assigned, we'd bogusly assume it was a 32-bit bar and return 1.
2005-09-03 23:15:46 +00:00
return (barlen);
The size of I/O ranges can be anything from 16 bytes to 2G bytes. Lower the minimum for memory mapped I/O from 32 bytes to 16 bytes. This fixes bus enumeration on ia64 now that the Diva auxiliary serial port is attached to.
2006-04-27 04:53:18 +00:00
if ((type == SYS_RES_MEMORY && ln2size < 4) ||
Add system tunable to turn off power state changes. Default to off until we get the resource allocation stuff hammered out. Fix and off by one error that caused unnecessary filtering of valid BARs for only 4 bytes than ICH3 and other PCI IDE controllers have. Andrew Gallatin submitted this, although it doesn't solve the problems ICH3 controllers have with the new code, it does restore the former resource list on the probe line.
2004-04-11 07:02:49 +00:00
(type == SYS_RES_IOPORT && ln2size < 2))
More consistantly return the correct BAR size. Before, we'd only return the correct bar size if we encountered a 64-bit BAR that had its resources already assigned. If the resources weren't yet assigned, we'd bogusly assume it was a 32-bit bar and return 1.
2005-09-03 23:15:46 +00:00
return (barlen);
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
if (ln2range == 64)
Fix some resource allocation peculiarities of the intpm device.
1999-10-28 08:06:59 +00:00
/* Read the other half of a 64bit map register */
Prefer the uintXX_t to the u_intXX_t names.
2003-08-22 03:11:53 +00:00
base |= (uint64_t) PCIB_READ_CONFIG(pcib, b, s, f, reg + 4, 4) << 32;
* Implement bus_set/get/delete_resource for pci. * Change the hack used on the alpha for mapping devices into DENSE or BWX memory spaces to a simpler one. Its still a hack and should be a seperate api to explicitly map the resource. * Add $FreeBSD$ as necessary.
1999-10-14 21:38:33 +00:00
Print the PCI resources even if they are disabled. This shows more clearly when the BIOS is forgetful about initialising the USB controllers.
2000-03-18 19:18:36 +00:00
if (bootverbose) {
printf("\tmap[%02x]: type %x, range %2d, base %08x, size %2d",
Make this file mostly conform to style(9). Approved by: msmith in principle before walkabout
2002-06-01 03:41:02 +00:00
reg, pci_maptype(map), ln2range,
(unsigned int) base, ln2size);
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
if (type == SYS_RES_IOPORT && !pci_porten(pcib, b, s, f))
Print the PCI resources even if they are disabled. This shows more clearly when the BIOS is forgetful about initialising the USB controllers.
2000-03-18 19:18:36 +00:00
printf(", port disabled\n");
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
else if (type == SYS_RES_MEMORY && !pci_memen(pcib, b, s, f))
Print the PCI resources even if they are disabled. This shows more clearly when the BIOS is forgetful about initialising the USB controllers.
2000-03-18 19:18:36 +00:00
printf(", memory disabled\n");
else
printf(", enabled\n");
}
Unconditionally turning on the I/O and memory enable bits in the PCI command register is too aggressive. Revert to the previous behaviour, but leave the new behaviour available as an undocumented option. It's not clear what the Right, Right Thing is to do here, but the more conservative approach is safer.
2000-10-28 23:07:13 +00:00
Don't enable I/O or memory mode in a device's command register if the BAR we are processing has a base address of zero. Note that this will only change behavior for devices where all the BARs of a given type have a base address of 0 since we will enable the appropriate access when we encounter the first BAR with a base that is not 0. Specifically, this allows certain Toshiba laptops to no longer require 'hw.pci.enable_io_modes=0' to avoid hangs during boot. PR: kern/20040 PR: i386/63776 (possibly) PR: i386/68900 (possibly) PR: i386/74532 (possibly) MFC after: 1 week
2005-05-31 21:33:33 +00:00
/*
* If base is 0, then we have problems. It is best to ignore
Typo. Submitted by: njl
2005-06-01 14:07:43 +00:00
* such entries for the moment. These will be allocated later if
Expose pci_add_resources to the outside world, add a 'force' flag to force allocation of unallocated BARs (cardbus uses this to preallocate everything). Add a prefetchmask to allow for busses that get prefetch hints to set them. Addjust pci_add_map and pci_ata_maps to take a new force flag which pci_add_resources will pass in. Implement 'force' in pci_add_map. Write new value of allocated resource into the bar, if the allocation succeeded (we should have done this before, but with the new force the bug was very obvious).
2005-12-30 19:28:26 +00:00
* the driver specifically requests them. However, some
* removable busses look better when all resources are allocated,
* so allow '0' to be overriden.
Treat resources that are 0xfffff.... as being 'unassigned'. Reviewed by: jhb Tested by: Mark Kirkwood MFC After: 3 days
2005-09-01 02:42:34 +00:00
*
Typo in comment.
2005-09-01 16:41:42 +00:00
* Similarly treat maps whose values is the same as the test value
Treat resources that are 0xfffff.... as being 'unassigned'. Reviewed by: jhb Tested by: Mark Kirkwood MFC After: 3 days
2005-09-01 02:42:34 +00:00
* read back. These maps have had all f's written to them by the
* BIOS in an attempt to disable the resources.
Don't enable I/O or memory mode in a device's command register if the BAR we are processing has a base address of zero. Note that this will only change behavior for devices where all the BARs of a given type have a base address of 0 since we will enable the appropriate access when we encounter the first BAR with a base that is not 0. Specifically, this allows certain Toshiba laptops to no longer require 'hw.pci.enable_io_modes=0' to avoid hangs during boot. PR: kern/20040 PR: i386/63776 (possibly) PR: i386/68900 (possibly) PR: i386/74532 (possibly) MFC after: 1 week
2005-05-31 21:33:33 +00:00
*/
Remove stray debug from p4 integration.
2006-01-01 08:26:39 +00:00
if (!force && (base == 0 || map == testval))
More consistantly return the correct BAR size. Before, we'd only return the correct bar size if we encountered a 64-bit BAR that had its resources already assigned. If the resources weren't yet assigned, we'd bogusly assume it was a 32-bit bar and return 1.
2005-09-03 23:15:46 +00:00
return (barlen);
Treat resources that are 0xfffff.... as being 'unassigned'. Reviewed by: jhb Tested by: Mark Kirkwood MFC After: 3 days
2005-09-01 02:42:34 +00:00
Unconditionally turning on the I/O and memory enable bits in the PCI command register is too aggressive. Revert to the previous behaviour, but leave the new behaviour available as an undocumented option. It's not clear what the Right, Right Thing is to do here, but the more conservative approach is safer.
2000-10-28 23:07:13 +00:00
/*
* This code theoretically does the right thing, but has
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
* undesirable side effects in some cases where peripherals
* respond oddly to having these bits enabled. Let the user
* be able to turn them off (since pci_enable_io_modes is 1 by
* default).
Unconditionally turning on the I/O and memory enable bits in the PCI command register is too aggressive. Revert to the previous behaviour, but leave the new behaviour available as an undocumented option. It's not clear what the Right, Right Thing is to do here, but the more conservative approach is safer.
2000-10-28 23:07:13 +00:00
*/
Make PCI_ENABLE_IO_MODES a sysctl hw.pci.enable_io_modes. It can also be set at boot time. It defaults to 1 now since it can be set in the boot loader. If this proves unwise, we can reset it to defaulting to 0.
2002-07-26 07:58:16 +00:00
if (pci_enable_io_modes) {
/* Turn on resources that have been left off by a lazy BIOS */
if (type == SYS_RES_IOPORT && !pci_porten(pcib, b, s, f)) {
cmd = PCIB_READ_CONFIG(pcib, b, s, f, PCIR_COMMAND, 2);
cmd |= PCIM_CMD_PORTEN;
PCIB_WRITE_CONFIG(pcib, b, s, f, PCIR_COMMAND, cmd, 2);
}
if (type == SYS_RES_MEMORY && !pci_memen(pcib, b, s, f)) {
cmd = PCIB_READ_CONFIG(pcib, b, s, f, PCIR_COMMAND, 2);
cmd |= PCIM_CMD_MEMEN;
PCIB_WRITE_CONFIG(pcib, b, s, f, PCIR_COMMAND, cmd, 2);
}
} else {
if (type == SYS_RES_IOPORT && !pci_porten(pcib, b, s, f))
More consistantly return the correct BAR size. Before, we'd only return the correct bar size if we encountered a 64-bit BAR that had its resources already assigned. If the resources weren't yet assigned, we'd bogusly assume it was a 32-bit bar and return 1.
2005-09-03 23:15:46 +00:00
return (barlen);
Make PCI_ENABLE_IO_MODES a sysctl hw.pci.enable_io_modes. It can also be set at boot time. It defaults to 1 now since it can be set in the boot loader. If this proves unwise, we can reset it to defaulting to 0.
2002-07-26 07:58:16 +00:00
if (type == SYS_RES_MEMORY && !pci_memen(pcib, b, s, f))
More consistantly return the correct BAR size. Before, we'd only return the correct bar size if we encountered a 64-bit BAR that had its resources already assigned. If the resources weren't yet assigned, we'd bogusly assume it was a 32-bit bar and return 1.
2005-09-03 23:15:46 +00:00
return (barlen);
If a base address register has been set up by the BIOS, but the relevant enable bit hasn't been set in the command register, set the bit and honour the register. It seems that quite a few lazy BIOS writers aren't bothering to do this, which upsets the existing code and causes us to miss out on properly-configured devices.
2000-09-01 23:09:02 +00:00
}
Do not pre-allocate resources for BAR's on ATA MASTERDEV's thats on the standard ATA primary and secondary addresses. Reintroduce the size 1 ALTIO space so that we can have both ATA and floppies back working.
2004-04-20 20:57:29 +00:00
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
count = 1 << ln2size;
Expose pci_add_resources to the outside world, add a 'force' flag to force allocation of unallocated BARs (cardbus uses this to preallocate everything). Add a prefetchmask to allow for busses that get prefetch hints to set them. Addjust pci_add_map and pci_ata_maps to take a new force flag which pci_add_resources will pass in. Implement 'force' in pci_add_map. Write new value of allocated resource into the bar, if the allocation succeeded (we should have done this before, but with the new force the bug was very obvious).
2005-12-30 19:28:26 +00:00
if (base == 0 || base == pci_mapbase(testval)) {
start = 0; /* Let the parent deside */
end = ~0ULL;
} else {
start = base;
end = base + (1 << ln2size) - 1;
}
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
resource_list_add(rl, type, reg, start, end, count);
* Implement bus_set/get/delete_resource for pci. * Change the hack used on the alpha for mapping devices into DENSE or BWX memory spaces to a simpler one. Its still a hack and should be a seperate api to explicitly map the resource. * Add $FreeBSD$ as necessary.
1999-10-14 21:38:33 +00:00
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
/*
* Not quite sure what to do on failure of allocating the resource
* since I can postulate several right answers.
*/
Expose pci_add_resources to the outside world, add a 'force' flag to force allocation of unallocated BARs (cardbus uses this to preallocate everything). Add a prefetchmask to allow for busses that get prefetch hints to set them. Addjust pci_add_map and pci_ata_maps to take a new force flag which pci_add_resources will pass in. Implement 'force' in pci_add_map. Write new value of allocated resource into the bar, if the allocation succeeded (we should have done this before, but with the new force the bug was very obvious).
2005-12-30 19:28:26 +00:00
res = resource_list_alloc(rl, bus, dev, type, &reg, start, end, count,
prefetch ? RF_PREFETCHABLE : 0);
if (res != NULL)
pci_write_config(dev, reg, rman_get_start(res), 4);
More consistantly return the correct BAR size. Before, we'd only return the correct bar size if we encountered a 64-bit BAR that had its resources already assigned. If the resources weren't yet assigned, we'd bogusly assume it was a 32-bit bar and return 1.
2005-09-03 23:15:46 +00:00
return (barlen);
Fix some resource allocation peculiarities of the intpm device.
1999-10-28 08:06:59 +00:00
}
ata devices in legacy are special, and we must treat them as such. While I would have prefered to have a solution that didn't move knowledge of this into the pci layer. However, this is literally the only exception that's listed in the PCI standard to the usual way of decoding BARs. atapci devices in legacy mode now ignore the first 4 bars and hard code the values to the legacy ide values (well, for each of the controllers that are in legacy mode). The 5th bar is handled normally. Remove the zero bar handling. zero bars should be ignored at all other times, and since we handle that specially, we don't need the older workaround.
2004-04-21 20:19:56 +00:00
/*
Update the special handling code for ATA devices to allow usage of PCI native addressing. That means that if the HW says that using "real" addresses instead of the hardwired legacy compat ones is allowed, we will use them.
2004-06-29 20:25:43 +00:00
* For ATA devices we need to decide early what addressing mode to use.
* Legacy demands that the primary and secondary ATA ports sits on the
* same addresses that old ISA hardware did. This dictates that we use
* those addresses and ignore the BAR's if we cannot set PCI native
* addressing mode.
ata devices in legacy are special, and we must treat them as such. While I would have prefered to have a solution that didn't move knowledge of this into the pci layer. However, this is literally the only exception that's listed in the PCI standard to the usual way of decoding BARs. atapci devices in legacy mode now ignore the first 4 bars and hard code the values to the legacy ide values (well, for each of the controllers that are in legacy mode). The 5th bar is handled normally. Remove the zero bar handling. zero bars should be ignored at all other times, and since we handle that specially, we don't need the older workaround.
2004-04-21 20:19:56 +00:00
*/
static void
Update the special handling code for ATA devices to allow usage of PCI native addressing. That means that if the HW says that using "real" addresses instead of the hardwired legacy compat ones is allowed, we will use them.
2004-06-29 20:25:43 +00:00
pci_ata_maps(device_t pcib, device_t bus, device_t dev, int b,
Expose pci_add_resources to the outside world, add a 'force' flag to force allocation of unallocated BARs (cardbus uses this to preallocate everything). Add a prefetchmask to allow for busses that get prefetch hints to set them. Addjust pci_add_map and pci_ata_maps to take a new force flag which pci_add_resources will pass in. Implement 'force' in pci_add_map. Write new value of allocated resource into the bar, if the allocation succeeded (we should have done this before, but with the new force the bug was very obvious).
2005-12-30 19:28:26 +00:00
int s, int f, struct resource_list *rl, int force, uint32_t prefetchmask)
ata devices in legacy are special, and we must treat them as such. While I would have prefered to have a solution that didn't move knowledge of this into the pci layer. However, this is literally the only exception that's listed in the PCI standard to the usual way of decoding BARs. atapci devices in legacy mode now ignore the first 4 bars and hard code the values to the legacy ide values (well, for each of the controllers that are in legacy mode). The 5th bar is handled normally. Remove the zero bar handling. zero bars should be ignored at all other times, and since we handle that specially, we don't need the older workaround.
2004-04-21 20:19:56 +00:00
{
Update the special handling code for ATA devices to allow usage of PCI native addressing. That means that if the HW says that using "real" addresses instead of the hardwired legacy compat ones is allowed, we will use them.
2004-06-29 20:25:43 +00:00
int rid, type, progif;
Disable native ata support for now, too much breaks
2004-07-02 13:42:36 +00:00
#if 0
Update the special handling code for ATA devices to allow usage of PCI native addressing. That means that if the HW says that using "real" addresses instead of the hardwired legacy compat ones is allowed, we will use them.
2004-06-29 20:25:43 +00:00
/* if this device supports PCI native addressing use it */
progif = pci_read_config(dev, PCIR_PROGIF, 1);
if ((progif & 0x8a) == 0x8a) {
if (pci_mapbase(pci_read_config(dev, PCIR_BAR(0), 4)) &&
pci_mapbase(pci_read_config(dev, PCIR_BAR(2), 4))) {
printf("Trying ATA native PCI addressing mode\n");
pci_write_config(dev, PCIR_PROGIF, progif | 0x05, 1);
}
}
#endif
progif = pci_read_config(dev, PCIR_PROGIF, 1);
ata devices in legacy are special, and we must treat them as such. While I would have prefered to have a solution that didn't move knowledge of this into the pci layer. However, this is literally the only exception that's listed in the PCI standard to the usual way of decoding BARs. atapci devices in legacy mode now ignore the first 4 bars and hard code the values to the legacy ide values (well, for each of the controllers that are in legacy mode). The 5th bar is handled normally. Remove the zero bar handling. zero bars should be ignored at all other times, and since we handle that specially, we don't need the older workaround.
2004-04-21 20:19:56 +00:00
type = SYS_RES_IOPORT;
Update the special handling code for ATA devices to allow usage of PCI native addressing. That means that if the HW says that using "real" addresses instead of the hardwired legacy compat ones is allowed, we will use them.
2004-06-29 20:25:43 +00:00
if (progif & PCIP_STORAGE_IDE_MODEPRIM) {
Expose pci_add_resources to the outside world, add a 'force' flag to force allocation of unallocated BARs (cardbus uses this to preallocate everything). Add a prefetchmask to allow for busses that get prefetch hints to set them. Addjust pci_add_map and pci_ata_maps to take a new force flag which pci_add_resources will pass in. Implement 'force' in pci_add_map. Write new value of allocated resource into the bar, if the allocation succeeded (we should have done this before, but with the new force the bug was very obvious).
2005-12-30 19:28:26 +00:00
pci_add_map(pcib, bus, dev, b, s, f, PCIR_BAR(0), rl, force,
prefetchmask & (1 << 0));
pci_add_map(pcib, bus, dev, b, s, f, PCIR_BAR(1), rl, force,
prefetchmask & (1 << 1));
Minor style(9) nitage.
2005-10-28 05:56:50 +00:00
} else {
ata devices in legacy are special, and we must treat them as such. While I would have prefered to have a solution that didn't move knowledge of this into the pci layer. However, this is literally the only exception that's listed in the PCI standard to the usual way of decoding BARs. atapci devices in legacy mode now ignore the first 4 bars and hard code the values to the legacy ide values (well, for each of the controllers that are in legacy mode). The 5th bar is handled normally. Remove the zero bar handling. zero bars should be ignored at all other times, and since we handle that specially, we don't need the older workaround.
2004-04-21 20:19:56 +00:00
rid = PCIR_BAR(0);
resource_list_add(rl, type, rid, 0x1f0, 0x1f7, 8);
Expose pci_add_resources to the outside world, add a 'force' flag to force allocation of unallocated BARs (cardbus uses this to preallocate everything). Add a prefetchmask to allow for busses that get prefetch hints to set them. Addjust pci_add_map and pci_ata_maps to take a new force flag which pci_add_resources will pass in. Implement 'force' in pci_add_map. Write new value of allocated resource into the bar, if the allocation succeeded (we should have done this before, but with the new force the bug was very obvious).
2005-12-30 19:28:26 +00:00
resource_list_alloc(rl, bus, dev, type, &rid, 0x1f0, 0x1f7, 8,
0);
ata devices in legacy are special, and we must treat them as such. While I would have prefered to have a solution that didn't move knowledge of this into the pci layer. However, this is literally the only exception that's listed in the PCI standard to the usual way of decoding BARs. atapci devices in legacy mode now ignore the first 4 bars and hard code the values to the legacy ide values (well, for each of the controllers that are in legacy mode). The 5th bar is handled normally. Remove the zero bar handling. zero bars should be ignored at all other times, and since we handle that specially, we don't need the older workaround.
2004-04-21 20:19:56 +00:00
rid = PCIR_BAR(1);
resource_list_add(rl, type, rid, 0x3f6, 0x3f6, 1);
Expose pci_add_resources to the outside world, add a 'force' flag to force allocation of unallocated BARs (cardbus uses this to preallocate everything). Add a prefetchmask to allow for busses that get prefetch hints to set them. Addjust pci_add_map and pci_ata_maps to take a new force flag which pci_add_resources will pass in. Implement 'force' in pci_add_map. Write new value of allocated resource into the bar, if the allocation succeeded (we should have done this before, but with the new force the bug was very obvious).
2005-12-30 19:28:26 +00:00
resource_list_alloc(rl, bus, dev, type, &rid, 0x3f6, 0x3f6, 1,
0);
Update the special handling code for ATA devices to allow usage of PCI native addressing. That means that if the HW says that using "real" addresses instead of the hardwired legacy compat ones is allowed, we will use them.
2004-06-29 20:25:43 +00:00
}
if (progif & PCIP_STORAGE_IDE_MODESEC) {
Expose pci_add_resources to the outside world, add a 'force' flag to force allocation of unallocated BARs (cardbus uses this to preallocate everything). Add a prefetchmask to allow for busses that get prefetch hints to set them. Addjust pci_add_map and pci_ata_maps to take a new force flag which pci_add_resources will pass in. Implement 'force' in pci_add_map. Write new value of allocated resource into the bar, if the allocation succeeded (we should have done this before, but with the new force the bug was very obvious).
2005-12-30 19:28:26 +00:00
pci_add_map(pcib, bus, dev, b, s, f, PCIR_BAR(2), rl, force,
prefetchmask & (1 << 2));
pci_add_map(pcib, bus, dev, b, s, f, PCIR_BAR(3), rl, force,
prefetchmask & (1 << 3));
Minor style(9) nitage.
2005-10-28 05:56:50 +00:00
} else {
ata devices in legacy are special, and we must treat them as such. While I would have prefered to have a solution that didn't move knowledge of this into the pci layer. However, this is literally the only exception that's listed in the PCI standard to the usual way of decoding BARs. atapci devices in legacy mode now ignore the first 4 bars and hard code the values to the legacy ide values (well, for each of the controllers that are in legacy mode). The 5th bar is handled normally. Remove the zero bar handling. zero bars should be ignored at all other times, and since we handle that specially, we don't need the older workaround.
2004-04-21 20:19:56 +00:00
rid = PCIR_BAR(2);
resource_list_add(rl, type, rid, 0x170, 0x177, 8);
Expose pci_add_resources to the outside world, add a 'force' flag to force allocation of unallocated BARs (cardbus uses this to preallocate everything). Add a prefetchmask to allow for busses that get prefetch hints to set them. Addjust pci_add_map and pci_ata_maps to take a new force flag which pci_add_resources will pass in. Implement 'force' in pci_add_map. Write new value of allocated resource into the bar, if the allocation succeeded (we should have done this before, but with the new force the bug was very obvious).
2005-12-30 19:28:26 +00:00
resource_list_alloc(rl, bus, dev, type, &rid, 0x170, 0x177, 8,
0);
ata devices in legacy are special, and we must treat them as such. While I would have prefered to have a solution that didn't move knowledge of this into the pci layer. However, this is literally the only exception that's listed in the PCI standard to the usual way of decoding BARs. atapci devices in legacy mode now ignore the first 4 bars and hard code the values to the legacy ide values (well, for each of the controllers that are in legacy mode). The 5th bar is handled normally. Remove the zero bar handling. zero bars should be ignored at all other times, and since we handle that specially, we don't need the older workaround.
2004-04-21 20:19:56 +00:00
rid = PCIR_BAR(3);
resource_list_add(rl, type, rid, 0x376, 0x376, 1);
Expose pci_add_resources to the outside world, add a 'force' flag to force allocation of unallocated BARs (cardbus uses this to preallocate everything). Add a prefetchmask to allow for busses that get prefetch hints to set them. Addjust pci_add_map and pci_ata_maps to take a new force flag which pci_add_resources will pass in. Implement 'force' in pci_add_map. Write new value of allocated resource into the bar, if the allocation succeeded (we should have done this before, but with the new force the bug was very obvious).
2005-12-30 19:28:26 +00:00
resource_list_alloc(rl, bus, dev, type, &rid, 0x376, 0x376, 1,
0);
ata devices in legacy are special, and we must treat them as such. While I would have prefered to have a solution that didn't move knowledge of this into the pci layer. However, this is literally the only exception that's listed in the PCI standard to the usual way of decoding BARs. atapci devices in legacy mode now ignore the first 4 bars and hard code the values to the legacy ide values (well, for each of the controllers that are in legacy mode). The 5th bar is handled normally. Remove the zero bar handling. zero bars should be ignored at all other times, and since we handle that specially, we don't need the older workaround.
2004-04-21 20:19:56 +00:00
}
Expose pci_add_resources to the outside world, add a 'force' flag to force allocation of unallocated BARs (cardbus uses this to preallocate everything). Add a prefetchmask to allow for busses that get prefetch hints to set them. Addjust pci_add_map and pci_ata_maps to take a new force flag which pci_add_resources will pass in. Implement 'force' in pci_add_map. Write new value of allocated resource into the bar, if the allocation succeeded (we should have done this before, but with the new force the bug was very obvious).
2005-12-30 19:28:26 +00:00
pci_add_map(pcib, bus, dev, b, s, f, PCIR_BAR(4), rl, force,
prefetchmask & (1 << 4));
pci_add_map(pcib, bus, dev, b, s, f, PCIR_BAR(5), rl, force,
prefetchmask & (1 << 5));
ata devices in legacy are special, and we must treat them as such. While I would have prefered to have a solution that didn't move knowledge of this into the pci layer. However, this is literally the only exception that's listed in the PCI standard to the usual way of decoding BARs. atapci devices in legacy mode now ignore the first 4 bars and hard code the values to the legacy ide values (well, for each of the controllers that are in legacy mode). The 5th bar is handled normally. Remove the zero bar handling. zero bars should be ignored at all other times, and since we handle that specially, we don't need the older workaround.
2004-04-21 20:19:56 +00:00
}
- Consolidate duplicated code for assigning interrupts to PCI devices via routing, etc. in a static pci_assign_interrupt() function. - Add a sledgehammer that allows the user to override the interrupt assignment of any PCI device via a tunable (e.g. "hw.pci0.7.INTB=5" would force any functions on the pci device in slot 7 of bus 0 that use B# to use IRQ 5). This should be used with great caution! Generally, if the interrupt routing in use provides specific tunables (such as hard-wiring the IRQ for a given $PIR or ACPI PCI link device), then those should be used instead. One instance where this tunable might be useful is if a box has an MPTable with duplicate entries for the same PCI device with different IRQs. MFC after: 1 week
2005-09-29 15:04:41 +00:00
static void
pci_assign_interrupt(device_t bus, device_t dev, int force_route)
{
struct pci_devinfo *dinfo = device_get_ivars(dev);
pcicfgregs *cfg = &dinfo->cfg;
char tunable_name[64];
int irq;
/* Has to have an intpin to have an interrupt. */
if (cfg->intpin == 0)
return;
/* Let the user override the IRQ with a tunable. */
irq = PCI_INVALID_IRQ;
snprintf(tunable_name, sizeof(tunable_name), "hw.pci%d.%d.INT%c.irq",
cfg->bus, cfg->slot, cfg->intpin + 'A' - 1);
if (TUNABLE_INT_FETCH(tunable_name, &irq) && (irq >= 255 || irq <= 0))
irq = PCI_INVALID_IRQ;
/*
* If we didn't get an IRQ via the tunable, then we either use the
* IRQ value in the intline register or we ask the bus to route an
* interrupt for us. If force_route is true, then we only use the
* value in the intline register if the bus was unable to assign an
* IRQ.
*/
if (!PCI_INTERRUPT_VALID(irq)) {
if (!PCI_INTERRUPT_VALID(cfg->intline) || force_route)
irq = PCI_ASSIGN_INTERRUPT(bus, dev);
if (!PCI_INTERRUPT_VALID(irq))
irq = cfg->intline;
}
/* If after all that we don't have an IRQ, just bail. */
if (!PCI_INTERRUPT_VALID(irq))
return;
/* Update the config register if it changed. */
if (irq != cfg->intline) {
cfg->intline = irq;
pci_write_config(dev, PCIR_INTLINE, irq, 1);
}
/* Add this IRQ as rid 0 interrupt resource. */
resource_list_add(&dinfo->resources, SYS_RES_IRQ, 0, irq, irq, 1);
}
Expose pci_add_resources to the outside world, add a 'force' flag to force allocation of unallocated BARs (cardbus uses this to preallocate everything). Add a prefetchmask to allow for busses that get prefetch hints to set them. Addjust pci_add_map and pci_ata_maps to take a new force flag which pci_add_resources will pass in. Implement 'force' in pci_add_map. Write new value of allocated resource into the bar, if the allocation succeeded (we should have done this before, but with the new force the bug was very obvious).
2005-12-30 19:28:26 +00:00
void
pci_add_resources(device_t bus, device_t dev, int force, uint32_t prefetchmask)
Fix some resource allocation peculiarities of the intpm device.
1999-10-28 08:06:59 +00:00
{
Expose pci_add_resources to the outside world, add a 'force' flag to force allocation of unallocated BARs (cardbus uses this to preallocate everything). Add a prefetchmask to allow for busses that get prefetch hints to set them. Addjust pci_add_map and pci_ata_maps to take a new force flag which pci_add_resources will pass in. Implement 'force' in pci_add_map. Write new value of allocated resource into the bar, if the allocation succeeded (we should have done this before, but with the new force the bug was very obvious).
2005-12-30 19:28:26 +00:00
device_t pcib;
Fix some resource allocation peculiarities of the intpm device.
1999-10-28 08:06:59 +00:00
struct pci_devinfo *dinfo = device_get_ivars(dev);
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
pcicfgregs *cfg = &dinfo->cfg;
Fix some resource allocation peculiarities of the intpm device.
1999-10-28 08:06:59 +00:00
struct resource_list *rl = &dinfo->resources;
struct pci_quirk *q;
- Consolidate duplicated code for assigning interrupts to PCI devices via routing, etc. in a static pci_assign_interrupt() function. - Add a sledgehammer that allows the user to override the interrupt assignment of any PCI device via a tunable (e.g. "hw.pci0.7.INTB=5" would force any functions on the pci device in slot 7 of bus 0 that use B# to use IRQ 5). This should be used with great caution! Generally, if the interrupt routing in use provides specific tunables (such as hard-wiring the IRQ for a given $PIR or ACPI PCI link device), then those should be used instead. One instance where this tunable might be useful is if a box has an MPTable with duplicate entries for the same PCI device with different IRQs. MFC after: 1 week
2005-09-29 15:04:41 +00:00
int b, i, f, s;
Fix some resource allocation peculiarities of the intpm device.
1999-10-28 08:06:59 +00:00
Expose pci_add_resources to the outside world, add a 'force' flag to force allocation of unallocated BARs (cardbus uses this to preallocate everything). Add a prefetchmask to allow for busses that get prefetch hints to set them. Addjust pci_add_map and pci_ata_maps to take a new force flag which pci_add_resources will pass in. Implement 'force' in pci_add_map. Write new value of allocated resource into the bar, if the allocation succeeded (we should have done this before, but with the new force the bug was very obvious).
2005-12-30 19:28:26 +00:00
pcib = device_get_parent(bus);
Various changes to make it easier to subclass the PCI bus device. - Make the pci devclass a global variable. - Add child devices in pci_attach() instead of pci_probe(). Change pci_probe() to just check for a valid bus number from the associated bridge and return -1000 if successful. This allows subclasses of the PCI bus driver to override the generic driver. - Move the code to load the vendor data into its own public function. Really though, doing this at attach is just plain wrong. This should really be done in the module load routine instead. As a side effect, the 'busno' variable in pci_attach() is now no longer static (minor bug that was harmless so far.) - Change pci_add_children() to take an extra argument that is the size of the device info structure passed to pci_read_device() and make it public so subclasses of the PCI bus can call it in their attach routines. - Move the bits to attach a probed PCI child to a PCI bus into a global pci_add_child() function. This will allow subclasses that can detect a PCI device not found in the normal PCI probe to add those devices in their own attach routine. (I have seen this in the ACPI tree on my laptop for example.) As a side effect, change the static function pci_add_resources() to get the busno, slot, and func from the passed in dinfo structure instead of requiring them as function arguments. Tested on: i386, alpha, ia64, sparc64
2002-08-26 15:23:52 +00:00
b = cfg->bus;
s = cfg->slot;
f = cfg->func;
ata devices in legacy are special, and we must treat them as such. While I would have prefered to have a solution that didn't move knowledge of this into the pci layer. However, this is literally the only exception that's listed in the PCI standard to the usual way of decoding BARs. atapci devices in legacy mode now ignore the first 4 bars and hard code the values to the legacy ide values (well, for each of the controllers that are in legacy mode). The 5th bar is handled normally. Remove the zero bar handling. zero bars should be ignored at all other times, and since we handle that specially, we don't need the older workaround.
2004-04-21 20:19:56 +00:00
Update the special handling code for ATA devices to allow usage of PCI native addressing. That means that if the HW says that using "real" addresses instead of the hardwired legacy compat ones is allowed, we will use them.
2004-06-29 20:25:43 +00:00
/* ATA devices needs special map treatment */
if ((pci_get_class(dev) == PCIC_STORAGE) &&
(pci_get_subclass(dev) == PCIS_STORAGE_IDE) &&
(pci_get_progif(dev) & PCIP_STORAGE_IDE_MASTERDEV))
Expose pci_add_resources to the outside world, add a 'force' flag to force allocation of unallocated BARs (cardbus uses this to preallocate everything). Add a prefetchmask to allow for busses that get prefetch hints to set them. Addjust pci_add_map and pci_ata_maps to take a new force flag which pci_add_resources will pass in. Implement 'force' in pci_add_map. Write new value of allocated resource into the bar, if the allocation succeeded (we should have done this before, but with the new force the bug was very obvious).
2005-12-30 19:28:26 +00:00
pci_ata_maps(pcib, bus, dev, b, s, f, rl, force, prefetchmask);
ata devices in legacy are special, and we must treat them as such. While I would have prefered to have a solution that didn't move knowledge of this into the pci layer. However, this is literally the only exception that's listed in the PCI standard to the usual way of decoding BARs. atapci devices in legacy mode now ignore the first 4 bars and hard code the values to the legacy ide values (well, for each of the controllers that are in legacy mode). The 5th bar is handled normally. Remove the zero bar handling. zero bars should be ignored at all other times, and since we handle that specially, we don't need the older workaround.
2004-04-21 20:19:56 +00:00
else
for (i = 0; i < cfg->nummaps;)
i += pci_add_map(pcib, bus, dev, b, s, f, PCIR_BAR(i),
Expose pci_add_resources to the outside world, add a 'force' flag to force allocation of unallocated BARs (cardbus uses this to preallocate everything). Add a prefetchmask to allow for busses that get prefetch hints to set them. Addjust pci_add_map and pci_ata_maps to take a new force flag which pci_add_resources will pass in. Implement 'force' in pci_add_map. Write new value of allocated resource into the bar, if the allocation succeeded (we should have done this before, but with the new force the bug was very obvious).
2005-12-30 19:28:26 +00:00
rl, force, prefetchmask & (1 << i));
Fix some resource allocation peculiarities of the intpm device.
1999-10-28 08:06:59 +00:00
Expose pci_add_resources to the outside world, add a 'force' flag to force allocation of unallocated BARs (cardbus uses this to preallocate everything). Add a prefetchmask to allow for busses that get prefetch hints to set them. Addjust pci_add_map and pci_ata_maps to take a new force flag which pci_add_resources will pass in. Implement 'force' in pci_add_map. Write new value of allocated resource into the bar, if the allocation succeeded (we should have done this before, but with the new force the bug was very obvious).
2005-12-30 19:28:26 +00:00
/*
* Add additional, quirked resources.
*/
Fix some resource allocation peculiarities of the intpm device.
1999-10-28 08:06:59 +00:00
for (q = &pci_quirks[0]; q->devid; q++) {
if (q->devid == ((cfg->device << 16) | cfg->vendor)
&& q->type == PCI_QUIRK_MAP_REG)
Expose pci_add_resources to the outside world, add a 'force' flag to force allocation of unallocated BARs (cardbus uses this to preallocate everything). Add a prefetchmask to allow for busses that get prefetch hints to set them. Addjust pci_add_map and pci_ata_maps to take a new force flag which pci_add_resources will pass in. Implement 'force' in pci_add_map. Write new value of allocated resource into the bar, if the allocation succeeded (we should have done this before, but with the new force the bug was very obvious).
2005-12-30 19:28:26 +00:00
pci_add_map(pcib, bus, dev, b, s, f, q->arg1, rl,
force, 0);
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
}
Fix some resource allocation peculiarities of the intpm device.
1999-10-28 08:06:59 +00:00
Use PCI_INTERRUPT_VALI in stead of hard coded 255
2002-06-01 05:48:58 +00:00
if (cfg->intpin > 0 && PCI_INTERRUPT_VALID(cfg->intline)) {
Use __HAVE_ACPI and __PCI_REROUTE_INTERRUPT as appropriate rather than the complicated #ifdefs.
2006-01-01 21:04:31 +00:00
#ifdef __PCI_REROUTE_INTERRUPT
Re-route interrupts on ia64 so that we can get the I/O SAPIC interrupt numbers (the BIOS leaves legacy PIC interrupt numbers in the intline registers).
2001-10-05 10:33:42 +00:00
/*
- Adjust the comment about re-routing PCI interrupts to be less ia64-specific. - When trying to re-route interrupts, don't change cfg->intline if the re-route fails by returning an invalid vector. This fixes machines without any way of routing interrupts such as older PC's without a $PIR table. We do not currently write the new intline value back to the hardware, but we should. That will likely be added in a later commit.
2003-06-07 15:00:19 +00:00
* Try to re-route interrupts. Sometimes the BIOS or
* firmware may leave bogus values in these registers.
* If the re-route fails, then just stick with what we
* have.
Re-route interrupts on ia64 so that we can get the I/O SAPIC interrupt numbers (the BIOS leaves legacy PIC interrupt numbers in the intline registers).
2001-10-05 10:33:42 +00:00
*/
- Consolidate duplicated code for assigning interrupts to PCI devices via routing, etc. in a static pci_assign_interrupt() function. - Add a sledgehammer that allows the user to override the interrupt assignment of any PCI device via a tunable (e.g. "hw.pci0.7.INTB=5" would force any functions on the pci device in slot 7 of bus 0 that use B# to use IRQ 5). This should be used with great caution! Generally, if the interrupt routing in use provides specific tunables (such as hard-wiring the IRQ for a given $PIR or ACPI PCI link device), then those should be used instead. One instance where this tunable might be useful is if a box has an MPTable with duplicate entries for the same PCI device with different IRQs. MFC after: 1 week
2005-09-29 15:04:41 +00:00
pci_assign_interrupt(bus, dev, 1);
#else
pci_assign_interrupt(bus, dev, 0);
Re-route interrupts on ia64 so that we can get the I/O SAPIC interrupt numbers (the BIOS leaves legacy PIC interrupt numbers in the intline registers).
2001-10-05 10:33:42 +00:00
#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
}
Various changes to make it easier to subclass the PCI bus device. - Make the pci devclass a global variable. - Add child devices in pci_attach() instead of pci_probe(). Change pci_probe() to just check for a valid bus number from the associated bridge and return -1000 if successful. This allows subclasses of the PCI bus driver to override the generic driver. - Move the code to load the vendor data into its own public function. Really though, doing this at attach is just plain wrong. This should really be done in the module load routine instead. As a side effect, the 'busno' variable in pci_attach() is now no longer static (minor bug that was harmless so far.) - Change pci_add_children() to take an extra argument that is the size of the device info structure passed to pci_read_device() and make it public so subclasses of the PCI bus can call it in their attach routines. - Move the bits to attach a probed PCI child to a PCI bus into a global pci_add_child() function. This will allow subclasses that can detect a PCI device not found in the normal PCI probe to add those devices in their own attach routine. (I have seen this in the ACPI tree on my laptop for example.) As a side effect, change the static function pci_add_resources() to get the busno, slot, and func from the passed in dinfo structure instead of requiring them as function arguments. Tested on: i386, alpha, ia64, sparc64
2002-08-26 15:23:52 +00:00
void
pci_add_children(device_t dev, int busno, size_t dinfo_size)
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
{
use a REG macro that was already defined. Reorder how the pci probing in handled. before adding devices, check to see if the slot is a multi-function device to see if we should probe all the functions. Original idea by: imp
2003-06-22 02:26:17 +00:00
#define REG(n, w) PCIB_READ_CONFIG(pcib, busno, s, f, n, w)
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
device_t pcib = device_get_parent(dev);
Various changes to make it easier to subclass the PCI bus device. - Make the pci devclass a global variable. - Add child devices in pci_attach() instead of pci_probe(). Change pci_probe() to just check for a valid bus number from the associated bridge and return -1000 if successful. This allows subclasses of the PCI bus driver to override the generic driver. - Move the code to load the vendor data into its own public function. Really though, doing this at attach is just plain wrong. This should really be done in the module load routine instead. As a side effect, the 'busno' variable in pci_attach() is now no longer static (minor bug that was harmless so far.) - Change pci_add_children() to take an extra argument that is the size of the device info structure passed to pci_read_device() and make it public so subclasses of the PCI bus can call it in their attach routines. - Move the bits to attach a probed PCI child to a PCI bus into a global pci_add_child() function. This will allow subclasses that can detect a PCI device not found in the normal PCI probe to add those devices in their own attach routine. (I have seen this in the ACPI tree on my laptop for example.) As a side effect, change the static function pci_add_resources() to get the busno, slot, and func from the passed in dinfo structure instead of requiring them as function arguments. Tested on: i386, alpha, ia64, sparc64
2002-08-26 15:23:52 +00:00
struct pci_devinfo *dinfo;
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
int maxslots;
Various changes to make it easier to subclass the PCI bus device. - Make the pci devclass a global variable. - Add child devices in pci_attach() instead of pci_probe(). Change pci_probe() to just check for a valid bus number from the associated bridge and return -1000 if successful. This allows subclasses of the PCI bus driver to override the generic driver. - Move the code to load the vendor data into its own public function. Really though, doing this at attach is just plain wrong. This should really be done in the module load routine instead. As a side effect, the 'busno' variable in pci_attach() is now no longer static (minor bug that was harmless so far.) - Change pci_add_children() to take an extra argument that is the size of the device info structure passed to pci_read_device() and make it public so subclasses of the PCI bus can call it in their attach routines. - Move the bits to attach a probed PCI child to a PCI bus into a global pci_add_child() function. This will allow subclasses that can detect a PCI device not found in the normal PCI probe to add those devices in their own attach routine. (I have seen this in the ACPI tree on my laptop for example.) As a side effect, change the static function pci_add_resources() to get the busno, slot, and func from the passed in dinfo structure instead of requiring them as function arguments. Tested on: i386, alpha, ia64, sparc64
2002-08-26 15:23:52 +00:00
int s, f, pcifunchigh;
Prefer the uintXX_t to the u_intXX_t names.
2003-08-22 03:11:53 +00:00
uint8_t hdrtype;
Add support for multiple PCI "hoses" used on various alpha platforms. The specific intent of this commit is to pave the way for importing Compaq XP1000 support. These changes should not affect the i386 port. Reviewed by: Doug Rabson <dfr@nlsystems.com> (actually, he walked me through most of it & deserves more than reviewd-by credit )
1999-05-20 15:33:33 +00:00
Various changes to make it easier to subclass the PCI bus device. - Make the pci devclass a global variable. - Add child devices in pci_attach() instead of pci_probe(). Change pci_probe() to just check for a valid bus number from the associated bridge and return -1000 if successful. This allows subclasses of the PCI bus driver to override the generic driver. - Move the code to load the vendor data into its own public function. Really though, doing this at attach is just plain wrong. This should really be done in the module load routine instead. As a side effect, the 'busno' variable in pci_attach() is now no longer static (minor bug that was harmless so far.) - Change pci_add_children() to take an extra argument that is the size of the device info structure passed to pci_read_device() and make it public so subclasses of the PCI bus can call it in their attach routines. - Move the bits to attach a probed PCI child to a PCI bus into a global pci_add_child() function. This will allow subclasses that can detect a PCI device not found in the normal PCI probe to add those devices in their own attach routine. (I have seen this in the ACPI tree on my laptop for example.) As a side effect, change the static function pci_add_resources() to get the busno, slot, and func from the passed in dinfo structure instead of requiring them as function arguments. Tested on: i386, alpha, ia64, sparc64
2002-08-26 15:23:52 +00:00
KASSERT(dinfo_size >= sizeof(struct pci_devinfo),
("dinfo_size too small"));
maxslots = PCIB_MAXSLOTS(pcib);
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
for (s = 0; s <= maxslots; s++) {
Various changes to make it easier to subclass the PCI bus device. - Make the pci devclass a global variable. - Add child devices in pci_attach() instead of pci_probe(). Change pci_probe() to just check for a valid bus number from the associated bridge and return -1000 if successful. This allows subclasses of the PCI bus driver to override the generic driver. - Move the code to load the vendor data into its own public function. Really though, doing this at attach is just plain wrong. This should really be done in the module load routine instead. As a side effect, the 'busno' variable in pci_attach() is now no longer static (minor bug that was harmless so far.) - Change pci_add_children() to take an extra argument that is the size of the device info structure passed to pci_read_device() and make it public so subclasses of the PCI bus can call it in their attach routines. - Move the bits to attach a probed PCI child to a PCI bus into a global pci_add_child() function. This will allow subclasses that can detect a PCI device not found in the normal PCI probe to add those devices in their own attach routine. (I have seen this in the ACPI tree on my laptop for example.) As a side effect, change the static function pci_add_resources() to get the busno, slot, and func from the passed in dinfo structure instead of requiring them as function arguments. Tested on: i386, alpha, ia64, sparc64
2002-08-26 15:23:52 +00:00
pcifunchigh = 0;
use a REG macro that was already defined. Reorder how the pci probing in handled. before adding devices, check to see if the slot is a multi-function device to see if we should probe all the functions. Original idea by: imp
2003-06-22 02:26:17 +00:00
f = 0;
Add a 1 microsecond delay in pci_add_children(), right before the read of the PCIR_HDRTYPE register. It's the value returned from this read access that determines whether or not we decide a device is present at the current slot index. For some reason that I can't adequately explain, this read fails on my machine when probing the USB controller on my machine (which happens a multifunction device at slot index 3 hung off the PCI-PCI bridge on the AMD8111 (bus index 1)). The read will return 0xFF even though it should return 0x80 to indicate the presence of a multifunction device. As near as I can tell, there's some timing issue involved with reading the 'dead' slot indexes 0 through 2 that causes the read of the actual device at slot 3 to fail. I tried a couple of different tricks to correct the problem (the patch to amd64/pci/pci_cfgreg.c fixes it for the amd64 arch), but adding this delay is the only thing that always allows the USB controllers to be correctly probed 100% of the time. Whatever the problem is, it's likely confined to the AMD8111 chipset. However, a simple 1us delay is fairly harmless and should have no side effects for other hardware. I consider this to be voodoo, but it's fairly benign voodoo and it makes my USB keyboard and mouse work again. Note that this is the second time that I've had to resort to a 1us delay to fix a PCI-related problem with this AMD8111/Opteron system (the first being a fix I made a while back to the NDISulator). It's possible the delay really belongs in the cfgreg code itself, or that pci_cfgreg needs some custom hackery for an errata in the 8111. (I checked but couldn't find any documented errata on AMD's site that could account for these problems.)
2005-10-25 06:53:45 +00:00
DELAY(1);
- Rename PCIx_HEADERTYPE* to PCIx_HDRTYPE* so the constants aren't so long. - Add a new PCIM_HDRTYPE constant for the field in PCIR_HDRTYPE that holds the header type. - Replace several magic numbers with appropriate constants for the header type register and a couple of PCI_FUNCMAX. - Merge to amd64 the fix to the i386 bridge code to skip devices with unknown header types. Requested by: imp (1, 2)
2003-08-28 21:22:25 +00:00
hdrtype = REG(PCIR_HDRTYPE, 1);
if ((hdrtype & PCIM_HDRTYPE) > PCI_MAXHDRTYPE)
use a REG macro that was already defined. Reorder how the pci probing in handled. before adding devices, check to see if the slot is a multi-function device to see if we should probe all the functions. Original idea by: imp
2003-06-22 02:26:17 +00:00
continue;
if (hdrtype & PCIM_MFDEV)
pcifunchigh = PCI_FUNCMAX;
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
for (f = 0; f <= pcifunchigh; f++) {
Various changes to make it easier to subclass the PCI bus device. - Make the pci devclass a global variable. - Add child devices in pci_attach() instead of pci_probe(). Change pci_probe() to just check for a valid bus number from the associated bridge and return -1000 if successful. This allows subclasses of the PCI bus driver to override the generic driver. - Move the code to load the vendor data into its own public function. Really though, doing this at attach is just plain wrong. This should really be done in the module load routine instead. As a side effect, the 'busno' variable in pci_attach() is now no longer static (minor bug that was harmless so far.) - Change pci_add_children() to take an extra argument that is the size of the device info structure passed to pci_read_device() and make it public so subclasses of the PCI bus can call it in their attach routines. - Move the bits to attach a probed PCI child to a PCI bus into a global pci_add_child() function. This will allow subclasses that can detect a PCI device not found in the normal PCI probe to add those devices in their own attach routine. (I have seen this in the ACPI tree on my laptop for example.) As a side effect, change the static function pci_add_resources() to get the busno, slot, and func from the passed in dinfo structure instead of requiring them as function arguments. Tested on: i386, alpha, ia64, sparc64
2002-08-26 15:23:52 +00:00
dinfo = pci_read_device(pcib, busno, s, f, dinfo_size);
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
if (dinfo != NULL) {
Various changes to make it easier to subclass the PCI bus device. - Make the pci devclass a global variable. - Add child devices in pci_attach() instead of pci_probe(). Change pci_probe() to just check for a valid bus number from the associated bridge and return -1000 if successful. This allows subclasses of the PCI bus driver to override the generic driver. - Move the code to load the vendor data into its own public function. Really though, doing this at attach is just plain wrong. This should really be done in the module load routine instead. As a side effect, the 'busno' variable in pci_attach() is now no longer static (minor bug that was harmless so far.) - Change pci_add_children() to take an extra argument that is the size of the device info structure passed to pci_read_device() and make it public so subclasses of the PCI bus can call it in their attach routines. - Move the bits to attach a probed PCI child to a PCI bus into a global pci_add_child() function. This will allow subclasses that can detect a PCI device not found in the normal PCI probe to add those devices in their own attach routine. (I have seen this in the ACPI tree on my laptop for example.) As a side effect, change the static function pci_add_resources() to get the busno, slot, and func from the passed in dinfo structure instead of requiring them as function arguments. Tested on: i386, alpha, ia64, sparc64
2002-08-26 15:23:52 +00:00
pci_add_child(dev, dinfo);
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
}
}
}
use a REG macro that was already defined. Reorder how the pci probing in handled. before adding devices, check to see if the slot is a multi-function device to see if we should probe all the functions. Original idea by: imp
2003-06-22 02:26:17 +00:00
#undef REG
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
}
Various changes to make it easier to subclass the PCI bus device. - Make the pci devclass a global variable. - Add child devices in pci_attach() instead of pci_probe(). Change pci_probe() to just check for a valid bus number from the associated bridge and return -1000 if successful. This allows subclasses of the PCI bus driver to override the generic driver. - Move the code to load the vendor data into its own public function. Really though, doing this at attach is just plain wrong. This should really be done in the module load routine instead. As a side effect, the 'busno' variable in pci_attach() is now no longer static (minor bug that was harmless so far.) - Change pci_add_children() to take an extra argument that is the size of the device info structure passed to pci_read_device() and make it public so subclasses of the PCI bus can call it in their attach routines. - Move the bits to attach a probed PCI child to a PCI bus into a global pci_add_child() function. This will allow subclasses that can detect a PCI device not found in the normal PCI probe to add those devices in their own attach routine. (I have seen this in the ACPI tree on my laptop for example.) As a side effect, change the static function pci_add_resources() to get the busno, slot, and func from the passed in dinfo structure instead of requiring them as function arguments. Tested on: i386, alpha, ia64, sparc64
2002-08-26 15:23:52 +00:00
void
pci_add_child(device_t bus, struct pci_devinfo *dinfo)
{
dinfo->cfg.dev = device_add_child(bus, NULL, -1);
device_set_ivars(dinfo->cfg.dev, dinfo);
Use STAILQ in preference to SLIST for the resources. Insert new resources last in the list rather than first. This makes the resouces print in the 4.x order rather than the 5.x order (eg fdc0 at 0x3f0-0x3f5,0x3f7 is 4.x, but 0x3f7,0x3f0-0x3f5 is 5.x). This also means that the pci code will once again print the resources in BAR ascending order.
2005-03-18 05:19:50 +00:00
resource_list_init(&dinfo->resources);
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
pci_cfg_save(dinfo->cfg.dev, dinfo, 0);
pci_cfg_restore(dinfo->cfg.dev, dinfo);
Various changes to make it easier to subclass the PCI bus device. - Make the pci devclass a global variable. - Add child devices in pci_attach() instead of pci_probe(). Change pci_probe() to just check for a valid bus number from the associated bridge and return -1000 if successful. This allows subclasses of the PCI bus driver to override the generic driver. - Move the code to load the vendor data into its own public function. Really though, doing this at attach is just plain wrong. This should really be done in the module load routine instead. As a side effect, the 'busno' variable in pci_attach() is now no longer static (minor bug that was harmless so far.) - Change pci_add_children() to take an extra argument that is the size of the device info structure passed to pci_read_device() and make it public so subclasses of the PCI bus can call it in their attach routines. - Move the bits to attach a probed PCI child to a PCI bus into a global pci_add_child() function. This will allow subclasses that can detect a PCI device not found in the normal PCI probe to add those devices in their own attach routine. (I have seen this in the ACPI tree on my laptop for example.) As a side effect, change the static function pci_add_resources() to get the busno, slot, and func from the passed in dinfo structure instead of requiring them as function arguments. Tested on: i386, alpha, ia64, sparc64
2002-08-26 15:23:52 +00:00
pci_print_verbose(dinfo);
Expose pci_add_resources to the outside world, add a 'force' flag to force allocation of unallocated BARs (cardbus uses this to preallocate everything). Add a prefetchmask to allow for busses that get prefetch hints to set them. Addjust pci_add_map and pci_ata_maps to take a new force flag which pci_add_resources will pass in. Implement 'force' in pci_add_map. Write new value of allocated resource into the bar, if the allocation succeeded (we should have done this before, but with the new force the bug was very obvious).
2005-12-30 19:28:26 +00:00
pci_add_resources(bus, dinfo->cfg.dev, 0, 0);
Various changes to make it easier to subclass the PCI bus device. - Make the pci devclass a global variable. - Add child devices in pci_attach() instead of pci_probe(). Change pci_probe() to just check for a valid bus number from the associated bridge and return -1000 if successful. This allows subclasses of the PCI bus driver to override the generic driver. - Move the code to load the vendor data into its own public function. Really though, doing this at attach is just plain wrong. This should really be done in the module load routine instead. As a side effect, the 'busno' variable in pci_attach() is now no longer static (minor bug that was harmless so far.) - Change pci_add_children() to take an extra argument that is the size of the device info structure passed to pci_read_device() and make it public so subclasses of the PCI bus can call it in their attach routines. - Move the bits to attach a probed PCI child to a PCI bus into a global pci_add_child() function. This will allow subclasses that can detect a PCI device not found in the normal PCI probe to add those devices in their own attach routine. (I have seen this in the ACPI tree on my laptop for example.) As a side effect, change the static function pci_add_resources() to get the busno, slot, and func from the passed in dinfo structure instead of requiring them as function arguments. Tested on: i386, alpha, ia64, sparc64
2002-08-26 15:23:52 +00:00
}
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
static int
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
pci_probe(device_t dev)
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
{
Convert DEVFS hooks in (most) drivers to make_dev(). Diskslice/label code not yet handled. Vinum, i4b, alpha, pc98 not dealt with (left to respective Maintainers) Add the correct hook for devfs to kern_conf.c The net result of this excercise is that a lot less files depends on DEVFS, and devtoname() gets more sensible output in many cases. A few drivers had minor additional cleanups performed relating to cdevsw registration. A few drivers don't register a cdevsw{} anymore, but only use make_dev().
1999-08-23 20:59:21 +00:00
device_set_desc(dev, "PCI bus");
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
Various changes to make it easier to subclass the PCI bus device. - Make the pci devclass a global variable. - Add child devices in pci_attach() instead of pci_probe(). Change pci_probe() to just check for a valid bus number from the associated bridge and return -1000 if successful. This allows subclasses of the PCI bus driver to override the generic driver. - Move the code to load the vendor data into its own public function. Really though, doing this at attach is just plain wrong. This should really be done in the module load routine instead. As a side effect, the 'busno' variable in pci_attach() is now no longer static (minor bug that was harmless so far.) - Change pci_add_children() to take an extra argument that is the size of the device info structure passed to pci_read_device() and make it public so subclasses of the PCI bus can call it in their attach routines. - Move the bits to attach a probed PCI child to a PCI bus into a global pci_add_child() function. This will allow subclasses that can detect a PCI device not found in the normal PCI probe to add those devices in their own attach routine. (I have seen this in the ACPI tree on my laptop for example.) As a side effect, change the static function pci_add_resources() to get the busno, slot, and func from the passed in dinfo structure instead of requiring them as function arguments. Tested on: i386, alpha, ia64, sparc64
2002-08-26 15:23:52 +00:00
/* Allow other subclasses to override this driver. */
return (-1000);
}
static int
pci_attach(device_t dev)
{
int busno;
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
/*
* Since there can be multiple independantly numbered PCI
Fixup some comments to allow for the fact that PCI domains are not specific to Alpha hoses.
2006-05-11 22:13:21 +00:00
* busses on systems with multiple PCI domains, we can't use
* the unit number to decide which bus we are probing. We ask
* the parent pcib what our bus number is.
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
*/
Validate the PCI bus number that we fetch from our parent, since there's no guarantee that everything attached to *it* is a PCI bus.
2000-10-09 00:43:45 +00:00
busno = pcib_get_bus(dev);
Various changes to make it easier to subclass the PCI bus device. - Make the pci devclass a global variable. - Add child devices in pci_attach() instead of pci_probe(). Change pci_probe() to just check for a valid bus number from the associated bridge and return -1000 if successful. This allows subclasses of the PCI bus driver to override the generic driver. - Move the code to load the vendor data into its own public function. Really though, doing this at attach is just plain wrong. This should really be done in the module load routine instead. As a side effect, the 'busno' variable in pci_attach() is now no longer static (minor bug that was harmless so far.) - Change pci_add_children() to take an extra argument that is the size of the device info structure passed to pci_read_device() and make it public so subclasses of the PCI bus can call it in their attach routines. - Move the bits to attach a probed PCI child to a PCI bus into a global pci_add_child() function. This will allow subclasses that can detect a PCI device not found in the normal PCI probe to add those devices in their own attach routine. (I have seen this in the ACPI tree on my laptop for example.) As a side effect, change the static function pci_add_resources() to get the busno, slot, and func from the passed in dinfo structure instead of requiring them as function arguments. Tested on: i386, alpha, ia64, sparc64
2002-08-26 15:23:52 +00:00
if (bootverbose)
device_printf(dev, "physical bus=%d\n", busno);
pci_add_children(dev, busno, sizeof(struct pci_devinfo));
return (bus_generic_attach(dev));
}
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
int
pci_suspend(device_t dev)
{
Turn ACPI and PCI devices off or to a lower power state in suspend and back on again in resume. Override the default of D3 with the value the BIOS specifies in _SxD, if present. Skip serial devices (PNP05xx) since they seem to hang when set to D3 and may require special driver support. Also, skip non-type 0 PCI devices (i.e., bridges) since our we don't yet save/restore their config space and that seems to be necessary. If this gives you trouble with suspend/resume, you can disable the new ACPI and PCI power behavior separately with these tunables & sysctls: debug.acpi.do_powerstate hw.pci.do_powerstate Approved by: imp (pci) Tested by: acpi@ (numerous)
2004-12-02 08:07:12 +00:00
int dstate, error, i, numdevs;
device_t acpi_dev, child, *devlist;
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
struct pci_devinfo *dinfo;
/*
Turn ACPI and PCI devices off or to a lower power state in suspend and back on again in resume. Override the default of D3 with the value the BIOS specifies in _SxD, if present. Skip serial devices (PNP05xx) since they seem to hang when set to D3 and may require special driver support. Also, skip non-type 0 PCI devices (i.e., bridges) since our we don't yet save/restore their config space and that seems to be necessary. If this gives you trouble with suspend/resume, you can disable the new ACPI and PCI power behavior separately with these tunables & sysctls: debug.acpi.do_powerstate hw.pci.do_powerstate Approved by: imp (pci) Tested by: acpi@ (numerous)
2004-12-02 08:07:12 +00:00
* Save the PCI configuration space for each child and set the
* device in the appropriate power state for this sleep state.
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
*/
Turn ACPI and PCI devices off or to a lower power state in suspend and back on again in resume. Override the default of D3 with the value the BIOS specifies in _SxD, if present. Skip serial devices (PNP05xx) since they seem to hang when set to D3 and may require special driver support. Also, skip non-type 0 PCI devices (i.e., bridges) since our we don't yet save/restore their config space and that seems to be necessary. If this gives you trouble with suspend/resume, you can disable the new ACPI and PCI power behavior separately with these tunables & sysctls: debug.acpi.do_powerstate hw.pci.do_powerstate Approved by: imp (pci) Tested by: acpi@ (numerous)
2004-12-02 08:07:12 +00:00
acpi_dev = NULL;
Split power state control into two variables. hw.pci.do_powerstate has been removed. It has been replaced by hw.pci.do_power_nodriver and hw.pci.do_power_resume. The former defaults to 0 while the latter defaults to 1. When do_powerstate was set to 0, it broke suspend/resume for a lot of people as an unintended consequence. This change will only affect the areas that were intended to affect. This change will have no effect on servers, but will help laptops quite a bit. MFC After: 3 days.
2005-09-21 19:47:00 +00:00
if (pci_do_power_resume)
Turn ACPI and PCI devices off or to a lower power state in suspend and back on again in resume. Override the default of D3 with the value the BIOS specifies in _SxD, if present. Skip serial devices (PNP05xx) since they seem to hang when set to D3 and may require special driver support. Also, skip non-type 0 PCI devices (i.e., bridges) since our we don't yet save/restore their config space and that seems to be necessary. If this gives you trouble with suspend/resume, you can disable the new ACPI and PCI power behavior separately with these tunables & sysctls: debug.acpi.do_powerstate hw.pci.do_powerstate Approved by: imp (pci) Tested by: acpi@ (numerous)
2004-12-02 08:07:12 +00:00
acpi_dev = devclass_get_device(devclass_find("acpi"), 0);
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
device_get_children(dev, &devlist, &numdevs);
for (i = 0; i < numdevs; i++) {
child = devlist[i];
dinfo = (struct pci_devinfo *) device_get_ivars(child);
pci_cfg_save(child, dinfo, 0);
}
Turn ACPI and PCI devices off or to a lower power state in suspend and back on again in resume. Override the default of D3 with the value the BIOS specifies in _SxD, if present. Skip serial devices (PNP05xx) since they seem to hang when set to D3 and may require special driver support. Also, skip non-type 0 PCI devices (i.e., bridges) since our we don't yet save/restore their config space and that seems to be necessary. If this gives you trouble with suspend/resume, you can disable the new ACPI and PCI power behavior separately with these tunables & sysctls: debug.acpi.do_powerstate hw.pci.do_powerstate Approved by: imp (pci) Tested by: acpi@ (numerous)
2004-12-02 08:07:12 +00:00
/* Suspend devices before potentially powering them down. */
error = bus_generic_suspend(dev);
If bus_generic_susped returns an error, devlist is not freed. Free it. Submitted by: Ted Unangst (using the Coverity Prevent analysis tool)
2005-03-15 22:53:31 +00:00
if (error) {
free(devlist, M_TEMP);
Turn ACPI and PCI devices off or to a lower power state in suspend and back on again in resume. Override the default of D3 with the value the BIOS specifies in _SxD, if present. Skip serial devices (PNP05xx) since they seem to hang when set to D3 and may require special driver support. Also, skip non-type 0 PCI devices (i.e., bridges) since our we don't yet save/restore their config space and that seems to be necessary. If this gives you trouble with suspend/resume, you can disable the new ACPI and PCI power behavior separately with these tunables & sysctls: debug.acpi.do_powerstate hw.pci.do_powerstate Approved by: imp (pci) Tested by: acpi@ (numerous)
2004-12-02 08:07:12 +00:00
return (error);
If bus_generic_susped returns an error, devlist is not freed. Free it. Submitted by: Ted Unangst (using the Coverity Prevent analysis tool)
2005-03-15 22:53:31 +00:00
}
Turn ACPI and PCI devices off or to a lower power state in suspend and back on again in resume. Override the default of D3 with the value the BIOS specifies in _SxD, if present. Skip serial devices (PNP05xx) since they seem to hang when set to D3 and may require special driver support. Also, skip non-type 0 PCI devices (i.e., bridges) since our we don't yet save/restore their config space and that seems to be necessary. If this gives you trouble with suspend/resume, you can disable the new ACPI and PCI power behavior separately with these tunables & sysctls: debug.acpi.do_powerstate hw.pci.do_powerstate Approved by: imp (pci) Tested by: acpi@ (numerous)
2004-12-02 08:07:12 +00:00
/*
* Always set the device to D3. If ACPI suggests a different
* power state, use it instead. If ACPI is not present, the
* firmware is responsible for managing device power. Skip
* children who aren't attached since they are powered down
* separately. Only manage type 0 devices for now.
*/
for (i = 0; acpi_dev && i < numdevs; i++) {
child = devlist[i];
dinfo = (struct pci_devinfo *) device_get_ivars(child);
if (device_is_attached(child) && dinfo->cfg.hdrtype == 0) {
dstate = PCI_POWERSTATE_D3;
ACPI_PWR_FOR_SLEEP(acpi_dev, child, &dstate);
pci_set_powerstate(child, dstate);
}
}
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
free(devlist, M_TEMP);
Turn ACPI and PCI devices off or to a lower power state in suspend and back on again in resume. Override the default of D3 with the value the BIOS specifies in _SxD, if present. Skip serial devices (PNP05xx) since they seem to hang when set to D3 and may require special driver support. Also, skip non-type 0 PCI devices (i.e., bridges) since our we don't yet save/restore their config space and that seems to be necessary. If this gives you trouble with suspend/resume, you can disable the new ACPI and PCI power behavior separately with these tunables & sysctls: debug.acpi.do_powerstate hw.pci.do_powerstate Approved by: imp (pci) Tested by: acpi@ (numerous)
2004-12-02 08:07:12 +00:00
return (0);
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
}
int
pci_resume(device_t dev)
{
Turn ACPI and PCI devices off or to a lower power state in suspend and back on again in resume. Override the default of D3 with the value the BIOS specifies in _SxD, if present. Skip serial devices (PNP05xx) since they seem to hang when set to D3 and may require special driver support. Also, skip non-type 0 PCI devices (i.e., bridges) since our we don't yet save/restore their config space and that seems to be necessary. If this gives you trouble with suspend/resume, you can disable the new ACPI and PCI power behavior separately with these tunables & sysctls: debug.acpi.do_powerstate hw.pci.do_powerstate Approved by: imp (pci) Tested by: acpi@ (numerous)
2004-12-02 08:07:12 +00:00
int i, numdevs;
device_t acpi_dev, child, *devlist;
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
struct pci_devinfo *dinfo;
/*
Turn ACPI and PCI devices off or to a lower power state in suspend and back on again in resume. Override the default of D3 with the value the BIOS specifies in _SxD, if present. Skip serial devices (PNP05xx) since they seem to hang when set to D3 and may require special driver support. Also, skip non-type 0 PCI devices (i.e., bridges) since our we don't yet save/restore their config space and that seems to be necessary. If this gives you trouble with suspend/resume, you can disable the new ACPI and PCI power behavior separately with these tunables & sysctls: debug.acpi.do_powerstate hw.pci.do_powerstate Approved by: imp (pci) Tested by: acpi@ (numerous)
2004-12-02 08:07:12 +00:00
* Set each child to D0 and restore its PCI configuration space.
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
*/
Turn ACPI and PCI devices off or to a lower power state in suspend and back on again in resume. Override the default of D3 with the value the BIOS specifies in _SxD, if present. Skip serial devices (PNP05xx) since they seem to hang when set to D3 and may require special driver support. Also, skip non-type 0 PCI devices (i.e., bridges) since our we don't yet save/restore their config space and that seems to be necessary. If this gives you trouble with suspend/resume, you can disable the new ACPI and PCI power behavior separately with these tunables & sysctls: debug.acpi.do_powerstate hw.pci.do_powerstate Approved by: imp (pci) Tested by: acpi@ (numerous)
2004-12-02 08:07:12 +00:00
acpi_dev = NULL;
Split power state control into two variables. hw.pci.do_powerstate has been removed. It has been replaced by hw.pci.do_power_nodriver and hw.pci.do_power_resume. The former defaults to 0 while the latter defaults to 1. When do_powerstate was set to 0, it broke suspend/resume for a lot of people as an unintended consequence. This change will only affect the areas that were intended to affect. This change will have no effect on servers, but will help laptops quite a bit. MFC After: 3 days.
2005-09-21 19:47:00 +00:00
if (pci_do_power_resume)
Turn ACPI and PCI devices off or to a lower power state in suspend and back on again in resume. Override the default of D3 with the value the BIOS specifies in _SxD, if present. Skip serial devices (PNP05xx) since they seem to hang when set to D3 and may require special driver support. Also, skip non-type 0 PCI devices (i.e., bridges) since our we don't yet save/restore their config space and that seems to be necessary. If this gives you trouble with suspend/resume, you can disable the new ACPI and PCI power behavior separately with these tunables & sysctls: debug.acpi.do_powerstate hw.pci.do_powerstate Approved by: imp (pci) Tested by: acpi@ (numerous)
2004-12-02 08:07:12 +00:00
acpi_dev = devclass_get_device(devclass_find("acpi"), 0);
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
device_get_children(dev, &devlist, &numdevs);
for (i = 0; i < numdevs; i++) {
Turn ACPI and PCI devices off or to a lower power state in suspend and back on again in resume. Override the default of D3 with the value the BIOS specifies in _SxD, if present. Skip serial devices (PNP05xx) since they seem to hang when set to D3 and may require special driver support. Also, skip non-type 0 PCI devices (i.e., bridges) since our we don't yet save/restore their config space and that seems to be necessary. If this gives you trouble with suspend/resume, you can disable the new ACPI and PCI power behavior separately with these tunables & sysctls: debug.acpi.do_powerstate hw.pci.do_powerstate Approved by: imp (pci) Tested by: acpi@ (numerous)
2004-12-02 08:07:12 +00:00
/*
* Notify ACPI we're going to D0 but ignore the result. If
* ACPI is not present, the firmware is responsible for
* managing device power. Only manage type 0 devices for now.
*/
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
child = devlist[i];
dinfo = (struct pci_devinfo *) device_get_ivars(child);
Turn ACPI and PCI devices off or to a lower power state in suspend and back on again in resume. Override the default of D3 with the value the BIOS specifies in _SxD, if present. Skip serial devices (PNP05xx) since they seem to hang when set to D3 and may require special driver support. Also, skip non-type 0 PCI devices (i.e., bridges) since our we don't yet save/restore their config space and that seems to be necessary. If this gives you trouble with suspend/resume, you can disable the new ACPI and PCI power behavior separately with these tunables & sysctls: debug.acpi.do_powerstate hw.pci.do_powerstate Approved by: imp (pci) Tested by: acpi@ (numerous)
2004-12-02 08:07:12 +00:00
if (acpi_dev && device_is_attached(child) &&
dinfo->cfg.hdrtype == 0) {
ACPI_PWR_FOR_SLEEP(acpi_dev, child, NULL);
pci_set_powerstate(child, PCI_POWERSTATE_D0);
}
/* Now the device is powered up, restore its config space. */
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
pci_cfg_restore(child, dinfo);
}
free(devlist, M_TEMP);
return (bus_generic_resume(dev));
}
- Make pci_load_vendor_data() static and do it during MOD_LOAD instead of when the first PCI bus attaches. - Create /dev/pci during MOD_LOAD as well. - Destroy /dev/pci during MOD_UNLOAD (not that you can kldunload pci, but might as well get the code right)
2002-09-04 03:13:16 +00:00
static void
Various changes to make it easier to subclass the PCI bus device. - Make the pci devclass a global variable. - Add child devices in pci_attach() instead of pci_probe(). Change pci_probe() to just check for a valid bus number from the associated bridge and return -1000 if successful. This allows subclasses of the PCI bus driver to override the generic driver. - Move the code to load the vendor data into its own public function. Really though, doing this at attach is just plain wrong. This should really be done in the module load routine instead. As a side effect, the 'busno' variable in pci_attach() is now no longer static (minor bug that was harmless so far.) - Change pci_add_children() to take an extra argument that is the size of the device info structure passed to pci_read_device() and make it public so subclasses of the PCI bus can call it in their attach routines. - Move the bits to attach a probed PCI child to a PCI bus into a global pci_add_child() function. This will allow subclasses that can detect a PCI device not found in the normal PCI probe to add those devices in their own attach routine. (I have seen this in the ACPI tree on my laptop for example.) As a side effect, change the static function pci_add_resources() to get the busno, slot, and func from the passed in dinfo structure instead of requiring them as function arguments. Tested on: i386, alpha, ia64, sparc64
2002-08-26 15:23:52 +00:00
pci_load_vendor_data(void)
{
caddr_t vendordata, info;
- Make pci_load_vendor_data() static and do it during MOD_LOAD instead of when the first PCI bus attaches. - Create /dev/pci during MOD_LOAD as well. - Destroy /dev/pci during MOD_UNLOAD (not that you can kldunload pci, but might as well get the code right)
2002-09-04 03:13:16 +00:00
if ((vendordata = preload_search_by_type("pci_vendor_data")) != NULL) {
info = preload_search_info(vendordata, MODINFO_ADDR);
pci_vendordata = *(char **)info;
info = preload_search_info(vendordata, MODINFO_SIZE);
pci_vendordata_size = *(size_t *)info;
/* terminate the database */
pci_vendordata[pci_vendordata_size] = '\n';
Convert DEVFS hooks in (most) drivers to make_dev(). Diskslice/label code not yet handled. Vinum, i4b, alpha, pc98 not dealt with (left to respective Maintainers) Add the correct hook for devfs to kern_conf.c The net result of this excercise is that a lot less files depends on DEVFS, and devtoname() gets more sensible output in many cases. A few drivers had minor additional cleanups performed relating to cdevsw registration. A few drivers don't register a cdevsw{} anymore, but only use make_dev().
1999-08-23 20:59:21 +00:00
}
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
}
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
void
pci_driver_added(device_t dev, driver_t *driver)
{
int numdevs;
device_t *devlist;
device_t child;
struct pci_devinfo *dinfo;
int i;
make the pci power state and resource code a lot less chatty. The chattiness was left in for debugging, but now that nearly all of the problems relating to the changes have been fixed, it is only annoying. It is still available via bootverbose. Prodded by: jhb
2004-05-21 06:03:26 +00:00
if (bootverbose)
device_printf(dev, "driver added\n");
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
DEVICE_IDENTIFY(driver, dev);
device_get_children(dev, &devlist, &numdevs);
for (i = 0; i < numdevs; i++) {
child = devlist[i];
if (device_get_state(child) != DS_NOTPRESENT)
continue;
dinfo = device_get_ivars(child);
pci_print_verbose(dinfo);
make the pci power state and resource code a lot less chatty. The chattiness was left in for debugging, but now that nearly all of the problems relating to the changes have been fixed, it is only annoying. It is still available via bootverbose. Prodded by: jhb
2004-05-21 06:03:26 +00:00
if (bootverbose)
printf("pci%d:%d:%d: reprobing on driver added\n",
dinfo->cfg.bus, dinfo->cfg.slot, dinfo->cfg.func);
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
pci_cfg_restore(child, dinfo);
if (device_probe_and_attach(child) != 0)
pci_cfg_save(child, dinfo, 1);
}
free(devlist, M_TEMP);
}
Use the pci.c code wherever possible, rather than copying all the pci code into cardbus and s/pci/cardbus. This exposes a few pci_* functions that are now static. This work is similar to work Justin posted to the mobile list about a year or two ago, which I have neglected since then. This is a subset of his current work with the multiple inheritance newbus architecutre. When completed, that will eliminate the need for pci/pci_private.h. Similar work is needed for the cardbus_cis and pccard_cis code as well.
2002-02-27 05:09:14 +00:00
int
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
pci_print_child(device_t dev, device_t child)
{
Print 'irq nn' on the device attach line like the old pci code did. However, we are not printing 'int a/b/c/d' yet, is it worth it on non-SMP systems? (It's useful when tracing PCI->IO-APIC routing on SMP systems)
1999-05-08 20:28:01 +00:00
struct pci_devinfo *dinfo;
Show the port/mem/irq of pci devices too.
2000-01-08 10:12:21 +00:00
struct resource_list *rl;
Alter the behavior of sys/kern/subr_bus.c:device_print_child() - device_print_child() either lets the BUS_PRINT_CHILD method produce the entire device announcement message or it prints "foo0: not found\n" Alter sys/kern/subr_bus.c:bus_generic_print_child() to take on the previous behavior of device_print_child() (printing the "foo0: <FooDevice 1.1>" bit of the announce message.) Provide bus_print_child_header() and bus_print_child_footer() to actually print the output for bus_generic_print_child(). These functions should be used whenever possible (unless you can just use bus_generic_print_child()) The BUS_PRINT_CHILD method now returns int instead of void. Modify everything else that defines or uses a BUS_PRINT_CHILD method to comply with the above changes. - Devices are 'on' a bus, not 'at' it. - If a custom BUS_PRINT_CHILD method does the same thing as bus_generic_print_child(), use bus_generic_print_child() - Use device_get_nameunit() instead of both device_get_name() and device_get_unit() - All BUS_PRINT_CHILD methods return the number of characters output. Reviewed by: dfr, peter
1999-07-29 01:03:04 +00:00
int retval = 0;
Print 'irq nn' on the device attach line like the old pci code did. However, we are not printing 'int a/b/c/d' yet, is it worth it on non-SMP systems? (It's useful when tracing PCI->IO-APIC routing on SMP systems)
1999-05-08 20:28:01 +00:00
dinfo = device_get_ivars(child);
Show the port/mem/irq of pci devices too.
2000-01-08 10:12:21 +00:00
rl = &dinfo->resources;
Alter the behavior of sys/kern/subr_bus.c:device_print_child() - device_print_child() either lets the BUS_PRINT_CHILD method produce the entire device announcement message or it prints "foo0: not found\n" Alter sys/kern/subr_bus.c:bus_generic_print_child() to take on the previous behavior of device_print_child() (printing the "foo0: <FooDevice 1.1>" bit of the announce message.) Provide bus_print_child_header() and bus_print_child_footer() to actually print the output for bus_generic_print_child(). These functions should be used whenever possible (unless you can just use bus_generic_print_child()) The BUS_PRINT_CHILD method now returns int instead of void. Modify everything else that defines or uses a BUS_PRINT_CHILD method to comply with the above changes. - Devices are 'on' a bus, not 'at' it. - If a custom BUS_PRINT_CHILD method does the same thing as bus_generic_print_child(), use bus_generic_print_child() - Use device_get_nameunit() instead of both device_get_name() and device_get_unit() - All BUS_PRINT_CHILD methods return the number of characters output. Reviewed by: dfr, peter
1999-07-29 01:03:04 +00:00
retval += bus_print_child_header(dev, child);
Use the new resource_list_print_type() function to print resource list contents, and the new __BUS_ACCESSOR macro to construct the accessor functions.
2001-12-21 21:49:57 +00:00
retval += resource_list_print_type(rl, "port", SYS_RES_IOPORT, "%#lx");
retval += resource_list_print_type(rl, "mem", SYS_RES_MEMORY, "%#lx");
retval += resource_list_print_type(rl, "irq", SYS_RES_IRQ, "%ld");
Show the port/mem/irq of pci devices too.
2000-01-08 10:12:21 +00:00
if (device_get_flags(dev))
retval += printf(" flags %#x", device_get_flags(dev));
Alter the behavior of sys/kern/subr_bus.c:device_print_child() - device_print_child() either lets the BUS_PRINT_CHILD method produce the entire device announcement message or it prints "foo0: not found\n" Alter sys/kern/subr_bus.c:bus_generic_print_child() to take on the previous behavior of device_print_child() (printing the "foo0: <FooDevice 1.1>" bit of the announce message.) Provide bus_print_child_header() and bus_print_child_footer() to actually print the output for bus_generic_print_child(). These functions should be used whenever possible (unless you can just use bus_generic_print_child()) The BUS_PRINT_CHILD method now returns int instead of void. Modify everything else that defines or uses a BUS_PRINT_CHILD method to comply with the above changes. - Devices are 'on' a bus, not 'at' it. - If a custom BUS_PRINT_CHILD method does the same thing as bus_generic_print_child(), use bus_generic_print_child() - Use device_get_nameunit() instead of both device_get_name() and device_get_unit() - All BUS_PRINT_CHILD methods return the number of characters output. Reviewed by: dfr, peter
1999-07-29 01:03:04 +00:00
retval += printf(" at device %d.%d", pci_get_slot(child),
Make this file mostly conform to style(9). Approved by: msmith in principle before walkabout
2002-06-01 03:41:02 +00:00
pci_get_function(child));
Alter the behavior of sys/kern/subr_bus.c:device_print_child() - device_print_child() either lets the BUS_PRINT_CHILD method produce the entire device announcement message or it prints "foo0: not found\n" Alter sys/kern/subr_bus.c:bus_generic_print_child() to take on the previous behavior of device_print_child() (printing the "foo0: <FooDevice 1.1>" bit of the announce message.) Provide bus_print_child_header() and bus_print_child_footer() to actually print the output for bus_generic_print_child(). These functions should be used whenever possible (unless you can just use bus_generic_print_child()) The BUS_PRINT_CHILD method now returns int instead of void. Modify everything else that defines or uses a BUS_PRINT_CHILD method to comply with the above changes. - Devices are 'on' a bus, not 'at' it. - If a custom BUS_PRINT_CHILD method does the same thing as bus_generic_print_child(), use bus_generic_print_child() - Use device_get_nameunit() instead of both device_get_name() and device_get_unit() - All BUS_PRINT_CHILD methods return the number of characters output. Reviewed by: dfr, peter
1999-07-29 01:03:04 +00:00
retval += bus_print_child_footer(dev, child);
return (retval);
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
}
Next phase in the PCI subsystem cleanup. - Move PCI core code to dev/pci. - Split bridge code out into separate modules. - Remove the descriptive strings from the bridge drivers. If you want to know what a device is, use pciconf. Add support for broadly identifying devices based on class/subclass, and for parsing a preloaded device identification database so that if you want to waste the memory, you can identify *anything* we know about. - Remove machine-dependant code from the core PCI code. APIC interrupt mapping is performed by shadowing the intline register in machine- dependant code. - Bring interrupt routing support to the Alpha (although many platforms don't yet support routing or mapping interrupts entirely correctly). This resulted in spamming <sys/bus.h> into more places than it really should have gone. - Put sys/dev on the kernel/modules include path. This avoids having to change *all* the pci*.h includes.
2000-12-08 22:11:23 +00:00
static struct
{
int class;
int subclass;
char *desc;
} pci_nomatch_tab[] = {
{PCIC_OLD, -1, "old"},
{PCIC_OLD, PCIS_OLD_NONVGA, "non-VGA display device"},
{PCIC_OLD, PCIS_OLD_VGA, "VGA-compatible display device"},
{PCIC_STORAGE, -1, "mass storage"},
{PCIC_STORAGE, PCIS_STORAGE_SCSI, "SCSI"},
{PCIC_STORAGE, PCIS_STORAGE_IDE, "ATA"},
{PCIC_STORAGE, PCIS_STORAGE_FLOPPY, "floppy disk"},
{PCIC_STORAGE, PCIS_STORAGE_IPI, "IPI"},
{PCIC_STORAGE, PCIS_STORAGE_RAID, "RAID"},
{PCIC_NETWORK, -1, "network"},
{PCIC_NETWORK, PCIS_NETWORK_ETHERNET, "ethernet"},
{PCIC_NETWORK, PCIS_NETWORK_TOKENRING, "token ring"},
{PCIC_NETWORK, PCIS_NETWORK_FDDI, "fddi"},
{PCIC_NETWORK, PCIS_NETWORK_ATM, "ATM"},
add some additional pci classes and sub-classes.. Reviewed by: imp (almost 6 months ago)
2005-03-26 20:31:09 +00:00
{PCIC_NETWORK, PCIS_NETWORK_ISDN, "ISDN"},
Next phase in the PCI subsystem cleanup. - Move PCI core code to dev/pci. - Split bridge code out into separate modules. - Remove the descriptive strings from the bridge drivers. If you want to know what a device is, use pciconf. Add support for broadly identifying devices based on class/subclass, and for parsing a preloaded device identification database so that if you want to waste the memory, you can identify *anything* we know about. - Remove machine-dependant code from the core PCI code. APIC interrupt mapping is performed by shadowing the intline register in machine- dependant code. - Bring interrupt routing support to the Alpha (although many platforms don't yet support routing or mapping interrupts entirely correctly). This resulted in spamming <sys/bus.h> into more places than it really should have gone. - Put sys/dev on the kernel/modules include path. This avoids having to change *all* the pci*.h includes.
2000-12-08 22:11:23 +00:00
{PCIC_DISPLAY, -1, "display"},
{PCIC_DISPLAY, PCIS_DISPLAY_VGA, "VGA"},
{PCIC_DISPLAY, PCIS_DISPLAY_XGA, "XGA"},
add some additional pci classes and sub-classes.. Reviewed by: imp (almost 6 months ago)
2005-03-26 20:31:09 +00:00
{PCIC_DISPLAY, PCIS_DISPLAY_3D, "3D"},
Next phase in the PCI subsystem cleanup. - Move PCI core code to dev/pci. - Split bridge code out into separate modules. - Remove the descriptive strings from the bridge drivers. If you want to know what a device is, use pciconf. Add support for broadly identifying devices based on class/subclass, and for parsing a preloaded device identification database so that if you want to waste the memory, you can identify *anything* we know about. - Remove machine-dependant code from the core PCI code. APIC interrupt mapping is performed by shadowing the intline register in machine- dependant code. - Bring interrupt routing support to the Alpha (although many platforms don't yet support routing or mapping interrupts entirely correctly). This resulted in spamming <sys/bus.h> into more places than it really should have gone. - Put sys/dev on the kernel/modules include path. This avoids having to change *all* the pci*.h includes.
2000-12-08 22:11:23 +00:00
{PCIC_MULTIMEDIA, -1, "multimedia"},
{PCIC_MULTIMEDIA, PCIS_MULTIMEDIA_VIDEO, "video"},
{PCIC_MULTIMEDIA, PCIS_MULTIMEDIA_AUDIO, "audio"},
add some additional pci classes and sub-classes.. Reviewed by: imp (almost 6 months ago)
2005-03-26 20:31:09 +00:00
{PCIC_MULTIMEDIA, PCIS_MULTIMEDIA_TELE, "telephony"},
Next phase in the PCI subsystem cleanup. - Move PCI core code to dev/pci. - Split bridge code out into separate modules. - Remove the descriptive strings from the bridge drivers. If you want to know what a device is, use pciconf. Add support for broadly identifying devices based on class/subclass, and for parsing a preloaded device identification database so that if you want to waste the memory, you can identify *anything* we know about. - Remove machine-dependant code from the core PCI code. APIC interrupt mapping is performed by shadowing the intline register in machine- dependant code. - Bring interrupt routing support to the Alpha (although many platforms don't yet support routing or mapping interrupts entirely correctly). This resulted in spamming <sys/bus.h> into more places than it really should have gone. - Put sys/dev on the kernel/modules include path. This avoids having to change *all* the pci*.h includes.
2000-12-08 22:11:23 +00:00
{PCIC_MEMORY, -1, "memory"},
{PCIC_MEMORY, PCIS_MEMORY_RAM, "RAM"},
{PCIC_MEMORY, PCIS_MEMORY_FLASH, "flash"},
{PCIC_BRIDGE, -1, "bridge"},
{PCIC_BRIDGE, PCIS_BRIDGE_HOST, "HOST-PCI"},
{PCIC_BRIDGE, PCIS_BRIDGE_ISA, "PCI-ISA"},
{PCIC_BRIDGE, PCIS_BRIDGE_EISA, "PCI-EISA"},
{PCIC_BRIDGE, PCIS_BRIDGE_MCA, "PCI-MCA"},
{PCIC_BRIDGE, PCIS_BRIDGE_PCI, "PCI-PCI"},
{PCIC_BRIDGE, PCIS_BRIDGE_PCMCIA, "PCI-PCMCIA"},
{PCIC_BRIDGE, PCIS_BRIDGE_NUBUS, "PCI-NuBus"},
{PCIC_BRIDGE, PCIS_BRIDGE_CARDBUS, "PCI-CardBus"},
add some additional pci classes and sub-classes.. Reviewed by: imp (almost 6 months ago)
2005-03-26 20:31:09 +00:00
{PCIC_BRIDGE, PCIS_BRIDGE_RACEWAY, "PCI-RACEway"},
Next phase in the PCI subsystem cleanup. - Move PCI core code to dev/pci. - Split bridge code out into separate modules. - Remove the descriptive strings from the bridge drivers. If you want to know what a device is, use pciconf. Add support for broadly identifying devices based on class/subclass, and for parsing a preloaded device identification database so that if you want to waste the memory, you can identify *anything* we know about. - Remove machine-dependant code from the core PCI code. APIC interrupt mapping is performed by shadowing the intline register in machine- dependant code. - Bring interrupt routing support to the Alpha (although many platforms don't yet support routing or mapping interrupts entirely correctly). This resulted in spamming <sys/bus.h> into more places than it really should have gone. - Put sys/dev on the kernel/modules include path. This avoids having to change *all* the pci*.h includes.
2000-12-08 22:11:23 +00:00
{PCIC_SIMPLECOMM, -1, "simple comms"},
{PCIC_SIMPLECOMM, PCIS_SIMPLECOMM_UART, "UART"}, /* could detect 16550 */
{PCIC_SIMPLECOMM, PCIS_SIMPLECOMM_PAR, "parallel port"},
add some additional pci classes and sub-classes.. Reviewed by: imp (almost 6 months ago)
2005-03-26 20:31:09 +00:00
{PCIC_SIMPLECOMM, PCIS_SIMPLECOMM_MULSER, "multiport serial"},
{PCIC_SIMPLECOMM, PCIS_SIMPLECOMM_MODEM, "generic modem"},
Next phase in the PCI subsystem cleanup. - Move PCI core code to dev/pci. - Split bridge code out into separate modules. - Remove the descriptive strings from the bridge drivers. If you want to know what a device is, use pciconf. Add support for broadly identifying devices based on class/subclass, and for parsing a preloaded device identification database so that if you want to waste the memory, you can identify *anything* we know about. - Remove machine-dependant code from the core PCI code. APIC interrupt mapping is performed by shadowing the intline register in machine- dependant code. - Bring interrupt routing support to the Alpha (although many platforms don't yet support routing or mapping interrupts entirely correctly). This resulted in spamming <sys/bus.h> into more places than it really should have gone. - Put sys/dev on the kernel/modules include path. This avoids having to change *all* the pci*.h includes.
2000-12-08 22:11:23 +00:00
{PCIC_BASEPERIPH, -1, "base peripheral"},
{PCIC_BASEPERIPH, PCIS_BASEPERIPH_PIC, "interrupt controller"},
{PCIC_BASEPERIPH, PCIS_BASEPERIPH_DMA, "DMA controller"},
{PCIC_BASEPERIPH, PCIS_BASEPERIPH_TIMER, "timer"},
{PCIC_BASEPERIPH, PCIS_BASEPERIPH_RTC, "realtime clock"},
add some additional pci classes and sub-classes.. Reviewed by: imp (almost 6 months ago)
2005-03-26 20:31:09 +00:00
{PCIC_BASEPERIPH, PCIS_BASEPERIPH_PCIHOT, "PCI hot-plug controller"},
Next phase in the PCI subsystem cleanup. - Move PCI core code to dev/pci. - Split bridge code out into separate modules. - Remove the descriptive strings from the bridge drivers. If you want to know what a device is, use pciconf. Add support for broadly identifying devices based on class/subclass, and for parsing a preloaded device identification database so that if you want to waste the memory, you can identify *anything* we know about. - Remove machine-dependant code from the core PCI code. APIC interrupt mapping is performed by shadowing the intline register in machine- dependant code. - Bring interrupt routing support to the Alpha (although many platforms don't yet support routing or mapping interrupts entirely correctly). This resulted in spamming <sys/bus.h> into more places than it really should have gone. - Put sys/dev on the kernel/modules include path. This avoids having to change *all* the pci*.h includes.
2000-12-08 22:11:23 +00:00
{PCIC_INPUTDEV, -1, "input device"},
{PCIC_INPUTDEV, PCIS_INPUTDEV_KEYBOARD, "keyboard"},
{PCIC_INPUTDEV, PCIS_INPUTDEV_DIGITIZER,"digitizer"},
{PCIC_INPUTDEV, PCIS_INPUTDEV_MOUSE, "mouse"},
add some additional pci classes and sub-classes.. Reviewed by: imp (almost 6 months ago)
2005-03-26 20:31:09 +00:00
{PCIC_INPUTDEV, PCIS_INPUTDEV_SCANNER, "scanner"},
{PCIC_INPUTDEV, PCIS_INPUTDEV_GAMEPORT, "gameport"},
Next phase in the PCI subsystem cleanup. - Move PCI core code to dev/pci. - Split bridge code out into separate modules. - Remove the descriptive strings from the bridge drivers. If you want to know what a device is, use pciconf. Add support for broadly identifying devices based on class/subclass, and for parsing a preloaded device identification database so that if you want to waste the memory, you can identify *anything* we know about. - Remove machine-dependant code from the core PCI code. APIC interrupt mapping is performed by shadowing the intline register in machine- dependant code. - Bring interrupt routing support to the Alpha (although many platforms don't yet support routing or mapping interrupts entirely correctly). This resulted in spamming <sys/bus.h> into more places than it really should have gone. - Put sys/dev on the kernel/modules include path. This avoids having to change *all* the pci*.h includes.
2000-12-08 22:11:23 +00:00
{PCIC_DOCKING, -1, "docking station"},
{PCIC_PROCESSOR, -1, "processor"},
{PCIC_SERIALBUS, -1, "serial bus"},
{PCIC_SERIALBUS, PCIS_SERIALBUS_FW, "FireWire"},
{PCIC_SERIALBUS, PCIS_SERIALBUS_ACCESS, "AccessBus"},
{PCIC_SERIALBUS, PCIS_SERIALBUS_SSA, "SSA"},
{PCIC_SERIALBUS, PCIS_SERIALBUS_USB, "USB"},
{PCIC_SERIALBUS, PCIS_SERIALBUS_FC, "Fibre Channel"},
{PCIC_SERIALBUS, PCIS_SERIALBUS_SMBUS, "SMBus"},
add some additional pci classes and sub-classes.. Reviewed by: imp (almost 6 months ago)
2005-03-26 20:31:09 +00:00
{PCIC_WIRELESS, -1, "wireless controller"},
{PCIC_WIRELESS, PCIS_WIRELESS_IRDA, "iRDA"},
{PCIC_WIRELESS, PCIS_WIRELESS_IR, "IR"},
{PCIC_WIRELESS, PCIS_WIRELESS_RF, "RF"},
{PCIC_INTELLIIO, -1, "intelligent I/O controller"},
{PCIC_INTELLIIO, PCIS_INTELLIIO_I2O, "I2O"},
{PCIC_SATCOM, -1, "satellite communication"},
{PCIC_SATCOM, PCIS_SATCOM_TV, "sat TV"},
{PCIC_SATCOM, PCIS_SATCOM_AUDIO, "sat audio"},
{PCIC_SATCOM, PCIS_SATCOM_VOICE, "sat voice"},
{PCIC_SATCOM, PCIS_SATCOM_DATA, "sat data"},
{PCIC_CRYPTO, -1, "encrypt/decrypt"},
{PCIC_CRYPTO, PCIS_CRYPTO_NETCOMP, "network/computer crypto"},
spell PCIS_CRYPTO_ENTERTAIN properly... MFC after: 3 days
2006-09-20 06:47:14 +00:00
{PCIC_CRYPTO, PCIS_CRYPTO_ENTERTAIN, "entertainment crypto"},
add some additional pci classes and sub-classes.. Reviewed by: imp (almost 6 months ago)
2005-03-26 20:31:09 +00:00
{PCIC_DASP, -1, "dasp"},
{PCIC_DASP, PCIS_DASP_DPIO, "DPIO module"},
Next phase in the PCI subsystem cleanup. - Move PCI core code to dev/pci. - Split bridge code out into separate modules. - Remove the descriptive strings from the bridge drivers. If you want to know what a device is, use pciconf. Add support for broadly identifying devices based on class/subclass, and for parsing a preloaded device identification database so that if you want to waste the memory, you can identify *anything* we know about. - Remove machine-dependant code from the core PCI code. APIC interrupt mapping is performed by shadowing the intline register in machine- dependant code. - Bring interrupt routing support to the Alpha (although many platforms don't yet support routing or mapping interrupts entirely correctly). This resulted in spamming <sys/bus.h> into more places than it really should have gone. - Put sys/dev on the kernel/modules include path. This avoids having to change *all* the pci*.h includes.
2000-12-08 22:11:23 +00:00
{0, 0, NULL}
};
Use the pci.c code wherever possible, rather than copying all the pci code into cardbus and s/pci/cardbus. This exposes a few pci_* functions that are now static. This work is similar to work Justin posted to the mobile list about a year or two ago, which I have neglected since then. This is a subset of his current work with the multiple inheritance newbus architecutre. When completed, that will eliminate the need for pci/pci_private.h. Similar work is needed for the cardbus_cis and pccard_cis code as well.
2002-02-27 05:09:14 +00:00
void
Implement the BUS_PROBE_NOMATCH method for the PCI bus. This function is called for each device for which no driver was found. Output is similar to the eisa_probe_nomatch() function but with the added benefit of displaying the assigned IRQ (since PCI gives us this information up front.) Output is like so: pci0: unknown card CPQ0508 (vendor=0x0e11, dev=0x0508) at 11.0 irq 9 pci0: unknown card DFZ0508 (vendor=0x10da, dev=0x0508) at 11.0 irq 9 pci0: unknown card DBL0508 (vendor=0x104c, dev=0x0508) at 11.0 irq 9 pci0: unknown card DDM0011 (vendor=0x108d, dev=0x0011) at 11.0 irq 9 I'm not happy with the 3 lines of macro cruft that got added but I consider it a temporary annoyance as those bits will be moved to some place where PCI, EISA and ISAPNP code will be able to use them. (Not surprisingly, this message is longer than the code in question.) Reviewed by: peter, dfr
1999-07-27 04:28:14 +00:00
pci_probe_nomatch(device_t dev, device_t child)
{
Next phase in the PCI subsystem cleanup. - Move PCI core code to dev/pci. - Split bridge code out into separate modules. - Remove the descriptive strings from the bridge drivers. If you want to know what a device is, use pciconf. Add support for broadly identifying devices based on class/subclass, and for parsing a preloaded device identification database so that if you want to waste the memory, you can identify *anything* we know about. - Remove machine-dependant code from the core PCI code. APIC interrupt mapping is performed by shadowing the intline register in machine- dependant code. - Bring interrupt routing support to the Alpha (although many platforms don't yet support routing or mapping interrupts entirely correctly). This resulted in spamming <sys/bus.h> into more places than it really should have gone. - Put sys/dev on the kernel/modules include path. This avoids having to change *all* the pci*.h includes.
2000-12-08 22:11:23 +00:00
int i;
char *cp, *scp, *device;
/*
* Look for a listing for this device in a loaded device database.
*/
if ((device = pci_describe_device(child)) != NULL) {
- Fix the device database parsing code so that it actually works. - Improve the formatting for devices identified by the database. - Fix the pcib_route_interrupt method definition, as an old version snuck in here somehow 8( - Remove a couple of the vendor/device IDs for PCI:ISA bridges which correctly identify themselves. Submitted by: peter
2000-12-09 09:15:38 +00:00
device_printf(dev, "<%s>", device);
Next phase in the PCI subsystem cleanup. - Move PCI core code to dev/pci. - Split bridge code out into separate modules. - Remove the descriptive strings from the bridge drivers. If you want to know what a device is, use pciconf. Add support for broadly identifying devices based on class/subclass, and for parsing a preloaded device identification database so that if you want to waste the memory, you can identify *anything* we know about. - Remove machine-dependant code from the core PCI code. APIC interrupt mapping is performed by shadowing the intline register in machine- dependant code. - Bring interrupt routing support to the Alpha (although many platforms don't yet support routing or mapping interrupts entirely correctly). This resulted in spamming <sys/bus.h> into more places than it really should have gone. - Put sys/dev on the kernel/modules include path. This avoids having to change *all* the pci*.h includes.
2000-12-08 22:11:23 +00:00
free(device, M_DEVBUF);
} else {
Make this file mostly conform to style(9). Approved by: msmith in principle before walkabout
2002-06-01 03:41:02 +00:00
/*
* Scan the class/subclass descriptions for a general
* description.
*/
cp = "unknown";
scp = NULL;
for (i = 0; pci_nomatch_tab[i].desc != NULL; i++) {
if (pci_nomatch_tab[i].class == pci_get_class(child)) {
if (pci_nomatch_tab[i].subclass == -1) {
cp = pci_nomatch_tab[i].desc;
} else if (pci_nomatch_tab[i].subclass ==
pci_get_subclass(child)) {
scp = pci_nomatch_tab[i].desc;
}
}
Next phase in the PCI subsystem cleanup. - Move PCI core code to dev/pci. - Split bridge code out into separate modules. - Remove the descriptive strings from the bridge drivers. If you want to know what a device is, use pciconf. Add support for broadly identifying devices based on class/subclass, and for parsing a preloaded device identification database so that if you want to waste the memory, you can identify *anything* we know about. - Remove machine-dependant code from the core PCI code. APIC interrupt mapping is performed by shadowing the intline register in machine- dependant code. - Bring interrupt routing support to the Alpha (although many platforms don't yet support routing or mapping interrupts entirely correctly). This resulted in spamming <sys/bus.h> into more places than it really should have gone. - Put sys/dev on the kernel/modules include path. This avoids having to change *all* the pci*.h includes.
2000-12-08 22:11:23 +00:00
}
Make this file mostly conform to style(9). Approved by: msmith in principle before walkabout
2002-06-01 03:41:02 +00:00
device_printf(dev, "<%s%s%s>",
Don't use GCC shorthand for ?: unless it really matters.
2002-09-28 17:47:51 +00:00
cp ? cp : "",
((cp != NULL) && (scp != NULL)) ? ", " : "",
scp ? scp : "");
Next phase in the PCI subsystem cleanup. - Move PCI core code to dev/pci. - Split bridge code out into separate modules. - Remove the descriptive strings from the bridge drivers. If you want to know what a device is, use pciconf. Add support for broadly identifying devices based on class/subclass, and for parsing a preloaded device identification database so that if you want to waste the memory, you can identify *anything* we know about. - Remove machine-dependant code from the core PCI code. APIC interrupt mapping is performed by shadowing the intline register in machine- dependant code. - Bring interrupt routing support to the Alpha (although many platforms don't yet support routing or mapping interrupts entirely correctly). This resulted in spamming <sys/bus.h> into more places than it really should have gone. - Put sys/dev on the kernel/modules include path. This avoids having to change *all* the pci*.h includes.
2000-12-08 22:11:23 +00:00
}
Small nit: Make both prints use 'at device %d.%d'.
2001-09-01 23:06:14 +00:00
printf(" at device %d.%d (no driver attached)\n",
Make this file mostly conform to style(9). Approved by: msmith in principle before walkabout
2002-06-01 03:41:02 +00:00
pci_get_slot(child), pci_get_function(child));
Split power state control into two variables. hw.pci.do_powerstate has been removed. It has been replaced by hw.pci.do_power_nodriver and hw.pci.do_power_resume. The former defaults to 0 while the latter defaults to 1. When do_powerstate was set to 0, it broke suspend/resume for a lot of people as an unintended consequence. This change will only affect the areas that were intended to affect. This change will have no effect on servers, but will help laptops quite a bit. MFC After: 3 days.
2005-09-21 19:47:00 +00:00
if (pci_do_power_nodriver)
Move pci_do_powerstate up a level. Now it just means 'do not turn devices off into d3 state when there's no driver for the device'. This should help suspend/resume in the default case.
2004-05-21 06:39:09 +00:00
pci_cfg_save(child,
(struct pci_devinfo *) device_get_ivars(child), 1);
Next phase in the PCI subsystem cleanup. - Move PCI core code to dev/pci. - Split bridge code out into separate modules. - Remove the descriptive strings from the bridge drivers. If you want to know what a device is, use pciconf. Add support for broadly identifying devices based on class/subclass, and for parsing a preloaded device identification database so that if you want to waste the memory, you can identify *anything* we know about. - Remove machine-dependant code from the core PCI code. APIC interrupt mapping is performed by shadowing the intline register in machine- dependant code. - Bring interrupt routing support to the Alpha (although many platforms don't yet support routing or mapping interrupts entirely correctly). This resulted in spamming <sys/bus.h> into more places than it really should have gone. - Put sys/dev on the kernel/modules include path. This avoids having to change *all* the pci*.h includes.
2000-12-08 22:11:23 +00:00
return;
}
Move the pretty printing of the description for USB controllers to pci_probe_nomatch, so it won't be in the way when loading USB as a module. The reason for them being there in the first place is that every motherboard comes with USB kit and this way it looks more pretty (peter). The real solution will be to define some method of detaching a driver after it has attached.
1999-11-22 03:34:43 +00:00
Next phase in the PCI subsystem cleanup. - Move PCI core code to dev/pci. - Split bridge code out into separate modules. - Remove the descriptive strings from the bridge drivers. If you want to know what a device is, use pciconf. Add support for broadly identifying devices based on class/subclass, and for parsing a preloaded device identification database so that if you want to waste the memory, you can identify *anything* we know about. - Remove machine-dependant code from the core PCI code. APIC interrupt mapping is performed by shadowing the intline register in machine- dependant code. - Bring interrupt routing support to the Alpha (although many platforms don't yet support routing or mapping interrupts entirely correctly). This resulted in spamming <sys/bus.h> into more places than it really should have gone. - Put sys/dev on the kernel/modules include path. This avoids having to change *all* the pci*.h includes.
2000-12-08 22:11:23 +00:00
/*
* Parse the PCI device database, if loaded, and return a pointer to a
* description of the device.
*
* The database is flat text formatted as follows:
*
* Any line not in a valid format is ignored.
* Lines are terminated with newline '\n' characters.
*
* A VENDOR line consists of the 4 digit (hex) vendor code, a TAB, then
* the vendor name.
*
* A DEVICE line is entered immediately below the corresponding VENDOR ID.
* - devices cannot be listed without a corresponding VENDOR line.
* A DEVICE line consists of a TAB, the 4 digit (hex) device code,
* another TAB, then the device name.
*/
/*
* Assuming (ptr) points to the beginning of a line in the database,
* return the vendor or device and description of the next entry.
* The value of (vendor) or (device) inappropriate for the entry type
* is set to -1. Returns nonzero at the end of the database.
*
* Note that this is slightly unrobust in the face of corrupt data;
* we attempt to safeguard against this by spamming the end of the
* database with a newline when we initialise.
*/
static int
pci_describe_parse_line(char **ptr, int *vendor, int *device, char **desc)
{
char *cp = *ptr;
- Fix the device database parsing code so that it actually works. - Improve the formatting for devices identified by the database. - Fix the pcib_route_interrupt method definition, as an old version snuck in here somehow 8( - Remove a couple of the vendor/device IDs for PCI:ISA bridges which correctly identify themselves. Submitted by: peter
2000-12-09 09:15:38 +00:00
int left;
Next phase in the PCI subsystem cleanup. - Move PCI core code to dev/pci. - Split bridge code out into separate modules. - Remove the descriptive strings from the bridge drivers. If you want to know what a device is, use pciconf. Add support for broadly identifying devices based on class/subclass, and for parsing a preloaded device identification database so that if you want to waste the memory, you can identify *anything* we know about. - Remove machine-dependant code from the core PCI code. APIC interrupt mapping is performed by shadowing the intline register in machine- dependant code. - Bring interrupt routing support to the Alpha (although many platforms don't yet support routing or mapping interrupts entirely correctly). This resulted in spamming <sys/bus.h> into more places than it really should have gone. - Put sys/dev on the kernel/modules include path. This avoids having to change *all* the pci*.h includes.
2000-12-08 22:11:23 +00:00
*device = -1;
*vendor = -1;
- Fix the device database parsing code so that it actually works. - Improve the formatting for devices identified by the database. - Fix the pcib_route_interrupt method definition, as an old version snuck in here somehow 8( - Remove a couple of the vendor/device IDs for PCI:ISA bridges which correctly identify themselves. Submitted by: peter
2000-12-09 09:15:38 +00:00
**desc = '\0';
Next phase in the PCI subsystem cleanup. - Move PCI core code to dev/pci. - Split bridge code out into separate modules. - Remove the descriptive strings from the bridge drivers. If you want to know what a device is, use pciconf. Add support for broadly identifying devices based on class/subclass, and for parsing a preloaded device identification database so that if you want to waste the memory, you can identify *anything* we know about. - Remove machine-dependant code from the core PCI code. APIC interrupt mapping is performed by shadowing the intline register in machine- dependant code. - Bring interrupt routing support to the Alpha (although many platforms don't yet support routing or mapping interrupts entirely correctly). This resulted in spamming <sys/bus.h> into more places than it really should have gone. - Put sys/dev on the kernel/modules include path. This avoids having to change *all* the pci*.h includes.
2000-12-08 22:11:23 +00:00
for (;;) {
- Fix the device database parsing code so that it actually works. - Improve the formatting for devices identified by the database. - Fix the pcib_route_interrupt method definition, as an old version snuck in here somehow 8( - Remove a couple of the vendor/device IDs for PCI:ISA bridges which correctly identify themselves. Submitted by: peter
2000-12-09 09:15:38 +00:00
left = pci_vendordata_size - (cp - pci_vendordata);
if (left <= 0) {
*ptr = cp;
Next phase in the PCI subsystem cleanup. - Move PCI core code to dev/pci. - Split bridge code out into separate modules. - Remove the descriptive strings from the bridge drivers. If you want to know what a device is, use pciconf. Add support for broadly identifying devices based on class/subclass, and for parsing a preloaded device identification database so that if you want to waste the memory, you can identify *anything* we know about. - Remove machine-dependant code from the core PCI code. APIC interrupt mapping is performed by shadowing the intline register in machine- dependant code. - Bring interrupt routing support to the Alpha (although many platforms don't yet support routing or mapping interrupts entirely correctly). This resulted in spamming <sys/bus.h> into more places than it really should have gone. - Put sys/dev on the kernel/modules include path. This avoids having to change *all* the pci*.h includes.
2000-12-08 22:11:23 +00:00
return(1);
- Fix the device database parsing code so that it actually works. - Improve the formatting for devices identified by the database. - Fix the pcib_route_interrupt method definition, as an old version snuck in here somehow 8( - Remove a couple of the vendor/device IDs for PCI:ISA bridges which correctly identify themselves. Submitted by: peter
2000-12-09 09:15:38 +00:00
}
Next phase in the PCI subsystem cleanup. - Move PCI core code to dev/pci. - Split bridge code out into separate modules. - Remove the descriptive strings from the bridge drivers. If you want to know what a device is, use pciconf. Add support for broadly identifying devices based on class/subclass, and for parsing a preloaded device identification database so that if you want to waste the memory, you can identify *anything* we know about. - Remove machine-dependant code from the core PCI code. APIC interrupt mapping is performed by shadowing the intline register in machine- dependant code. - Bring interrupt routing support to the Alpha (although many platforms don't yet support routing or mapping interrupts entirely correctly). This resulted in spamming <sys/bus.h> into more places than it really should have gone. - Put sys/dev on the kernel/modules include path. This avoids having to change *all* the pci*.h includes.
2000-12-08 22:11:23 +00:00
/* vendor entry? */
Make this file mostly conform to style(9). Approved by: msmith in principle before walkabout
2002-06-01 03:41:02 +00:00
if (*cp != '\t' &&
sscanf(cp, "%x\t%80[^\n]", vendor, *desc) == 2)
Next phase in the PCI subsystem cleanup. - Move PCI core code to dev/pci. - Split bridge code out into separate modules. - Remove the descriptive strings from the bridge drivers. If you want to know what a device is, use pciconf. Add support for broadly identifying devices based on class/subclass, and for parsing a preloaded device identification database so that if you want to waste the memory, you can identify *anything* we know about. - Remove machine-dependant code from the core PCI code. APIC interrupt mapping is performed by shadowing the intline register in machine- dependant code. - Bring interrupt routing support to the Alpha (although many platforms don't yet support routing or mapping interrupts entirely correctly). This resulted in spamming <sys/bus.h> into more places than it really should have gone. - Put sys/dev on the kernel/modules include path. This avoids having to change *all* the pci*.h includes.
2000-12-08 22:11:23 +00:00
break;
/* device entry? */
Make this file mostly conform to style(9). Approved by: msmith in principle before walkabout
2002-06-01 03:41:02 +00:00
if (*cp == '\t' &&
sscanf(cp, "%x\t%80[^\n]", device, *desc) == 2)
Next phase in the PCI subsystem cleanup. - Move PCI core code to dev/pci. - Split bridge code out into separate modules. - Remove the descriptive strings from the bridge drivers. If you want to know what a device is, use pciconf. Add support for broadly identifying devices based on class/subclass, and for parsing a preloaded device identification database so that if you want to waste the memory, you can identify *anything* we know about. - Remove machine-dependant code from the core PCI code. APIC interrupt mapping is performed by shadowing the intline register in machine- dependant code. - Bring interrupt routing support to the Alpha (although many platforms don't yet support routing or mapping interrupts entirely correctly). This resulted in spamming <sys/bus.h> into more places than it really should have gone. - Put sys/dev on the kernel/modules include path. This avoids having to change *all* the pci*.h includes.
2000-12-08 22:11:23 +00:00
break;
/* skip to next line */
- Fix the device database parsing code so that it actually works. - Improve the formatting for devices identified by the database. - Fix the pcib_route_interrupt method definition, as an old version snuck in here somehow 8( - Remove a couple of the vendor/device IDs for PCI:ISA bridges which correctly identify themselves. Submitted by: peter
2000-12-09 09:15:38 +00:00
while (*cp != '\n' && left > 0) {
cp++;
left--;
}
if (*cp == '\n') {
Next phase in the PCI subsystem cleanup. - Move PCI core code to dev/pci. - Split bridge code out into separate modules. - Remove the descriptive strings from the bridge drivers. If you want to know what a device is, use pciconf. Add support for broadly identifying devices based on class/subclass, and for parsing a preloaded device identification database so that if you want to waste the memory, you can identify *anything* we know about. - Remove machine-dependant code from the core PCI code. APIC interrupt mapping is performed by shadowing the intline register in machine- dependant code. - Bring interrupt routing support to the Alpha (although many platforms don't yet support routing or mapping interrupts entirely correctly). This resulted in spamming <sys/bus.h> into more places than it really should have gone. - Put sys/dev on the kernel/modules include path. This avoids having to change *all* the pci*.h includes.
2000-12-08 22:11:23 +00:00
cp++;
- Fix the device database parsing code so that it actually works. - Improve the formatting for devices identified by the database. - Fix the pcib_route_interrupt method definition, as an old version snuck in here somehow 8( - Remove a couple of the vendor/device IDs for PCI:ISA bridges which correctly identify themselves. Submitted by: peter
2000-12-09 09:15:38 +00:00
left--;
}
}
/* skip to next line */
while (*cp != '\n' && left > 0) {
Next phase in the PCI subsystem cleanup. - Move PCI core code to dev/pci. - Split bridge code out into separate modules. - Remove the descriptive strings from the bridge drivers. If you want to know what a device is, use pciconf. Add support for broadly identifying devices based on class/subclass, and for parsing a preloaded device identification database so that if you want to waste the memory, you can identify *anything* we know about. - Remove machine-dependant code from the core PCI code. APIC interrupt mapping is performed by shadowing the intline register in machine- dependant code. - Bring interrupt routing support to the Alpha (although many platforms don't yet support routing or mapping interrupts entirely correctly). This resulted in spamming <sys/bus.h> into more places than it really should have gone. - Put sys/dev on the kernel/modules include path. This avoids having to change *all* the pci*.h includes.
2000-12-08 22:11:23 +00:00
cp++;
- Fix the device database parsing code so that it actually works. - Improve the formatting for devices identified by the database. - Fix the pcib_route_interrupt method definition, as an old version snuck in here somehow 8( - Remove a couple of the vendor/device IDs for PCI:ISA bridges which correctly identify themselves. Submitted by: peter
2000-12-09 09:15:38 +00:00
left--;
o No need to print the vendor/device ID for things that matched succesfully. o ``<device name>'' versus ``device name'' for things that fall under nomatch. Reviewed by: dfr (in principle) Approved by: Baron von Hubbard
2000-02-22 21:44:39 +00:00
}
- Fix the device database parsing code so that it actually works. - Improve the formatting for devices identified by the database. - Fix the pcib_route_interrupt method definition, as an old version snuck in here somehow 8( - Remove a couple of the vendor/device IDs for PCI:ISA bridges which correctly identify themselves. Submitted by: peter
2000-12-09 09:15:38 +00:00
if (*cp == '\n' && left > 0)
cp++;
Next phase in the PCI subsystem cleanup. - Move PCI core code to dev/pci. - Split bridge code out into separate modules. - Remove the descriptive strings from the bridge drivers. If you want to know what a device is, use pciconf. Add support for broadly identifying devices based on class/subclass, and for parsing a preloaded device identification database so that if you want to waste the memory, you can identify *anything* we know about. - Remove machine-dependant code from the core PCI code. APIC interrupt mapping is performed by shadowing the intline register in machine- dependant code. - Bring interrupt routing support to the Alpha (although many platforms don't yet support routing or mapping interrupts entirely correctly). This resulted in spamming <sys/bus.h> into more places than it really should have gone. - Put sys/dev on the kernel/modules include path. This avoids having to change *all* the pci*.h includes.
2000-12-08 22:11:23 +00:00
*ptr = cp;
return(0);
}
static char *
pci_describe_device(device_t dev)
{
int vendor, device;
char *desc, *vp, *dp, *line;
desc = vp = dp = NULL;
/*
* If we have no vendor data, we can't do anything.
*/
if (pci_vendordata == NULL)
goto out;
/*
* Scan the vendor data looking for this device
*/
line = pci_vendordata;
if ((vp = malloc(80, M_DEVBUF, M_NOWAIT)) == NULL)
goto out;
for (;;) {
if (pci_describe_parse_line(&line, &vendor, &device, &vp))
goto out;
if (vendor == pci_get_vendor(dev))
break;
o No need to print the vendor/device ID for things that matched succesfully. o ``<device name>'' versus ``device name'' for things that fall under nomatch. Reviewed by: dfr (in principle) Approved by: Baron von Hubbard
2000-02-22 21:44:39 +00:00
}
Next phase in the PCI subsystem cleanup. - Move PCI core code to dev/pci. - Split bridge code out into separate modules. - Remove the descriptive strings from the bridge drivers. If you want to know what a device is, use pciconf. Add support for broadly identifying devices based on class/subclass, and for parsing a preloaded device identification database so that if you want to waste the memory, you can identify *anything* we know about. - Remove machine-dependant code from the core PCI code. APIC interrupt mapping is performed by shadowing the intline register in machine- dependant code. - Bring interrupt routing support to the Alpha (although many platforms don't yet support routing or mapping interrupts entirely correctly). This resulted in spamming <sys/bus.h> into more places than it really should have gone. - Put sys/dev on the kernel/modules include path. This avoids having to change *all* the pci*.h includes.
2000-12-08 22:11:23 +00:00
if ((dp = malloc(80, M_DEVBUF, M_NOWAIT)) == NULL)
goto out;
for (;;) {
- Fix the device database parsing code so that it actually works. - Improve the formatting for devices identified by the database. - Fix the pcib_route_interrupt method definition, as an old version snuck in here somehow 8( - Remove a couple of the vendor/device IDs for PCI:ISA bridges which correctly identify themselves. Submitted by: peter
2000-12-09 09:15:38 +00:00
if (pci_describe_parse_line(&line, &vendor, &device, &dp)) {
*dp = 0;
break;
}
Next phase in the PCI subsystem cleanup. - Move PCI core code to dev/pci. - Split bridge code out into separate modules. - Remove the descriptive strings from the bridge drivers. If you want to know what a device is, use pciconf. Add support for broadly identifying devices based on class/subclass, and for parsing a preloaded device identification database so that if you want to waste the memory, you can identify *anything* we know about. - Remove machine-dependant code from the core PCI code. APIC interrupt mapping is performed by shadowing the intline register in machine- dependant code. - Bring interrupt routing support to the Alpha (although many platforms don't yet support routing or mapping interrupts entirely correctly). This resulted in spamming <sys/bus.h> into more places than it really should have gone. - Put sys/dev on the kernel/modules include path. This avoids having to change *all* the pci*.h includes.
2000-12-08 22:11:23 +00:00
if (vendor != -1) {
*dp = 0;
break;
}
if (device == pci_get_device(dev))
break;
Implement the BUS_PROBE_NOMATCH method for the PCI bus. This function is called for each device for which no driver was found. Output is similar to the eisa_probe_nomatch() function but with the added benefit of displaying the assigned IRQ (since PCI gives us this information up front.) Output is like so: pci0: unknown card CPQ0508 (vendor=0x0e11, dev=0x0508) at 11.0 irq 9 pci0: unknown card DFZ0508 (vendor=0x10da, dev=0x0508) at 11.0 irq 9 pci0: unknown card DBL0508 (vendor=0x104c, dev=0x0508) at 11.0 irq 9 pci0: unknown card DDM0011 (vendor=0x108d, dev=0x0011) at 11.0 irq 9 I'm not happy with the 3 lines of macro cruft that got added but I consider it a temporary annoyance as those bits will be moved to some place where PCI, EISA and ISAPNP code will be able to use them. (Not surprisingly, this message is longer than the code in question.) Reviewed by: peter, dfr
1999-07-27 04:28:14 +00:00
}
- Fix the device database parsing code so that it actually works. - Improve the formatting for devices identified by the database. - Fix the pcib_route_interrupt method definition, as an old version snuck in here somehow 8( - Remove a couple of the vendor/device IDs for PCI:ISA bridges which correctly identify themselves. Submitted by: peter
2000-12-09 09:15:38 +00:00
if (dp[0] == '\0')
snprintf(dp, 80, "0x%x", pci_get_device(dev));
Make this file mostly conform to style(9). Approved by: msmith in principle before walkabout
2002-06-01 03:41:02 +00:00
if ((desc = malloc(strlen(vp) + strlen(dp) + 3, M_DEVBUF, M_NOWAIT)) !=
NULL)
- Fix the device database parsing code so that it actually works. - Improve the formatting for devices identified by the database. - Fix the pcib_route_interrupt method definition, as an old version snuck in here somehow 8( - Remove a couple of the vendor/device IDs for PCI:ISA bridges which correctly identify themselves. Submitted by: peter
2000-12-09 09:15:38 +00:00
sprintf(desc, "%s, %s", vp, dp);
Next phase in the PCI subsystem cleanup. - Move PCI core code to dev/pci. - Split bridge code out into separate modules. - Remove the descriptive strings from the bridge drivers. If you want to know what a device is, use pciconf. Add support for broadly identifying devices based on class/subclass, and for parsing a preloaded device identification database so that if you want to waste the memory, you can identify *anything* we know about. - Remove machine-dependant code from the core PCI code. APIC interrupt mapping is performed by shadowing the intline register in machine- dependant code. - Bring interrupt routing support to the Alpha (although many platforms don't yet support routing or mapping interrupts entirely correctly). This resulted in spamming <sys/bus.h> into more places than it really should have gone. - Put sys/dev on the kernel/modules include path. This avoids having to change *all* the pci*.h includes.
2000-12-08 22:11:23 +00:00
out:
if (vp != NULL)
free(vp, M_DEVBUF);
if (dp != NULL)
free(dp, M_DEVBUF);
return(desc);
Implement the BUS_PROBE_NOMATCH method for the PCI bus. This function is called for each device for which no driver was found. Output is similar to the eisa_probe_nomatch() function but with the added benefit of displaying the assigned IRQ (since PCI gives us this information up front.) Output is like so: pci0: unknown card CPQ0508 (vendor=0x0e11, dev=0x0508) at 11.0 irq 9 pci0: unknown card DFZ0508 (vendor=0x10da, dev=0x0508) at 11.0 irq 9 pci0: unknown card DBL0508 (vendor=0x104c, dev=0x0508) at 11.0 irq 9 pci0: unknown card DDM0011 (vendor=0x108d, dev=0x0011) at 11.0 irq 9 I'm not happy with the 3 lines of macro cruft that got added but I consider it a temporary annoyance as those bits will be moved to some place where PCI, EISA and ISAPNP code will be able to use them. (Not surprisingly, this message is longer than the code in question.) Reviewed by: peter, dfr
1999-07-27 04:28:14 +00:00
}
Use the pci.c code wherever possible, rather than copying all the pci code into cardbus and s/pci/cardbus. This exposes a few pci_* functions that are now static. This work is similar to work Justin posted to the mobile list about a year or two ago, which I have neglected since then. This is a subset of his current work with the multiple inheritance newbus architecutre. When completed, that will eliminate the need for pci/pci_private.h. Similar work is needed for the cardbus_cis and pccard_cis code as well.
2002-02-27 05:09:14 +00:00
int
Fixed the type of some ivar access functions. Ivars have type uintptr_t, not u_long. On i386's with 64-bit longs, returning u_longs indirectly in (more than) the space reserved for uintptr_t's tended to corrupt the previous frame pointer in the stack frame, so it was not easy to debug. The type mismatches are hidden by the bogus cast in DEVMETHOD().
2000-04-30 10:01:56 +00:00
pci_read_ivar(device_t dev, device_t child, int which, uintptr_t *result)
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
{
struct pci_devinfo *dinfo;
pcicfgregs *cfg;
dinfo = device_get_ivars(child);
cfg = &dinfo->cfg;
switch (which) {
Add a new cardbus specific ivar: PCI_IVAR_ETHADDR. Some pci-like buses support querying the MAC address in a standard-for-that-bus way. The base pci bus returns NULL for this IVAR always. Submitted by: sam Approved by: re (blanket for NEWCARD)
2002-11-27 06:41:28 +00:00
case PCI_IVAR_ETHADDR:
/*
* The generic accessor doesn't deal with failure, so
* we set the return value, then return an error.
*/
Prefer the uintXX_t to the u_intXX_t names.
2003-08-22 03:11:53 +00:00
*((uint8_t **) result) = NULL;
Add a new cardbus specific ivar: PCI_IVAR_ETHADDR. Some pci-like buses support querying the MAC address in a standard-for-that-bus way. The base pci bus returns NULL for this IVAR always. Submitted by: sam Approved by: re (blanket for NEWCARD)
2002-11-27 06:41:28 +00:00
return (EINVAL);
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
case PCI_IVAR_SUBVENDOR:
*result = cfg->subvendor;
break;
case PCI_IVAR_SUBDEVICE:
*result = cfg->subdevice;
break;
case PCI_IVAR_VENDOR:
*result = cfg->vendor;
break;
case PCI_IVAR_DEVICE:
*result = cfg->device;
break;
case PCI_IVAR_DEVID:
*result = (cfg->device << 16) | cfg->vendor;
break;
case PCI_IVAR_CLASS:
*result = cfg->baseclass;
break;
case PCI_IVAR_SUBCLASS:
*result = cfg->subclass;
break;
case PCI_IVAR_PROGIF:
*result = cfg->progif;
break;
case PCI_IVAR_REVID:
*result = cfg->revid;
break;
case PCI_IVAR_INTPIN:
*result = cfg->intpin;
break;
case PCI_IVAR_IRQ:
*result = cfg->intline;
break;
case PCI_IVAR_BUS:
*result = cfg->bus;
break;
case PCI_IVAR_SLOT:
*result = cfg->slot;
break;
case PCI_IVAR_FUNCTION:
*result = cfg->func;
break;
Allow one to access the cached values for CMDREG, CACHELNSZ, MINGNT, MAXLAT and LATTIMER. Improve error message when a bogus RID type is requested for a bar.
2005-09-11 03:22:03 +00:00
case PCI_IVAR_CMDREG:
*result = cfg->cmdreg;
break;
case PCI_IVAR_CACHELNSZ:
*result = cfg->cachelnsz;
break;
case PCI_IVAR_MINGNT:
*result = cfg->mingnt;
break;
case PCI_IVAR_MAXLAT:
*result = cfg->maxlat;
break;
case PCI_IVAR_LATTIMER:
*result = cfg->lattimer;
break;
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
default:
More style(9) nits
2002-06-01 05:44:45 +00:00
return (ENOENT);
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
}
More style(9) nits
2002-06-01 05:44:45 +00:00
return (0);
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
}
Use the pci.c code wherever possible, rather than copying all the pci code into cardbus and s/pci/cardbus. This exposes a few pci_* functions that are now static. This work is similar to work Justin posted to the mobile list about a year or two ago, which I have neglected since then. This is a subset of his current work with the multiple inheritance newbus architecutre. When completed, that will eliminate the need for pci/pci_private.h. Similar work is needed for the cardbus_cis and pccard_cis code as well.
2002-02-27 05:09:14 +00:00
int
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
pci_write_ivar(device_t dev, device_t child, int which, uintptr_t value)
{
struct pci_devinfo *dinfo;
dinfo = device_get_ivars(child);
switch (which) {
Allow to write the intpin ivar using the pci_set_intpin() accessor. There are some Sun PCI devices around which bogusly set intpin to 0, although they use the intline mechanism; this allows the device driver to correct that. Reviewed by: imp
2003-07-01 13:54:10 +00:00
case PCI_IVAR_INTPIN:
dinfo->cfg.intpin = value;
return (0);
Add a new cardbus specific ivar: PCI_IVAR_ETHADDR. Some pci-like buses support querying the MAC address in a standard-for-that-bus way. The base pci bus returns NULL for this IVAR always. Submitted by: sam Approved by: re (blanket for NEWCARD)
2002-11-27 06:41:28 +00:00
case PCI_IVAR_ETHADDR:
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
case PCI_IVAR_SUBVENDOR:
case PCI_IVAR_SUBDEVICE:
case PCI_IVAR_VENDOR:
case PCI_IVAR_DEVICE:
case PCI_IVAR_DEVID:
case PCI_IVAR_CLASS:
case PCI_IVAR_SUBCLASS:
case PCI_IVAR_PROGIF:
case PCI_IVAR_REVID:
case PCI_IVAR_IRQ:
case PCI_IVAR_BUS:
case PCI_IVAR_SLOT:
case PCI_IVAR_FUNCTION:
More style(9) nits
2002-06-01 05:44:45 +00:00
return (EINVAL); /* disallow for now */
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
default:
More style(9) nits
2002-06-01 05:44:45 +00:00
return (ENOENT);
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
}
}
Add a new command: show pciregs, equivalent to pciconf -l Reviewed by: Doug Rabson (quite some time ago) MFC after: 1 week
2002-08-28 10:02:59 +00:00
#include "opt_ddb.h"
#ifdef DDB
#include <ddb/ddb.h>
#include <sys/cons.h>
/*
* List resources based on pci map registers, used for within ddb
*/
DB_SHOW_COMMAND(pciregs, db_pci_dump)
{
struct pci_devinfo *dinfo;
struct devlist *devlist_head;
struct pci_conf *p;
const char *name;
Simplify the pager support in DDB. Allowing different db commands to install custom pager functions didn't actually happen in practice (they all just used the simple pager and passed in a local quit pointer). So, just hardcode the simple pager as the only pager and make it set a global db_pager_quit flag that db commands can check when the user hits 'q' (or a suitable variant) at the pager prompt. Also, now that it's easy to do so, enable paging by default for all ddb commands. Any command that wishes to honor the quit flag can do so by checking db_pager_quit. Note that the pager can also be effectively disabled by setting $lines to 0. Other fixes: - 'show idt' on i386 and pc98 now actually checks the quit flag and terminates early. - 'show intr' now actually checks the quit flag and terminates early.
2006-07-12 21:22:44 +00:00
int i, error, none_count;
Add a new command: show pciregs, equivalent to pciconf -l Reviewed by: Doug Rabson (quite some time ago) MFC after: 1 week
2002-08-28 10:02:59 +00:00
none_count = 0;
/* get the head of the device queue */
devlist_head = &pci_devq;
/*
* Go through the list of devices and print out devices
*/
Simplify the pager support in DDB. Allowing different db commands to install custom pager functions didn't actually happen in practice (they all just used the simple pager and passed in a local quit pointer). So, just hardcode the simple pager as the only pager and make it set a global db_pager_quit flag that db commands can check when the user hits 'q' (or a suitable variant) at the pager prompt. Also, now that it's easy to do so, enable paging by default for all ddb commands. Any command that wishes to honor the quit flag can do so by checking db_pager_quit. Note that the pager can also be effectively disabled by setting $lines to 0. Other fixes: - 'show idt' on i386 and pc98 now actually checks the quit flag and terminates early. - 'show intr' now actually checks the quit flag and terminates early.
2006-07-12 21:22:44 +00:00
for (error = 0, i = 0,
Add a new command: show pciregs, equivalent to pciconf -l Reviewed by: Doug Rabson (quite some time ago) MFC after: 1 week
2002-08-28 10:02:59 +00:00
dinfo = STAILQ_FIRST(devlist_head);
Simplify the pager support in DDB. Allowing different db commands to install custom pager functions didn't actually happen in practice (they all just used the simple pager and passed in a local quit pointer). So, just hardcode the simple pager as the only pager and make it set a global db_pager_quit flag that db commands can check when the user hits 'q' (or a suitable variant) at the pager prompt. Also, now that it's easy to do so, enable paging by default for all ddb commands. Any command that wishes to honor the quit flag can do so by checking db_pager_quit. Note that the pager can also be effectively disabled by setting $lines to 0. Other fixes: - 'show idt' on i386 and pc98 now actually checks the quit flag and terminates early. - 'show intr' now actually checks the quit flag and terminates early.
2006-07-12 21:22:44 +00:00
(dinfo != NULL) && (error == 0) && (i < pci_numdevs) && !db_pager_quit;
Add a new command: show pciregs, equivalent to pciconf -l Reviewed by: Doug Rabson (quite some time ago) MFC after: 1 week
2002-08-28 10:02:59 +00:00
dinfo = STAILQ_NEXT(dinfo, pci_links), i++) {
/* Populate pd_name and pd_unit */
name = NULL;
if (dinfo->cfg.dev)
name = device_get_name(dinfo->cfg.dev);
p = &dinfo->conf;
db_printf("%s%d@pci%d:%d:%d:\tclass=0x%06x card=0x%08x "
"chip=0x%08x rev=0x%02x hdr=0x%02x\n",
(name && *name) ? name : "none",
(name && *name) ? (int)device_get_unit(dinfo->cfg.dev) :
none_count++,
p->pc_sel.pc_bus, p->pc_sel.pc_dev,
p->pc_sel.pc_func, (p->pc_class << 16) |
(p->pc_subclass << 8) | p->pc_progif,
(p->pc_subdevice << 16) | p->pc_subvendor,
(p->pc_device << 16) | p->pc_vendor,
p->pc_revid, p->pc_hdr);
}
}
#endif /* DDB */
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
static struct resource *
pci_alloc_map(device_t dev, device_t child, int type, int *rid,
u_long start, u_long end, u_long count, u_int flags)
{
struct pci_devinfo *dinfo = device_get_ivars(child);
struct resource_list *rl = &dinfo->resources;
struct resource_list_entry *rle;
struct resource *res;
uint32_t map, testval;
int mapsize;
/*
Some devices have what appear to be invalid BARs. They are invalid in the sense that any write to them reads back as a 0. This presents a problem to our resource allocation scheme. If we encounter such vars, the code now treats them as special, allowing any allocation against them to succeed. I've not seen anything in the standard to clearify what host software should do when it encounters these sorts of BARs. Also cleaned up some output while I'm here and add commmented out bootverbose lines until I'm ready to reduce the verbosity of boot messages. This gets a number of south bridges and ata controllers made mostly by VIA, AMD and nVidia working again. Thanks to Soren Schmidt for his help in coming up with this patch.
2004-04-13 19:31:57 +00:00
* Weed out the bogons, and figure out how large the BAR/map
ata devices in legacy are special, and we must treat them as such. While I would have prefered to have a solution that didn't move knowledge of this into the pci layer. However, this is literally the only exception that's listed in the PCI standard to the usual way of decoding BARs. atapci devices in legacy mode now ignore the first 4 bars and hard code the values to the legacy ide values (well, for each of the controllers that are in legacy mode). The 5th bar is handled normally. Remove the zero bar handling. zero bars should be ignored at all other times, and since we handle that specially, we don't need the older workaround.
2004-04-21 20:19:56 +00:00
* is. Bars that read back 0 here are bogus and unimplemented.
* Note: atapci in legacy mode are special and handled elsewhere
* in the code. If you have a atapci device in legacy mode and
* it fails here, that other code is broken.
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
*/
Some devices have what appear to be invalid BARs. They are invalid in the sense that any write to them reads back as a 0. This presents a problem to our resource allocation scheme. If we encounter such vars, the code now treats them as special, allowing any allocation against them to succeed. I've not seen anything in the standard to clearify what host software should do when it encounters these sorts of BARs. Also cleaned up some output while I'm here and add commmented out bootverbose lines until I'm ready to reduce the verbosity of boot messages. This gets a number of south bridges and ata controllers made mostly by VIA, AMD and nVidia working again. Thanks to Soren Schmidt for his help in coming up with this patch.
2004-04-13 19:31:57 +00:00
res = NULL;
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
map = pci_read_config(child, *rid, 4);
pci_write_config(child, *rid, 0xffffffff, 4);
testval = pci_read_config(child, *rid, 4);
Mask off the bar's value after the probe test write before testing against 0 in pci_alloc_map, just like we do in pci_add_map. Also, make sure that we restore the value to the BAR that was there before if the bar is 0. Chances are that it was 0 before the write too and that the restoration is a nop, but better safe than sorry. Notice by: dwhite
2005-06-03 19:41:06 +00:00
if (pci_mapbase(testval) == 0)
goto out;
ata devices in legacy are special, and we must treat them as such. While I would have prefered to have a solution that didn't move knowledge of this into the pci layer. However, this is literally the only exception that's listed in the PCI standard to the usual way of decoding BARs. atapci devices in legacy mode now ignore the first 4 bars and hard code the values to the legacy ide values (well, for each of the controllers that are in legacy mode). The 5th bar is handled normally. Remove the zero bar handling. zero bars should be ignored at all other times, and since we handle that specially, we don't need the older workaround.
2004-04-21 20:19:56 +00:00
if (pci_maptype(testval) & PCI_MAPMEM) {
if (type != SYS_RES_MEMORY) {
It isn't a whinable offence to want memory when the bar says ioport. Put that behind bootverbose to make the ata driver less chatty on advanced hardware. Requested by: sos
2005-04-11 02:08:05 +00:00
if (bootverbose)
Allow one to access the cached values for CMDREG, CACHELNSZ, MINGNT, MAXLAT and LATTIMER. Improve error message when a bogus RID type is requested for a bar.
2005-09-11 03:22:03 +00:00
device_printf(dev,
"child %s requested type %d for rid %#x,"
" but the BAR says it is an memio\n",
device_get_nameunit(child), type, *rid);
ata devices in legacy are special, and we must treat them as such. While I would have prefered to have a solution that didn't move knowledge of this into the pci layer. However, this is literally the only exception that's listed in the PCI standard to the usual way of decoding BARs. atapci devices in legacy mode now ignore the first 4 bars and hard code the values to the legacy ide values (well, for each of the controllers that are in legacy mode). The 5th bar is handled normally. Remove the zero bar handling. zero bars should be ignored at all other times, and since we handle that specially, we don't need the older workaround.
2004-04-21 20:19:56 +00:00
goto out;
Some devices have what appear to be invalid BARs. They are invalid in the sense that any write to them reads back as a 0. This presents a problem to our resource allocation scheme. If we encounter such vars, the code now treats them as special, allowing any allocation against them to succeed. I've not seen anything in the standard to clearify what host software should do when it encounters these sorts of BARs. Also cleaned up some output while I'm here and add commmented out bootverbose lines until I'm ready to reduce the verbosity of boot messages. This gets a number of south bridges and ata controllers made mostly by VIA, AMD and nVidia working again. Thanks to Soren Schmidt for his help in coming up with this patch.
2004-04-13 19:31:57 +00:00
}
ata devices in legacy are special, and we must treat them as such. While I would have prefered to have a solution that didn't move knowledge of this into the pci layer. However, this is literally the only exception that's listed in the PCI standard to the usual way of decoding BARs. atapci devices in legacy mode now ignore the first 4 bars and hard code the values to the legacy ide values (well, for each of the controllers that are in legacy mode). The 5th bar is handled normally. Remove the zero bar handling. zero bars should be ignored at all other times, and since we handle that specially, we don't need the older workaround.
2004-04-21 20:19:56 +00:00
} else {
if (type != SYS_RES_IOPORT) {
It isn't a whinable offence to want memory when the bar says ioport. Put that behind bootverbose to make the ata driver less chatty on advanced hardware. Requested by: sos
2005-04-11 02:08:05 +00:00
if (bootverbose)
Allow one to access the cached values for CMDREG, CACHELNSZ, MINGNT, MAXLAT and LATTIMER. Improve error message when a bogus RID type is requested for a bar.
2005-09-11 03:22:03 +00:00
device_printf(dev,
"child %s requested type %d for rid %#x,"
" but the BAR says it is an ioport\n",
device_get_nameunit(child), type, *rid);
ata devices in legacy are special, and we must treat them as such. While I would have prefered to have a solution that didn't move knowledge of this into the pci layer. However, this is literally the only exception that's listed in the PCI standard to the usual way of decoding BARs. atapci devices in legacy mode now ignore the first 4 bars and hard code the values to the legacy ide values (well, for each of the controllers that are in legacy mode). The 5th bar is handled normally. Remove the zero bar handling. zero bars should be ignored at all other times, and since we handle that specially, we don't need the older workaround.
2004-04-21 20:19:56 +00:00
goto out;
}
Some devices have what appear to be invalid BARs. They are invalid in the sense that any write to them reads back as a 0. This presents a problem to our resource allocation scheme. If we encounter such vars, the code now treats them as special, allowing any allocation against them to succeed. I've not seen anything in the standard to clearify what host software should do when it encounters these sorts of BARs. Also cleaned up some output while I'm here and add commmented out bootverbose lines until I'm ready to reduce the verbosity of boot messages. This gets a number of south bridges and ata controllers made mostly by VIA, AMD and nVidia working again. Thanks to Soren Schmidt for his help in coming up with this patch.
2004-04-13 19:31:57 +00:00
}
ata devices in legacy are special, and we must treat them as such. While I would have prefered to have a solution that didn't move knowledge of this into the pci layer. However, this is literally the only exception that's listed in the PCI standard to the usual way of decoding BARs. atapci devices in legacy mode now ignore the first 4 bars and hard code the values to the legacy ide values (well, for each of the controllers that are in legacy mode). The 5th bar is handled normally. Remove the zero bar handling. zero bars should be ignored at all other times, and since we handle that specially, we don't need the older workaround.
2004-04-21 20:19:56 +00:00
/*
* For real BARs, we need to override the size that
* the driver requests, because that's what the BAR
* actually uses and we would otherwise have a
* situation where we might allocate the excess to
* another driver, which won't work.
*/
mapsize = pci_mapsize(testval);
count = 1 << mapsize;
if (RF_ALIGNMENT(flags) < mapsize)
flags = (flags & ~RF_ALIGNMENT_MASK) | RF_ALIGNMENT_LOG2(mapsize);
Some devices have what appear to be invalid BARs. They are invalid in the sense that any write to them reads back as a 0. This presents a problem to our resource allocation scheme. If we encounter such vars, the code now treats them as special, allowing any allocation against them to succeed. I've not seen anything in the standard to clearify what host software should do when it encounters these sorts of BARs. Also cleaned up some output while I'm here and add commmented out bootverbose lines until I'm ready to reduce the verbosity of boot messages. This gets a number of south bridges and ata controllers made mostly by VIA, AMD and nVidia working again. Thanks to Soren Schmidt for his help in coming up with this patch.
2004-04-13 19:31:57 +00:00
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
/*
* Allocate enough resource, and then write back the
Some devices have what appear to be invalid BARs. They are invalid in the sense that any write to them reads back as a 0. This presents a problem to our resource allocation scheme. If we encounter such vars, the code now treats them as special, allowing any allocation against them to succeed. I've not seen anything in the standard to clearify what host software should do when it encounters these sorts of BARs. Also cleaned up some output while I'm here and add commmented out bootverbose lines until I'm ready to reduce the verbosity of boot messages. This gets a number of south bridges and ata controllers made mostly by VIA, AMD and nVidia working again. Thanks to Soren Schmidt for his help in coming up with this patch.
2004-04-13 19:31:57 +00:00
* appropriate bar for that resource.
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
*/
res = BUS_ALLOC_RESOURCE(device_get_parent(dev), child, type, rid,
start, end, count, flags);
if (res == NULL) {
Improve diagnostic message.
2005-11-09 03:37:52 +00:00
device_printf(child,
"%#lx bytes of rid %#x res %d failed (%#lx, %#lx).\n",
count, *rid, type, start, end);
Some devices have what appear to be invalid BARs. They are invalid in the sense that any write to them reads back as a 0. This presents a problem to our resource allocation scheme. If we encounter such vars, the code now treats them as special, allowing any allocation against them to succeed. I've not seen anything in the standard to clearify what host software should do when it encounters these sorts of BARs. Also cleaned up some output while I'm here and add commmented out bootverbose lines until I'm ready to reduce the verbosity of boot messages. This gets a number of south bridges and ata controllers made mostly by VIA, AMD and nVidia working again. Thanks to Soren Schmidt for his help in coming up with this patch.
2004-04-13 19:31:57 +00:00
goto out;
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
}
resource_list_add(rl, type, *rid, start, end, count);
rle = resource_list_find(rl, type, *rid);
if (rle == NULL)
Fix a spelling error in a panic string.
2004-10-14 03:05:39 +00:00
panic("pci_alloc_map: unexpectedly can't find resource.");
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
rle->res = res;
MFp4: When doing lazy allocation, it turns out that we need to record the actual resource values we received from the system rather than the range we requested. Since we request a range starting at 0, we would record that number. Later, since this == 0, we'd allocate again. However, we wouldn't write the new resource into the BAR. This resulted in a resource leak as well as a BAR that couldn't access the resource at all since rman_get_start, et al, were wrong. MFC After: 1 week (assuming RELENG_6 is open for business)
2005-10-29 05:52:17 +00:00
rle->start = rman_get_start(res);
rle->end = rman_get_end(res);
rle->count = count;
make the pci power state and resource code a lot less chatty. The chattiness was left in for debugging, but now that nearly all of the problems relating to the changes have been fixed, it is only annoying. It is still available via bootverbose. Prodded by: jhb
2004-05-21 06:03:26 +00:00
if (bootverbose)
device_printf(child,
"Lazy allocation of %#lx bytes rid %#x type %d at %#lx\n",
count, *rid, type, rman_get_start(res));
Some devices have what appear to be invalid BARs. They are invalid in the sense that any write to them reads back as a 0. This presents a problem to our resource allocation scheme. If we encounter such vars, the code now treats them as special, allowing any allocation against them to succeed. I've not seen anything in the standard to clearify what host software should do when it encounters these sorts of BARs. Also cleaned up some output while I'm here and add commmented out bootverbose lines until I'm ready to reduce the verbosity of boot messages. This gets a number of south bridges and ata controllers made mostly by VIA, AMD and nVidia working again. Thanks to Soren Schmidt for his help in coming up with this patch.
2004-04-13 19:31:57 +00:00
map = rman_get_start(res);
out:;
pci_write_config(child, *rid, map, 4);
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
return (res);
}
Use the pci.c code wherever possible, rather than copying all the pci code into cardbus and s/pci/cardbus. This exposes a few pci_* functions that are now static. This work is similar to work Justin posted to the mobile list about a year or two ago, which I have neglected since then. This is a subset of his current work with the multiple inheritance newbus architecutre. When completed, that will eliminate the need for pci/pci_private.h. Similar work is needed for the cardbus_cis and pccard_cis code as well.
2002-02-27 05:09:14 +00:00
struct resource *
* Implement bus_set/get/delete_resource for pci. * Change the hack used on the alpha for mapping devices into DENSE or BWX memory spaces to a simpler one. Its still a hack and should be a seperate api to explicitly map the resource. * Add $FreeBSD$ as necessary.
1999-10-14 21:38:33 +00:00
pci_alloc_resource(device_t dev, device_t child, int type, int *rid,
u_long start, u_long end, u_long count, u_int flags)
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
{
* Implement bus_set/get/delete_resource for pci. * Change the hack used on the alpha for mapping devices into DENSE or BWX memory spaces to a simpler one. Its still a hack and should be a seperate api to explicitly map the resource. * Add $FreeBSD$ as necessary.
1999-10-14 21:38:33 +00:00
struct pci_devinfo *dinfo = device_get_ivars(child);
struct resource_list *rl = &dinfo->resources;
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
struct resource_list_entry *rle;
When a pci device hasn't had an interrupt routed to it (signified by the intline of 255) go ahead and route the interrupt when we allocate an interrupt. Submitted by: msmith
2000-10-16 07:24:00 +00:00
pcicfgregs *cfg = &dinfo->cfg;
/*
* Perform lazy resource allocation
*/
if (device_get_parent(child) == dev) {
When the driver allocates memory or I/O ports, enable that bit in the command config register. At the present, this represents a nop because these bits should have been set earlier in the process. In the future, we'll only set these bits when the driver requests the resource, not when the bus code detects the resource. Reviewed by: mdodd
2003-04-15 19:38:18 +00:00
switch (type) {
case SYS_RES_IRQ:
/*
* If the child device doesn't have an
* interrupt routed and is deserving of an
* interrupt, try to assign it one.
*/
if (!PCI_INTERRUPT_VALID(cfg->intline) &&
- Consolidate duplicated code for assigning interrupts to PCI devices via routing, etc. in a static pci_assign_interrupt() function. - Add a sledgehammer that allows the user to override the interrupt assignment of any PCI device via a tunable (e.g. "hw.pci0.7.INTB=5" would force any functions on the pci device in slot 7 of bus 0 that use B# to use IRQ 5). This should be used with great caution! Generally, if the interrupt routing in use provides specific tunables (such as hard-wiring the IRQ for a given $PIR or ACPI PCI link device), then those should be used instead. One instance where this tunable might be useful is if a box has an MPTable with duplicate entries for the same PCI device with different IRQs. MFC after: 1 week
2005-09-29 15:04:41 +00:00
(cfg->intpin != 0))
pci_assign_interrupt(dev, child, 0);
When the driver allocates memory or I/O ports, enable that bit in the command config register. At the present, this represents a nop because these bits should have been set earlier in the process. In the future, we'll only set these bits when the driver requests the resource, not when the bus code detects the resource. Reviewed by: mdodd
2003-04-15 19:38:18 +00:00
break;
case SYS_RES_IOPORT:
case SYS_RES_MEMORY:
Replace another instance of PCIR_MAPS with PCIR_BAR(x). Reminded by: dfr
2003-09-03 15:24:31 +00:00
if (*rid < PCIR_BAR(cfg->nummaps)) {
Don't try to enable io or memory access for non-standard resource addresses. This stops resource allocations for e.g. amdpm failing - this has its own special ways of enabling access.
2003-09-01 15:01:49 +00:00
/*
* Enable the I/O mode. We should
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
* also be assigning resources too
* when none are present. The
* resource_list_alloc kind of sorta does
* this...
Don't try to enable io or memory access for non-standard resource addresses. This stops resource allocations for e.g. amdpm failing - this has its own special ways of enabling access.
2003-09-01 15:01:49 +00:00
*/
if (PCI_ENABLE_IO(dev, child, type))
return (NULL);
}
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
rle = resource_list_find(rl, type, *rid);
if (rle == NULL)
return (pci_alloc_map(dev, child, type, rid,
start, end, count, flags));
When the driver allocates memory or I/O ports, enable that bit in the command config register. At the present, this represents a nop because these bits should have been set earlier in the process. In the future, we'll only set these bits when the driver requests the resource, not when the bus code detects the resource. Reviewed by: mdodd
2003-04-15 19:38:18 +00:00
break;
When a pci device hasn't had an interrupt routed to it (signified by the intline of 255) go ahead and route the interrupt when we allocate an interrupt. Submitted by: msmith
2000-10-16 07:24:00 +00:00
}
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
/*
* If we've already allocated the resource, then
* return it now. But first we may need to activate
* it, since we don't allocate the resource as active
* above. Normally this would be done down in the
* nexus, but since we short-circuit that path we have
* to do its job here. Not sure if we should free the
* resource if it fails to activate.
*/
rle = resource_list_find(rl, type, *rid);
if (rle != NULL && rle->res != NULL) {
make the pci power state and resource code a lot less chatty. The chattiness was left in for debugging, but now that nearly all of the problems relating to the changes have been fixed, it is only annoying. It is still available via bootverbose. Prodded by: jhb
2004-05-21 06:03:26 +00:00
if (bootverbose)
device_printf(child,
Some devices have what appear to be invalid BARs. They are invalid in the sense that any write to them reads back as a 0. This presents a problem to our resource allocation scheme. If we encounter such vars, the code now treats them as special, allowing any allocation against them to succeed. I've not seen anything in the standard to clearify what host software should do when it encounters these sorts of BARs. Also cleaned up some output while I'm here and add commmented out bootverbose lines until I'm ready to reduce the verbosity of boot messages. This gets a number of south bridges and ata controllers made mostly by VIA, AMD and nVidia working again. Thanks to Soren Schmidt for his help in coming up with this patch.
2004-04-13 19:31:57 +00:00
"Reserved %#lx bytes for rid %#x type %d at %#lx\n",
make the pci power state and resource code a lot less chatty. The chattiness was left in for debugging, but now that nearly all of the problems relating to the changes have been fixed, it is only annoying. It is still available via bootverbose. Prodded by: jhb
2004-05-21 06:03:26 +00:00
rman_get_size(rle->res), *rid, type,
rman_get_start(rle->res));
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
if ((flags & RF_ACTIVE) &&
bus_generic_activate_resource(dev, child, type,
*rid, rle->res) != 0)
return NULL;
return (rle->res);
}
When a pci device hasn't had an interrupt routed to it (signified by the intline of 255) go ahead and route the interrupt when we allocate an interrupt. Submitted by: msmith
2000-10-16 07:24:00 +00:00
}
More style(9) nits
2002-06-01 05:44:45 +00:00
return (resource_list_alloc(rl, dev, child, type, rid,
start, end, count, flags));
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
}
Use the pci.c code wherever possible, rather than copying all the pci code into cardbus and s/pci/cardbus. This exposes a few pci_* functions that are now static. This work is similar to work Justin posted to the mobile list about a year or two ago, which I have neglected since then. This is a subset of his current work with the multiple inheritance newbus architecutre. When completed, that will eliminate the need for pci/pci_private.h. Similar work is needed for the cardbus_cis and pccard_cis code as well.
2002-02-27 05:09:14 +00:00
void
Reduce code duplication by using the GET_RESOURCE_LIST bus method and related generic resource_list management functions. I'll deal with the EISA bits later. Not objected to by: new-bus
2000-11-28 07:12:12 +00:00
pci_delete_resource(device_t dev, device_t child, int type, int rid)
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
{
Use the pci.c code wherever possible, rather than copying all the pci code into cardbus and s/pci/cardbus. This exposes a few pci_* functions that are now static. This work is similar to work Justin posted to the mobile list about a year or two ago, which I have neglected since then. This is a subset of his current work with the multiple inheritance newbus architecutre. When completed, that will eliminate the need for pci/pci_private.h. Similar work is needed for the cardbus_cis and pccard_cis code as well.
2002-02-27 05:09:14 +00:00
struct pci_devinfo *dinfo;
struct resource_list *rl;
struct resource_list_entry *rle;
if (device_get_parent(child) != dev)
return;
dinfo = device_get_ivars(child);
rl = &dinfo->resources;
rle = resource_list_find(rl, type, rid);
if (rle) {
if (rle->res) {
Use rman_get_device rather than rle->resl->r_dev. make pci_hdrtypedata and pci_read_extcap accessible (but maybe in the end we'll make them private again).
2003-02-16 02:02:44 +00:00
if (rman_get_device(rle->res) != dev ||
Use the pci.c code wherever possible, rather than copying all the pci code into cardbus and s/pci/cardbus. This exposes a few pci_* functions that are now static. This work is similar to work Justin posted to the mobile list about a year or two ago, which I have neglected since then. This is a subset of his current work with the multiple inheritance newbus architecutre. When completed, that will eliminate the need for pci/pci_private.h. Similar work is needed for the cardbus_cis and pccard_cis code as well.
2002-02-27 05:09:14 +00:00
rman_get_flags(rle->res) & RF_ACTIVE) {
device_printf(dev, "delete_resource: "
"Resource still owned by child, oops. "
"(type=%d, rid=%d, addr=%lx)\n",
rle->type, rle->rid,
rman_get_start(rle->res));
return;
}
bus_release_resource(dev, type, rid, rle->res);
}
resource_list_delete(rl, type, rid);
}
More style(9) nits
2002-06-01 05:44:45 +00:00
/*
* Why do we turn off the PCI configuration BAR when we delete a
* resource? -- imp
*/
Use the pci.c code wherever possible, rather than copying all the pci code into cardbus and s/pci/cardbus. This exposes a few pci_* functions that are now static. This work is similar to work Justin posted to the mobile list about a year or two ago, which I have neglected since then. This is a subset of his current work with the multiple inheritance newbus architecutre. When completed, that will eliminate the need for pci/pci_private.h. Similar work is needed for the cardbus_cis and pccard_cis code as well.
2002-02-27 05:09:14 +00:00
pci_write_config(child, rid, 0, 4);
BUS_DELETE_RESOURCE(device_get_parent(dev), child, type, rid);
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
}
Use the pci.c code wherever possible, rather than copying all the pci code into cardbus and s/pci/cardbus. This exposes a few pci_* functions that are now static. This work is similar to work Justin posted to the mobile list about a year or two ago, which I have neglected since then. This is a subset of his current work with the multiple inheritance newbus architecutre. When completed, that will eliminate the need for pci/pci_private.h. Similar work is needed for the cardbus_cis and pccard_cis code as well.
2002-02-27 05:09:14 +00:00
struct resource_list *
Reduce code duplication by using the GET_RESOURCE_LIST bus method and related generic resource_list management functions. I'll deal with the EISA bits later. Not objected to by: new-bus
2000-11-28 07:12:12 +00:00
pci_get_resource_list (device_t dev, device_t child)
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
{
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
struct pci_devinfo *dinfo = device_get_ivars(child);
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
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
return (&dinfo->resources);
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
}
Prefer the uintXX_t to the u_intXX_t names.
2003-08-22 03:11:53 +00:00
uint32_t
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
pci_read_config_method(device_t dev, device_t child, int reg, int width)
{
struct pci_devinfo *dinfo = device_get_ivars(child);
pcicfgregs *cfg = &dinfo->cfg;
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
More style(9) nits
2002-06-01 05:44:45 +00:00
return (PCIB_READ_CONFIG(device_get_parent(dev),
cfg->bus, cfg->slot, cfg->func, reg, width));
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
}
Use the pci.c code wherever possible, rather than copying all the pci code into cardbus and s/pci/cardbus. This exposes a few pci_* functions that are now static. This work is similar to work Justin posted to the mobile list about a year or two ago, which I have neglected since then. This is a subset of his current work with the multiple inheritance newbus architecutre. When completed, that will eliminate the need for pci/pci_private.h. Similar work is needed for the cardbus_cis and pccard_cis code as well.
2002-02-27 05:09:14 +00:00
void
More style(9) nits
2002-06-01 05:44:45 +00:00
pci_write_config_method(device_t dev, device_t child, int reg,
Prefer the uintXX_t to the u_intXX_t names.
2003-08-22 03:11:53 +00:00
uint32_t val, int width)
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
{
struct pci_devinfo *dinfo = device_get_ivars(child);
pcicfgregs *cfg = &dinfo->cfg;
* Completely rewrite the alpha busspace to hide the implementation from the drivers. * Remove legacy inx/outx support from chipset and replace with macros which call busspace. * Rework pci config accesses to route through the pcib device instead of calling a MD function directly. With these changes it is possible to cleanly support machines which have more than one independantly numbered PCI busses. As a bonus, the new busspace implementation should be measurably faster than the old one.
2000-08-28 21:48:13 +00:00
PCIB_WRITE_CONFIG(device_get_parent(dev),
Make this file mostly conform to style(9). Approved by: msmith in principle before walkabout
2002-06-01 03:41:02 +00:00
cfg->bus, cfg->slot, cfg->func, reg, val, width);
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
}
Move the pnp and location info into the common pci bus. Make all known pci busses implement this. Also minor comment smithing in cardbus. Fix copyright to this year with my name on it since I've been doing a lot to this file. Reviewed by: jhb
2003-02-17 21:20:35 +00:00
int
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
pci_child_location_str_method(device_t dev, device_t child, char *buf,
Move the pnp and location info into the common pci bus. Make all known pci busses implement this. Also minor comment smithing in cardbus. Fix copyright to this year with my name on it since I've been doing a lot to this file. Reviewed by: jhb
2003-02-17 21:20:35 +00:00
size_t buflen)
{
snprintf(buf, buflen, "slot=%d function=%d", pci_get_slot(child),
pci_get_function(child));
return (0);
}
int
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
pci_child_pnpinfo_str_method(device_t dev, device_t child, char *buf,
Move the pnp and location info into the common pci bus. Make all known pci busses implement this. Also minor comment smithing in cardbus. Fix copyright to this year with my name on it since I've been doing a lot to this file. Reviewed by: jhb
2003-02-17 21:20:35 +00:00
size_t buflen)
{
struct pci_devinfo *dinfo;
pcicfgregs *cfg;
dinfo = device_get_ivars(child);
cfg = &dinfo->cfg;
snprintf(buf, buflen, "vendor=0x%04x device=0x%04x subvendor=0x%04x "
Include class designation in pnpinfo for generic driver loading
2003-02-18 03:25:57 +00:00
"subdevice=0x%04x class=0x%02x%02x%02x", cfg->vendor, cfg->device,
cfg->subvendor, cfg->subdevice, cfg->baseclass, cfg->subclass,
cfg->progif);
Move the pnp and location info into the common pci bus. Make all known pci busses implement this. Also minor comment smithing in cardbus. Fix copyright to this year with my name on it since I've been doing a lot to this file. Reviewed by: jhb
2003-02-17 21:20:35 +00:00
return (0);
}
Add a new PCI interface method, assign_interrupt, to determine the interrupt to be used for a device. This is intended solely for internal use of PCI bus implementations, and exists so that PCI bus drivers implementing special interrupt assignment methods which require additional work at the bus level to work right can be easily derived from the generic driver (or any other one) without resorting to hacks. It will be used in the sparc64 ofw_pcibus driver, which will be committed shortly. Make use of this method in the generic implementation, and add it to the method table of bus drivers derived from the PCI one. Reviewed by: imp, -hackers
2003-07-01 14:08:33 +00:00
int
pci_assign_interrupt_method(device_t dev, device_t child)
{
struct pci_devinfo *dinfo = device_get_ivars(child);
pcicfgregs *cfg = &dinfo->cfg;
return (PCIB_ROUTE_INTERRUPT(device_get_parent(dev), child,
cfg->intpin));
}
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
static int
pci_modevent(module_t mod, int what, void *arg)
{
Do the dreaded s/dev_t/struct cdev */ Bump __FreeBSD_version accordingly.
2004-06-16 09:47:26 +00:00
static struct cdev *pci_cdev;
- Make pci_load_vendor_data() static and do it during MOD_LOAD instead of when the first PCI bus attaches. - Create /dev/pci during MOD_LOAD as well. - Destroy /dev/pci during MOD_UNLOAD (not that you can kldunload pci, but might as well get the code right)
2002-09-04 03:13:16 +00:00
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
switch (what) {
case MOD_LOAD:
Argh, don't clobber the pci device list if there are multiple busses! (An AGP counts as a PCI bus, it seems...) This stopped 'pciconf -l' from working on AGP or PCI->PCI bridge systems.
1999-05-09 15:54:04 +00:00
STAILQ_INIT(&pci_devq);
Next round of PCI subsystem updates: - Break out the /dev/pci driver into a separate file. - Kill the COMPAT_OLDPCI support. - Make the EISA bridge attach a bit more like the old code; explicitly check for the existence of eisa0/isa0 and only attach if they don't already exist. Only make one bus_generic_attach() pass over the bridge, once both busses are attached. Note that the stupid Intel bridge's class is entirely unpredictable. - Add prototypes and re-layout the core PCI modules in line with current coding standards (not a major whitespace change, just moving the module data to the top of the file). - Remove redundant type-2 bridge support from the core PCI code; the PCI-CardBus code does this itself internally. Remove the now entirely redundant header-class-specific support, as well as the secondary and subordinate bus number fields. These are bridge attributes now. - Add support for PCI Extended Capabilities. - Add support for PCI Power Management. The interface currently allows a driver to query and set the power state of a device. - Add helper functions to allow drivers to enable/disable busmastering and the decoding of I/O and memory ranges. - Use PCI_SLOTMAX and PCI_FUNCMAX rather than magic numbers in some places. - Make the PCI-PCI bridge code a little more paranoid about valid I/O and memory decodes. - Add some more PCI register definitions for the command and status registers. Correct another bogus definition for type-1 bridges.
2000-12-13 01:25:11 +00:00
pci_generation = 0;
- Make pci_load_vendor_data() static and do it during MOD_LOAD instead of when the first PCI bus attaches. - Create /dev/pci during MOD_LOAD as well. - Destroy /dev/pci during MOD_UNLOAD (not that you can kldunload pci, but might as well get the code right)
2002-09-04 03:13:16 +00:00
pci_cdev = make_dev(&pcicdev, 0, UID_ROOT, GID_WHEEL, 0644,
"pci");
pci_load_vendor_data();
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
break;
case MOD_UNLOAD:
- Make pci_load_vendor_data() static and do it during MOD_LOAD instead of when the first PCI bus attaches. - Create /dev/pci during MOD_LOAD as well. - Destroy /dev/pci during MOD_UNLOAD (not that you can kldunload pci, but might as well get the code right)
2002-09-04 03:13:16 +00:00
destroy_dev(pci_cdev);
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
break;
}
More style(9) nits
2002-06-01 05:44:45 +00:00
return (0);
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
}
Add pci_resume() to reestablish interrupt routing after suspend/resume. Especially after hibernation, interrupt routing went back to initial status on some machines.
2003-09-17 08:32:44 +00:00
Expose pci_cfg_safe/restore for subclasses of pci to use.
2005-02-28 01:14:15 +00:00
void
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
pci_cfg_restore(device_t dev, struct pci_devinfo *dinfo)
Add pci_resume() to reestablish interrupt routing after suspend/resume. Especially after hibernation, interrupt routing went back to initial status on some machines.
2003-09-17 08:32:44 +00:00
{
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
int i;
Add pci_resume() to reestablish interrupt routing after suspend/resume. Especially after hibernation, interrupt routing went back to initial status on some machines.
2003-09-17 08:32:44 +00:00
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
/*
Add system tunable to turn off power state changes. Default to off until we get the resource allocation stuff hammered out. Fix and off by one error that caused unnecessary filtering of valid BARs for only 4 bytes than ICH3 and other PCI IDE controllers have. Andrew Gallatin submitted this, although it doesn't solve the problems ICH3 controllers have with the new code, it does restore the former resource list on the probe line.
2004-04-11 07:02:49 +00:00
* Only do header type 0 devices. Type 1 devices are bridges,
* which we know need special treatment. Type 2 devices are
* cardbus bridges which also require special treatment.
* Other types are unknown, and we err on the side of safety
* by ignoring them.
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
*/
if (dinfo->cfg.hdrtype != 0)
return;
Move pci_do_powerstate up a level. Now it just means 'do not turn devices off into d3 state when there's no driver for the device'. This should help suspend/resume in the default case.
2004-05-21 06:39:09 +00:00
Add system tunable to turn off power state changes. Default to off until we get the resource allocation stuff hammered out. Fix and off by one error that caused unnecessary filtering of valid BARs for only 4 bytes than ICH3 and other PCI IDE controllers have. Andrew Gallatin submitted this, although it doesn't solve the problems ICH3 controllers have with the new code, it does restore the former resource list on the probe line.
2004-04-11 07:02:49 +00:00
/*
* Restore the device to full power mode. We must do this
* before we restore the registers because moving from D3 to
* D0 will cause the chip's BARs and some other registers to
* be reset to some unknown power on reset values. Cut down
* the noise on boot by doing nothing if we are already in
* state D0.
*/
Move pci_do_powerstate up a level. Now it just means 'do not turn devices off into d3 state when there's no driver for the device'. This should help suspend/resume in the default case.
2004-05-21 06:39:09 +00:00
if (pci_get_powerstate(dev) != PCI_POWERSTATE_D0) {
Only print state change message for real state changes. When we set a device in D0 to D0, that's a no-op, however the messages seem to be confusing some people. Eventually, these messages will be parked behind a if (bootverbose). # I don't think this will fix any real bugs...
2004-04-09 20:41:18 +00:00
pci_set_powerstate(dev, PCI_POWERSTATE_D0);
}
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
for (i = 0; i < dinfo->cfg.nummaps; i++)
Use PCI_BAR() in preference to PCI_MAPS + x * 4. Submitted by: jhb
2004-05-24 17:41:05 +00:00
pci_write_config(dev, PCIR_BAR(i), dinfo->cfg.bar[i], 4);
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
pci_write_config(dev, PCIR_BIOS, dinfo->cfg.bios, 4);
pci_write_config(dev, PCIR_COMMAND, dinfo->cfg.cmdreg, 2);
pci_write_config(dev, PCIR_INTLINE, dinfo->cfg.intline, 1);
pci_write_config(dev, PCIR_INTPIN, dinfo->cfg.intpin, 1);
pci_write_config(dev, PCIR_MINGNT, dinfo->cfg.mingnt, 1);
pci_write_config(dev, PCIR_MAXLAT, dinfo->cfg.maxlat, 1);
pci_write_config(dev, PCIR_CACHELNSZ, dinfo->cfg.cachelnsz, 1);
pci_write_config(dev, PCIR_LATTIMER, dinfo->cfg.lattimer, 1);
MFp4: o save/restore subvendor, subdevice, vendor, device, baseclass, subclass, progif and revid. While these are typically read only fields, they aren't always read-only. progif is writable for ata devices, for example. It does no harm when they are read only, and helps when they aren't.
2004-05-21 06:36:36 +00:00
pci_write_config(dev, PCIR_PROGIF, dinfo->cfg.progif, 1);
pci_write_config(dev, PCIR_REVID, dinfo->cfg.revid, 1);
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
}
Add pci_resume() to reestablish interrupt routing after suspend/resume. Especially after hibernation, interrupt routing went back to initial status on some machines.
2003-09-17 08:32:44 +00:00
Expose pci_cfg_safe/restore for subclasses of pci to use.
2005-02-28 01:14:15 +00:00
void
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
pci_cfg_save(device_t dev, struct pci_devinfo *dinfo, int setstate)
{
int i;
uint32_t cls;
Add system tunable to turn off power state changes. Default to off until we get the resource allocation stuff hammered out. Fix and off by one error that caused unnecessary filtering of valid BARs for only 4 bytes than ICH3 and other PCI IDE controllers have. Andrew Gallatin submitted this, although it doesn't solve the problems ICH3 controllers have with the new code, it does restore the former resource list on the probe line.
2004-04-11 07:02:49 +00:00
int ps;
Add pci_resume() to reestablish interrupt routing after suspend/resume. Especially after hibernation, interrupt routing went back to initial status on some machines.
2003-09-17 08:32:44 +00:00
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
/*
* Only do header type 0 devices. Type 1 devices are bridges, which
* we know need special treatment. Type 2 devices are cardbus bridges
* which also require special treatment. Other types are unknown, and
* we err on the side of safety by ignoring them. Powering down
* bridges should not be undertaken lightly.
*/
if (dinfo->cfg.hdrtype != 0)
return;
for (i = 0; i < dinfo->cfg.nummaps; i++)
Use PCI_BAR() in preference to PCI_MAPS + x * 4. Submitted by: jhb
2004-05-24 17:41:05 +00:00
dinfo->cfg.bar[i] = pci_read_config(dev, PCIR_BAR(i), 4);
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
dinfo->cfg.bios = pci_read_config(dev, PCIR_BIOS, 4);
Add pci_resume() to reestablish interrupt routing after suspend/resume. Especially after hibernation, interrupt routing went back to initial status on some machines.
2003-09-17 08:32:44 +00:00
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
/*
Fix comments to match last commit, and minor reformatting...
2004-12-08 04:35:19 +00:00
* Some drivers apparently write to these registers w/o updating our
Fix typo.
2005-01-29 19:45:31 +00:00
* cached copy. No harm happens if we update the copy, so do so here
Fix comments to match last commit, and minor reformatting...
2004-12-08 04:35:19 +00:00
* so we can restore them. The COMMAND register is modified by the
* bus w/o updating the cache. This should represent the normally
* writable portion of the 'defined' part of type 0 headers. In
* theory we also need to save/restore the PCI capability structures
* we know about, but apart from power we don't know any that are
* writable.
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
*/
MFp4: o save/restore subvendor, subdevice, vendor, device, baseclass, subclass, progif and revid. While these are typically read only fields, they aren't always read-only. progif is writable for ata devices, for example. It does no harm when they are read only, and helps when they aren't.
2004-05-21 06:36:36 +00:00
dinfo->cfg.subvendor = pci_read_config(dev, PCIR_SUBVEND_0, 2);
dinfo->cfg.subdevice = pci_read_config(dev, PCIR_SUBDEV_0, 2);
dinfo->cfg.vendor = pci_read_config(dev, PCIR_VENDOR, 2);
dinfo->cfg.device = pci_read_config(dev, PCIR_DEVICE, 2);
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
dinfo->cfg.cmdreg = pci_read_config(dev, PCIR_COMMAND, 2);
dinfo->cfg.intline = pci_read_config(dev, PCIR_INTLINE, 1);
dinfo->cfg.intpin = pci_read_config(dev, PCIR_INTPIN, 1);
dinfo->cfg.mingnt = pci_read_config(dev, PCIR_MINGNT, 1);
dinfo->cfg.maxlat = pci_read_config(dev, PCIR_MAXLAT, 1);
dinfo->cfg.cachelnsz = pci_read_config(dev, PCIR_CACHELNSZ, 1);
dinfo->cfg.lattimer = pci_read_config(dev, PCIR_LATTIMER, 1);
MFp4: o save/restore subvendor, subdevice, vendor, device, baseclass, subclass, progif and revid. While these are typically read only fields, they aren't always read-only. progif is writable for ata devices, for example. It does no harm when they are read only, and helps when they aren't.
2004-05-21 06:36:36 +00:00
dinfo->cfg.baseclass = pci_read_config(dev, PCIR_CLASS, 1);
dinfo->cfg.subclass = pci_read_config(dev, PCIR_SUBCLASS, 1);
dinfo->cfg.progif = pci_read_config(dev, PCIR_PROGIF, 1);
dinfo->cfg.revid = pci_read_config(dev, PCIR_REVID, 1);
Add pci_resume() to reestablish interrupt routing after suspend/resume. Especially after hibernation, interrupt routing went back to initial status on some machines.
2003-09-17 08:32:44 +00:00
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
/*
Fix comments to match last commit, and minor reformatting...
2004-12-08 04:35:19 +00:00
* don't set the state for display devices, base peripherals and
* memory devices since bad things happen when they are powered down.
* We should (a) have drivers that can easily detach and (b) use
* generic drivers for these devices so that some device actually
* attaches. We need to make sure that when we implement (a) we don't
* power the device down on a reattach.
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
*/
cls = pci_get_class(dev);
Change hw.pci.do_powerstate from a boolean to a range. 0 means the same as today: do no power management. 1 means be conservative about what you power down (any device class that has caused problems gets added here). 2 means be agressive about what gets powered down (any device class that's fundamental to the system is here). 3 means power them all down, reguardless. The default is 1. The effect in the default system is to add mass storage devices to the list that we don't power down. From all the pciconf -l lists that I've seen for the aac and amr issue, the bad device has been a mass storage device class. This is an attempt at a compromise between the very small number of systems that have extreme issues with powerdown, and the very large number of systems that gain real benefits from powerdown (I get about 20% more battery life when I attach a minimal set of drivers on my Sony). Hopefully it will strike the proper balance. MFC After: 3 days (before next beta)
2005-09-11 04:09:44 +00:00
if (!setstate)
return;
Split power state control into two variables. hw.pci.do_powerstate has been removed. It has been replaced by hw.pci.do_power_nodriver and hw.pci.do_power_resume. The former defaults to 0 while the latter defaults to 1. When do_powerstate was set to 0, it broke suspend/resume for a lot of people as an unintended consequence. This change will only affect the areas that were intended to affect. This change will have no effect on servers, but will help laptops quite a bit. MFC After: 3 days.
2005-09-21 19:47:00 +00:00
switch (pci_do_power_nodriver)
Change hw.pci.do_powerstate from a boolean to a range. 0 means the same as today: do no power management. 1 means be conservative about what you power down (any device class that has caused problems gets added here). 2 means be agressive about what gets powered down (any device class that's fundamental to the system is here). 3 means power them all down, reguardless. The default is 1. The effect in the default system is to add mass storage devices to the list that we don't power down. From all the pciconf -l lists that I've seen for the aac and amr issue, the bad device has been a mass storage device class. This is an attempt at a compromise between the very small number of systems that have extreme issues with powerdown, and the very large number of systems that gain real benefits from powerdown (I get about 20% more battery life when I attach a minimal set of drivers on my Sony). Hopefully it will strike the proper balance. MFC After: 3 days (before next beta)
2005-09-11 04:09:44 +00:00
{
case 0: /* NO powerdown at all */
return;
case 1: /* Conservative about what to power down */
if (cls == PCIC_STORAGE)
return;
/*FALLTHROUGH*/
case 2: /* Agressive about what to power down */
if (cls == PCIC_DISPLAY || cls == PCIC_MEMORY ||
cls == PCIC_BASEPERIPH)
return;
/*FALLTHROUGH*/
case 3: /* Power down everything */
break;
Omnibus PCI commit: o Save and restore bars for suspend/resume as well as for D3->D0 transitions. o preallocate resources that the PCI devices use to avoid resource conflicts o lazy allocation of resources not allocated by the BIOS. o set unattached drivers to state D3. Set power state to D0 before probe/attach. Right now there's two special cases for this (display and memory devices) that need work in other areas of the tree. Please report any bugs to me.
2004-04-09 15:44:34 +00:00
}
Change hw.pci.do_powerstate from a boolean to a range. 0 means the same as today: do no power management. 1 means be conservative about what you power down (any device class that has caused problems gets added here). 2 means be agressive about what gets powered down (any device class that's fundamental to the system is here). 3 means power them all down, reguardless. The default is 1. The effect in the default system is to add mass storage devices to the list that we don't power down. From all the pciconf -l lists that I've seen for the aac and amr issue, the bad device has been a mass storage device class. This is an attempt at a compromise between the very small number of systems that have extreme issues with powerdown, and the very large number of systems that gain real benefits from powerdown (I get about 20% more battery life when I attach a minimal set of drivers on my Sony). Hopefully it will strike the proper balance. MFC After: 3 days (before next beta)
2005-09-11 04:09:44 +00:00
/*
* PCI spec says we can only go into D3 state from D0 state.
* Transition from D[12] into D0 before going to D3 state.
*/
ps = pci_get_powerstate(dev);
if (ps != PCI_POWERSTATE_D0 && ps != PCI_POWERSTATE_D3)
pci_set_powerstate(dev, PCI_POWERSTATE_D0);
if (pci_get_powerstate(dev) != PCI_POWERSTATE_D3)
pci_set_powerstate(dev, PCI_POWERSTATE_D3);
Add pci_resume() to reestablish interrupt routing after suspend/resume. Especially after hibernation, interrupt routing went back to initial status on some machines.
2003-09-17 08:32:44 +00:00
}
Reference in New Issue Copy Permalink