License cleanup: add SPDX GPL-2.0 license identifier to files with no license
Many source files in the tree are missing licensing information, which
makes it harder for compliance tools to determine the correct license.
By default all files without license information are under the default
license of the kernel, which is GPL version 2.
Update the files which contain no license information with the 'GPL-2.0'
SPDX license identifier. The SPDX identifier is a legally binding
shorthand, which can be used instead of the full boiler plate text.
This patch is based on work done by Thomas Gleixner and Kate Stewart and
Philippe Ombredanne.
How this work was done:
Patches were generated and checked against linux-4.14-rc6 for a subset of
the use cases:
- file had no licensing information it it.
- file was a */uapi/* one with no licensing information in it,
- file was a */uapi/* one with existing licensing information,
Further patches will be generated in subsequent months to fix up cases
where non-standard license headers were used, and references to license
had to be inferred by heuristics based on keywords.
The analysis to determine which SPDX License Identifier to be applied to
a file was done in a spreadsheet of side by side results from of the
output of two independent scanners (ScanCode & Windriver) producing SPDX
tag:value files created by Philippe Ombredanne. Philippe prepared the
base worksheet, and did an initial spot review of a few 1000 files.
The 4.13 kernel was the starting point of the analysis with 60,537 files
assessed. Kate Stewart did a file by file comparison of the scanner
results in the spreadsheet to determine which SPDX license identifier(s)
to be applied to the file. She confirmed any determination that was not
immediately clear with lawyers working with the Linux Foundation.
Criteria used to select files for SPDX license identifier tagging was:
- Files considered eligible had to be source code files.
- Make and config files were included as candidates if they contained >5
lines of source
- File already had some variant of a license header in it (even if <5
lines).
All documentation files were explicitly excluded.
The following heuristics were used to determine which SPDX license
identifiers to apply.
- when both scanners couldn't find any license traces, file was
considered to have no license information in it, and the top level
COPYING file license applied.
For non */uapi/* files that summary was:
SPDX license identifier # files
---------------------------------------------------|-------
GPL-2.0 11139
and resulted in the first patch in this series.
If that file was a */uapi/* path one, it was "GPL-2.0 WITH
Linux-syscall-note" otherwise it was "GPL-2.0". Results of that was:
SPDX license identifier # files
---------------------------------------------------|-------
GPL-2.0 WITH Linux-syscall-note 930
and resulted in the second patch in this series.
- if a file had some form of licensing information in it, and was one
of the */uapi/* ones, it was denoted with the Linux-syscall-note if
any GPL family license was found in the file or had no licensing in
it (per prior point). Results summary:
SPDX license identifier # files
---------------------------------------------------|------
GPL-2.0 WITH Linux-syscall-note 270
GPL-2.0+ WITH Linux-syscall-note 169
((GPL-2.0 WITH Linux-syscall-note) OR BSD-2-Clause) 21
((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause) 17
LGPL-2.1+ WITH Linux-syscall-note 15
GPL-1.0+ WITH Linux-syscall-note 14
((GPL-2.0+ WITH Linux-syscall-note) OR BSD-3-Clause) 5
LGPL-2.0+ WITH Linux-syscall-note 4
LGPL-2.1 WITH Linux-syscall-note 3
((GPL-2.0 WITH Linux-syscall-note) OR MIT) 3
((GPL-2.0 WITH Linux-syscall-note) AND MIT) 1
and that resulted in the third patch in this series.
- when the two scanners agreed on the detected license(s), that became
the concluded license(s).
- when there was disagreement between the two scanners (one detected a
license but the other didn't, or they both detected different
licenses) a manual inspection of the file occurred.
- In most cases a manual inspection of the information in the file
resulted in a clear resolution of the license that should apply (and
which scanner probably needed to revisit its heuristics).
- When it was not immediately clear, the license identifier was
confirmed with lawyers working with the Linux Foundation.
- If there was any question as to the appropriate license identifier,
the file was flagged for further research and to be revisited later
in time.
In total, over 70 hours of logged manual review was done on the
spreadsheet to determine the SPDX license identifiers to apply to the
source files by Kate, Philippe, Thomas and, in some cases, confirmation
by lawyers working with the Linux Foundation.
Kate also obtained a third independent scan of the 4.13 code base from
FOSSology, and compared selected files where the other two scanners
disagreed against that SPDX file, to see if there was new insights. The
Windriver scanner is based on an older version of FOSSology in part, so
they are related.
Thomas did random spot checks in about 500 files from the spreadsheets
for the uapi headers and agreed with SPDX license identifier in the
files he inspected. For the non-uapi files Thomas did random spot checks
in about 15000 files.
In initial set of patches against 4.14-rc6, 3 files were found to have
copy/paste license identifier errors, and have been fixed to reflect the
correct identifier.
Additionally Philippe spent 10 hours this week doing a detailed manual
inspection and review of the 12,461 patched files from the initial patch
version early this week with:
- a full scancode scan run, collecting the matched texts, detected
license ids and scores
- reviewing anything where there was a license detected (about 500+
files) to ensure that the applied SPDX license was correct
- reviewing anything where there was no detection but the patch license
was not GPL-2.0 WITH Linux-syscall-note to ensure that the applied
SPDX license was correct
This produced a worksheet with 20 files needing minor correction. This
worksheet was then exported into 3 different .csv files for the
different types of files to be modified.
These .csv files were then reviewed by Greg. Thomas wrote a script to
parse the csv files and add the proper SPDX tag to the file, in the
format that the file expected. This script was further refined by Greg
based on the output to detect more types of files automatically and to
distinguish between header and source .c files (which need different
comment types.) Finally Greg ran the script using the .csv files to
generate the patches.
Reviewed-by: Kate Stewart <kstewart@linuxfoundation.org>
Reviewed-by: Philippe Ombredanne <pombredanne@nexb.com>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2017-11-01 17:07:57 +03:00
// SPDX-License-Identifier: GPL-2.0
2005-04-17 02:20:36 +04:00
/*
2018-03-10 01:36:33 +03:00
* This file contains work - arounds for many known PCI hardware bugs .
* Devices present only on certain architectures ( host bridges et cetera )
* should be handled in arch - specific code .
2005-04-17 02:20:36 +04:00
*
2018-03-10 01:36:33 +03:00
* Note : any quirks for hotpluggable devices must _NOT_ be declared __init .
2005-04-17 02:20:36 +04:00
*
2018-03-10 01:36:33 +03:00
* Copyright ( c ) 1999 Martin Mares < mj @ ucw . cz >
2005-04-17 02:20:36 +04:00
*
2018-03-10 01:36:33 +03:00
* Init / reset quirks for USB host controllers should be in the USB quirks
* file , where their drivers can use them .
2005-04-17 02:20:36 +04:00
*/
# include <linux/types.h>
# include <linux/kernel.h>
2011-05-27 17:37:25 +04:00
# include <linux/export.h>
2005-04-17 02:20:36 +04:00
# include <linux/pci.h>
# include <linux/init.h>
# include <linux/delay.h>
2005-05-27 12:21:50 +04:00
# include <linux/acpi.h>
2008-11-21 07:42:25 +03:00
# include <linux/dmi.h>
2009-03-16 11:13:39 +03:00
# include <linux/ioport.h>
2012-01-31 08:52:07 +04:00
# include <linux/sched.h>
# include <linux/ktime.h>
2014-08-21 03:26:52 +04:00
# include <linux/mm.h>
2018-08-09 23:18:33 +03:00
# include <linux/nvme.h>
2017-08-01 15:10:41 +03:00
# include <linux/platform_data/x86/apple.h>
vga_switcheroo: Use device link for HDA controller
Back in 2013, runtime PM for GPUs with integrated HDA controller was
introduced with commits 0d69704ae348 ("gpu/vga_switcheroo: add driver
control power feature. (v3)") and 246efa4a072f ("snd/hda: add runtime
suspend/resume on optimus support (v4)").
Briefly, the idea was that the HDA controller is forced on and off in
unison with the GPU.
The original code is mostly still in place even though it was never a
100% perfect solution: E.g. on access to the HDA controller, the GPU
is powered up via vga_switcheroo_runtime_resume_hdmi_audio() but there
are no provisions to keep it resumed until access to the HDA controller
has ceased: The GPU autosuspends after 5 seconds, rendering the HDA
controller inaccessible.
Additionally, a kludge is required when hda_intel.c probes: It has to
check whether the GPU is powered down (check_hdmi_disabled()) and defer
probing if so.
However in the meantime (in v4.10) the driver core has gained a feature
called device links which promises to solve such issues in a clean way:
It allows us to declare a dependency from the HDA controller (consumer)
to the GPU (supplier). The PM core then automagically ensures that the
GPU is runtime resumed as long as the HDA controller's ->probe hook is
executed and whenever the HDA controller is accessed.
By default, the HDA controller has a dependency on its parent, a PCIe
Root Port. Adding a device link creates another dependency on its
sibling:
PCIe Root Port
^ ^
| |
| |
HDA ===> GPU
The device link is not only used for runtime PM, it also guarantees that
on system sleep, the HDA controller suspends before the GPU and resumes
after the GPU, and on system shutdown the HDA controller's ->shutdown
hook is executed before the one of the GPU. It is a complete solution.
Using this functionality is as simple as calling device_link_add(),
which results in a dmesg entry like this:
pci 0000:01:00.1: Linked as a consumer to 0000:01:00.0
The code for the GPU-governed audio power management can thus be removed
(except where it's still needed for legacy manual power control).
The device link is added in a PCI quirk rather than in hda_intel.c.
It is therefore legal for the GPU to runtime suspend to D3cold even if
the HDA controller is not bound to a driver or if CONFIG_SND_HDA_INTEL
is not enabled, for accesses to the HDA controller will cause the GPU to
wake up regardless if they're occurring outside of hda_intel.c (think
config space readout via sysfs).
Contrary to the previous implementation, the HDA controller's power
state is now self-governed, rather than GPU-governed, whereas the GPU's
power state is no longer fully self-governed. (The HDA controller needs
to runtime suspend before the GPU can.)
It is thus crucial that runtime PM is always activated on the HDA
controller even if CONFIG_SND_HDA_POWER_SAVE_DEFAULT is set to 0 (which
is the default), lest the GPU stays awake. This is achieved by setting
the auto_runtime_pm flag on every codec and the AZX_DCAPS_PM_RUNTIME
flag on the HDA controller.
A side effect is that power consumption might be reduced if the GPU is
in use but the HDA controller is not, because the HDA controller is now
allowed to go to D3hot. Before, it was forced to stay in D0 as long as
the GPU was in use. (There is no reduction in power consumption on my
Nvidia GK107, but there might be on other chips.)
The code paths for legacy manual power control are adjusted such that
runtime PM is disabled during power off, thereby preventing the PM core
from resuming the HDA controller.
Note that the device link is not only added on vga_switcheroo capable
systems, but for *any* GPU with integrated HDA controller. The idea is
that the HDA controller streams audio via connectors located on the GPU,
so the GPU needs to be on for the HDA controller to do anything useful.
This commit implicitly fixes an unbalanced runtime PM ref upon unbind of
hda_intel.c: On ->probe, a runtime PM ref was previously released under
the condition "azx_has_pm_runtime(chip) || hda->use_vga_switcheroo", but
on ->remove a runtime PM ref was only acquired under the first of those
conditions. Thus, binding and unbinding the driver twice on a
vga_switcheroo capable system caused the runtime PM refcount to drop
below zero. The issue is resolved because the AZX_DCAPS_PM_RUNTIME flag
is now always set if use_vga_switcheroo is true.
For more information on device links please refer to:
https://www.kernel.org/doc/html/latest/driver-api/device_link.html
Documentation/driver-api/device_link.rst
Cc: Dave Airlie <airlied@redhat.com>
Cc: Ben Skeggs <bskeggs@redhat.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Takashi Iwai <tiwai@suse.de>
Reviewed-by: Peter Wu <peter@lekensteyn.nl>
Tested-by: Kai Heng Feng <kai.heng.feng@canonical.com> # AMD PowerXpress
Tested-by: Mike Lothian <mike@fireburn.co.uk> # AMD PowerXpress
Tested-by: Denis Lisov <dennis.lissov@gmail.com> # Nvidia Optimus
Tested-by: Peter Wu <peter@lekensteyn.nl> # Nvidia Optimus
Tested-by: Lukas Wunner <lukas@wunner.de> # MacBook Pro
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Link: https://patchwork.freedesktop.org/patch/msgid/51bd38360ff502a8c42b1ebf4405ee1d3f27118d.1520068884.git.lukas@wunner.de
2018-03-03 12:53:24 +03:00
# include <linux/pm_runtime.h>
PCI: Add DMA alias quirk for Microsemi Switchtec NTB
Add a quirk for the Microsemi Switchtec parts to allow DMA access via
non-transparent bridging to work when the IOMMU is turned on.
This exclusively addresses the ability of a remote NT endpoint to perform
DMA accesses through the locally enumerated NT endpoint. Other aspects of
the Switchtec NTB functionality, such as interrupts for doorbells and
messages are independent of this quirk, and will work whether the IOMMU is
on or off.
When a requestor on one NT endpoint accesses memory on another NT endpoint,
it does this via a devfn proxy ID. Proxy IDs are statically assigned to
each NT endpoint by the NTB hardware as part of the release-from-reset
sequence prior to PCI enumeration. These proxy IDs cannot be modified
dynamically, and are not visible to the host during enumeration.
When the Switchtec NTB driver loads it will map local requestor IDs, such
as the root complex and transparent bridge DMA engines, to proxy IDs by
populating those requestor IDs in hardware mapping table table entries.
This establishes a fixed relationship between a requestor ID and a proxy
ID.
When a peer on a remote NT endpoint performs an access within a particular
translation window in it's NT endpoint BAR address space, that access is
translated to a DMA request on the local endpoint's bus. As part of the
translation process, the original requestor ID has its devfn replaced with
the proxy ID, and the bus portion of the BDF is replaced with the bus of
the local NT endpoint. Thus, the DMA access from a remote NT endpoint will
appear on the local bus to have come from the unknown devfn which the IOMMU
will reject.
Interrogate NTB hardware registers for each remote NT endpoint to obtain
the proxy IDs that have been assigned to it and alias them to the local
(enumerated) NT endpoint's device. The IOMMU then accepts the remote proxy
IDs as if they were requests coming directly from the enumerated endpoint,
giving remote requestors access to memory resources which the local host
has made available.
Note that the aliasing of the proxy IDs cannot be performed at the driver
level given the current IOMMU architecture. Superficially this is because
pci_add_dma_alias() symbol is not exported. Functionally, the current
IOMMU design requires the aliasing to be performed prior to the creation of
IOMMU groups. If a driver were to attempt to use pci_add_dma_alias() in
its probe routine it would fail since the IOMMU groups have been set up by
that time. If the Switchtec hardware supported dynamic proxy ID
(re-)assignment this would be an issue, but it does not.
To further clarify static proxy ID assignment: While the requester ID to
proxy ID mapping can be dynamically changed, the number and value of proxy
IDs given to an NT EP cannot, even for dynamic reconfiguration such as
hot-add. Therefore, the chip configuration must account a priori for the
proxy IDs needs, considering both static and dynamic system configurations.
For example, a port on the chip may not having anything plugged into it at
start of day; but it must have a sufficient number of proxy IDs assigned to
accommodate the supported devices which may be hot-added.
Switchtec NTB functionality with the IOMMU off is unchanged by this quirk.
Signed-off-by: Doug Meyer <dmeyer@gigaio.com>
[bhelgaas: use hard-coded Device IDs instead of adding #defines for each]
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Logan Gunthorpe <logang@deltatee.com>
2018-05-23 23:18:06 +03:00
# include <linux/switchtec.h>
2010-01-03 00:57:24 +03:00
# include <asm/dma.h> /* isa_dma_bridge_buggy */
2005-04-08 09:53:31 +04:00
# include "pci.h"
2005-04-17 02:20:36 +04:00
2018-05-02 20:50:55 +03:00
static ktime_t fixup_debug_start ( struct pci_dev * dev ,
void ( * fn ) ( struct pci_dev * dev ) )
{
if ( initcall_debug )
2019-03-25 22:32:28 +03:00
pci_info ( dev , " calling %pS @ %i \n " , fn , task_pid_nr ( current ) ) ;
2018-05-02 20:50:55 +03:00
return ktime_get ( ) ;
}
static void fixup_debug_report ( struct pci_dev * dev , ktime_t calltime ,
void ( * fn ) ( struct pci_dev * dev ) )
{
ktime_t delta , rettime ;
unsigned long long duration ;
rettime = ktime_get ( ) ;
delta = ktime_sub ( rettime , calltime ) ;
duration = ( unsigned long long ) ktime_to_ns ( delta ) > > 10 ;
if ( initcall_debug | | duration > 10000 )
2019-03-25 22:32:28 +03:00
pci_info ( dev , " %pS took %lld usecs \n " , fn , duration ) ;
2018-05-02 20:50:55 +03:00
}
static void pci_do_fixups ( struct pci_dev * dev , struct pci_fixup * f ,
struct pci_fixup * end )
{
ktime_t calltime ;
for ( ; f < end ; f + + )
if ( ( f - > class = = ( u32 ) ( dev - > class > > f - > class_shift ) | |
f - > class = = ( u32 ) PCI_ANY_ID ) & &
( f - > vendor = = dev - > vendor | |
f - > vendor = = ( u16 ) PCI_ANY_ID ) & &
( f - > device = = dev - > device | |
f - > device = = ( u16 ) PCI_ANY_ID ) ) {
2018-08-22 07:56:18 +03:00
void ( * hook ) ( struct pci_dev * dev ) ;
# ifdef CONFIG_HAVE_ARCH_PREL32_RELOCATIONS
hook = offset_to_ptr ( & f - > hook_offset ) ;
# else
hook = f - > hook ;
# endif
calltime = fixup_debug_start ( dev , hook ) ;
hook ( dev ) ;
fixup_debug_report ( dev , calltime , hook ) ;
2018-05-02 20:50:55 +03:00
}
}
extern struct pci_fixup __start_pci_fixups_early [ ] ;
extern struct pci_fixup __end_pci_fixups_early [ ] ;
extern struct pci_fixup __start_pci_fixups_header [ ] ;
extern struct pci_fixup __end_pci_fixups_header [ ] ;
extern struct pci_fixup __start_pci_fixups_final [ ] ;
extern struct pci_fixup __end_pci_fixups_final [ ] ;
extern struct pci_fixup __start_pci_fixups_enable [ ] ;
extern struct pci_fixup __end_pci_fixups_enable [ ] ;
extern struct pci_fixup __start_pci_fixups_resume [ ] ;
extern struct pci_fixup __end_pci_fixups_resume [ ] ;
extern struct pci_fixup __start_pci_fixups_resume_early [ ] ;
extern struct pci_fixup __end_pci_fixups_resume_early [ ] ;
extern struct pci_fixup __start_pci_fixups_suspend [ ] ;
extern struct pci_fixup __end_pci_fixups_suspend [ ] ;
extern struct pci_fixup __start_pci_fixups_suspend_late [ ] ;
extern struct pci_fixup __end_pci_fixups_suspend_late [ ] ;
static bool pci_apply_fixup_final_quirks ;
void pci_fixup_device ( enum pci_fixup_pass pass , struct pci_dev * dev )
{
struct pci_fixup * start , * end ;
switch ( pass ) {
case pci_fixup_early :
start = __start_pci_fixups_early ;
end = __end_pci_fixups_early ;
break ;
case pci_fixup_header :
start = __start_pci_fixups_header ;
end = __end_pci_fixups_header ;
break ;
case pci_fixup_final :
if ( ! pci_apply_fixup_final_quirks )
return ;
start = __start_pci_fixups_final ;
end = __end_pci_fixups_final ;
break ;
case pci_fixup_enable :
start = __start_pci_fixups_enable ;
end = __end_pci_fixups_enable ;
break ;
case pci_fixup_resume :
start = __start_pci_fixups_resume ;
end = __end_pci_fixups_resume ;
break ;
case pci_fixup_resume_early :
start = __start_pci_fixups_resume_early ;
end = __end_pci_fixups_resume_early ;
break ;
case pci_fixup_suspend :
start = __start_pci_fixups_suspend ;
end = __end_pci_fixups_suspend ;
break ;
case pci_fixup_suspend_late :
start = __start_pci_fixups_suspend_late ;
end = __end_pci_fixups_suspend_late ;
break ;
default :
/* stupid compiler warning, you would think with an enum... */
return ;
}
pci_do_fixups ( dev , start , end ) ;
}
EXPORT_SYMBOL ( pci_fixup_device ) ;
static int __init pci_apply_final_quirks ( void )
{
struct pci_dev * dev = NULL ;
u8 cls = 0 ;
u8 tmp ;
if ( pci_cache_line_size )
2019-04-20 07:07:20 +03:00
pr_info ( " PCI: CLS %u bytes \n " , pci_cache_line_size < < 2 ) ;
2018-05-02 20:50:55 +03:00
pci_apply_fixup_final_quirks = true ;
for_each_pci_dev ( dev ) {
pci_fixup_device ( pci_fixup_final , dev ) ;
/*
* If arch hasn ' t set it explicitly yet , use the CLS
* value shared by all PCI devices . If there ' s a
* mismatch , fall back to the default value .
*/
if ( ! pci_cache_line_size ) {
pci_read_config_byte ( dev , PCI_CACHE_LINE_SIZE , & tmp ) ;
if ( ! cls )
cls = tmp ;
if ( ! tmp | | cls = = tmp )
continue ;
2019-04-20 07:07:20 +03:00
pci_info ( dev , " CLS mismatch (%u != %u), using %u bytes \n " ,
cls < < 2 , tmp < < 2 ,
pci_dfl_cache_line_size < < 2 ) ;
2018-05-02 20:50:55 +03:00
pci_cache_line_size = pci_dfl_cache_line_size ;
}
}
if ( ! pci_cache_line_size ) {
2019-04-20 07:07:20 +03:00
pr_info ( " PCI: CLS %u bytes, default %u \n " , cls < < 2 ,
pci_dfl_cache_line_size < < 2 ) ;
2018-05-02 20:50:55 +03:00
pci_cache_line_size = cls ? cls : pci_dfl_cache_line_size ;
}
return 0 ;
}
fs_initcall_sync ( pci_apply_final_quirks ) ;
2010-07-16 21:19:22 +04:00
/*
* Decoding should be disabled for a PCI device during BAR sizing to avoid
* conflict . But doing so may cause problems on host bridge and perhaps other
* key system devices . For devices that need to have mmio decoding always - on ,
* we need to set the dev - > mmio_always_on bit .
*/
2012-11-22 00:35:00 +04:00
static void quirk_mmio_always_on ( struct pci_dev * dev )
2010-07-16 21:19:22 +04:00
{
2012-02-24 11:46:53 +04:00
dev - > mmio_always_on = 1 ;
2010-07-16 21:19:22 +04:00
}
2012-02-24 11:46:53 +04:00
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_ANY_ID , PCI_ANY_ID ,
PCI_CLASS_BRIDGE_HOST , 8 , quirk_mmio_always_on ) ;
2010-07-16 21:19:22 +04:00
2018-05-02 16:53:19 +03:00
/*
* The Mellanox Tavor device gives false positive parity errors . Mark this
* device with a broken_parity_status to allow PCI scanning code to " skip "
* this now blacklisted device .
2006-05-09 04:06:09 +04:00
*/
2012-11-22 00:35:00 +04:00
static void quirk_mellanox_tavor ( struct pci_dev * dev )
2006-05-09 04:06:09 +04:00
{
dev - > broken_parity_status = 1 ; /* This device gives false positives */
}
2014-04-19 04:13:49 +04:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_MELLANOX , PCI_DEVICE_ID_MELLANOX_TAVOR , quirk_mellanox_tavor ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_MELLANOX , PCI_DEVICE_ID_MELLANOX_TAVOR_BRIDGE , quirk_mellanox_tavor ) ;
2006-05-09 04:06:09 +04:00
2018-05-02 16:53:19 +03:00
/*
* Deal with broken BIOSes that neglect to enable passive release ,
* which can cause problems in combination with the 82441F X / PPro MTRRs
*/
2006-12-05 02:14:45 +03:00
static void quirk_passive_release ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
struct pci_dev * d = NULL ;
unsigned char dlc ;
2018-05-02 16:53:19 +03:00
/*
* We have to make sure a particular bit is set in the PIIX3
* ISA bridge , so we have to go out and find it .
*/
2005-04-17 02:20:36 +04:00
while ( ( d = pci_get_device ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82371SB_0 , d ) ) ) {
pci_read_config_byte ( d , 0x82 , & dlc ) ;
if ( ! ( dlc & 1 < < 1 ) ) {
2018-01-18 21:55:24 +03:00
pci_info ( d , " PIIX3: Enabling Passive Release \n " ) ;
2005-04-17 02:20:36 +04:00
dlc | = 1 < < 1 ;
pci_write_config_byte ( d , 0x82 , dlc ) ;
}
}
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82441 , quirk_passive_release ) ;
DECLARE_PCI_FIXUP_RESUME ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82441 , quirk_passive_release ) ;
2005-04-17 02:20:36 +04:00
2018-05-02 16:53:19 +03:00
/*
* The VIA VP2 / VP3 / MVP3 seem to have some ' features ' . There may be a
* workaround but VIA don ' t answer queries . If you happen to have good
* contacts at VIA ask them for me please - - Alan
*
* This appears to be BIOS not version dependent . So presumably there is a
* chipset level fix .
*/
2012-11-22 00:35:00 +04:00
static void quirk_isa_dma_hangs ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
if ( ! isa_dma_bridge_buggy ) {
2014-04-19 04:13:49 +04:00
isa_dma_bridge_buggy = 1 ;
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Activating ISA DMA hang workarounds \n " ) ;
2005-04-17 02:20:36 +04:00
}
}
2018-05-02 16:53:19 +03:00
/*
* It ' s not totally clear which chipsets are the problematic ones . We know
* 82 C586 and 82 C596 variants are affected .
*/
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_82C586_0 , quirk_isa_dma_hangs ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_82C596 , quirk_isa_dma_hangs ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82371SB_0 , quirk_isa_dma_hangs ) ;
2013-11-14 22:28:18 +04:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_AL , PCI_DEVICE_ID_AL_M1533 , quirk_isa_dma_hangs ) ;
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_NEC , PCI_DEVICE_ID_NEC_CBUS_1 , quirk_isa_dma_hangs ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_NEC , PCI_DEVICE_ID_NEC_CBUS_2 , quirk_isa_dma_hangs ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_NEC , PCI_DEVICE_ID_NEC_CBUS_3 , quirk_isa_dma_hangs ) ;
2005-04-17 02:20:36 +04:00
2010-09-25 05:02:27 +04:00
/*
* Intel NM10 " TigerPoint " LPC PM1a_STS . BM_STS must be clear
* for some HT machines to use C4 w / o hanging .
*/
2012-11-22 00:35:00 +04:00
static void quirk_tigerpoint_bm_sts ( struct pci_dev * dev )
2010-09-25 05:02:27 +04:00
{
u32 pmbase ;
u16 pm1a ;
pci_read_config_dword ( dev , 0x40 , & pmbase ) ;
pmbase = pmbase & 0xff80 ;
pm1a = inw ( pmbase ) ;
if ( pm1a & 0x10 ) {
2018-01-18 21:55:24 +03:00
pci_info ( dev , FW_BUG " TigerPoint LPC.BM_STS cleared \n " ) ;
2010-09-25 05:02:27 +04:00
outw ( 0x10 , pmbase ) ;
}
}
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_TGP_LPC , quirk_tigerpoint_bm_sts ) ;
2018-05-02 16:53:19 +03:00
/* Chipsets where PCI->PCI transfers vanish or hang */
2012-11-22 00:35:00 +04:00
static void quirk_nopcipci ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
2014-04-19 04:13:49 +04:00
if ( ( pci_pci_problems & PCIPCI_FAIL ) = = 0 ) {
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Disabling direct PCI/PCI transfers \n " ) ;
2005-04-17 02:20:36 +04:00
pci_pci_problems | = PCIPCI_FAIL ;
}
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_SI , PCI_DEVICE_ID_SI_5597 , quirk_nopcipci ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_SI , PCI_DEVICE_ID_SI_496 , quirk_nopcipci ) ;
2006-10-01 10:27:03 +04:00
2012-11-22 00:35:00 +04:00
static void quirk_nopciamd ( struct pci_dev * dev )
2006-10-01 10:27:03 +04:00
{
u8 rev ;
pci_read_config_byte ( dev , 0x08 , & rev ) ;
if ( rev = = 0x13 ) {
/* Erratum 24 */
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Chipset erratum: Disabling direct PCI/AGP transfers \n " ) ;
2006-10-01 10:27:03 +04:00
pci_pci_problems | = PCIAGP_FAIL ;
}
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_AMD , PCI_DEVICE_ID_AMD_8151_0 , quirk_nopciamd ) ;
2005-04-17 02:20:36 +04:00
2018-05-02 16:53:19 +03:00
/* Triton requires workarounds to be used by the drivers */
2012-11-22 00:35:00 +04:00
static void quirk_triton ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
2014-04-19 04:13:49 +04:00
if ( ( pci_pci_problems & PCIPCI_TRITON ) = = 0 ) {
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Limiting direct PCI/PCI transfers \n " ) ;
2005-04-17 02:20:36 +04:00
pci_pci_problems | = PCIPCI_TRITON ;
}
}
2013-11-14 22:28:18 +04:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82437 , quirk_triton ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82437VX , quirk_triton ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82439 , quirk_triton ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82439TX , quirk_triton ) ;
2005-04-17 02:20:36 +04:00
/*
2018-05-02 16:53:19 +03:00
* VIA Apollo KT133 needs PCI latency patch
* Made according to a Windows driver - based patch by George E . Breese ;
* see PCI Latency Adjust on http : //www.viahardware.com/download/viatweak.shtm
* Also see http : //www.au-ja.org/review-kt133a-1-en.phtml for the info on
* which Mr Breese based his work .
2005-04-17 02:20:36 +04:00
*
2018-05-02 16:53:19 +03:00
* Updated based on further information from the site and also on
* information provided by VIA
2005-04-17 02:20:36 +04:00
*/
2006-12-05 02:14:45 +03:00
static void quirk_vialatency ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
struct pci_dev * p ;
u8 busarb ;
2013-11-14 22:28:18 +04:00
2018-05-02 16:53:19 +03:00
/*
* Ok , we have a potential problem chipset here . Now see if we have
* a buggy southbridge .
*/
2005-04-17 02:20:36 +04:00
p = pci_get_device ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_82C686 , NULL ) ;
2014-04-19 04:13:49 +04:00
if ( p ! = NULL ) {
2018-05-02 16:53:19 +03:00
/*
* 0x40 - 0x4f = = 686 B , 0x10 - 0x2f = = 686 A ;
* thanks Dan Hollis .
* Check for buggy part revisions
*/
2007-10-30 00:55:02 +03:00
if ( p - > revision < 0x40 | | p - > revision > 0x42 )
2005-04-17 02:20:36 +04:00
goto exit ;
} else {
p = pci_get_device ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_8231 , NULL ) ;
2014-04-19 04:13:49 +04:00
if ( p = = NULL ) /* No problem parts */
2005-04-17 02:20:36 +04:00
goto exit ;
2018-05-02 16:53:19 +03:00
2005-04-17 02:20:36 +04:00
/* Check for buggy part revisions */
2007-10-30 00:55:02 +03:00
if ( p - > revision < 0x10 | | p - > revision > 0x12 )
2005-04-17 02:20:36 +04:00
goto exit ;
}
2013-11-14 22:28:18 +04:00
2005-04-17 02:20:36 +04:00
/*
2018-05-02 16:53:19 +03:00
* Ok we have the problem . Now set the PCI master grant to occur
* every master grant . The apparent bug is that under high PCI load
* ( quite common in Linux of course ) you can get data loss when the
* CPU is held off the bus for 3 bus master requests . This happens
* to include the IDE controllers . . . .
2005-04-17 02:20:36 +04:00
*
2018-05-02 16:53:19 +03:00
* VIA only apply this fix when an SB Live ! is present but under
* both Linux and Windows this isn ' t enough , and we have seen
* corruption without SB Live ! but with things like 3 UDMA IDE
* controllers . So we ignore that bit of the VIA recommendation . .
2005-04-17 02:20:36 +04:00
*/
pci_read_config_byte ( dev , 0x76 , & busarb ) ;
2018-05-02 16:53:19 +03:00
/*
* Set bit 4 and bit 5 of byte 76 to 0x01
* " Master priority rotation on every PCI master grant "
*/
2005-04-17 02:20:36 +04:00
busarb & = ~ ( 1 < < 5 ) ;
busarb | = ( 1 < < 4 ) ;
pci_write_config_byte ( dev , 0x76 , busarb ) ;
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Applying VIA southbridge workaround \n " ) ;
2005-04-17 02:20:36 +04:00
exit :
pci_dev_put ( p ) ;
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_8363_0 , quirk_vialatency ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_8371_1 , quirk_vialatency ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_8361 , quirk_vialatency ) ;
2006-12-05 02:14:45 +03:00
/* Must restore this on a resume from RAM */
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_RESUME ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_8363_0 , quirk_vialatency ) ;
DECLARE_PCI_FIXUP_RESUME ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_8371_1 , quirk_vialatency ) ;
DECLARE_PCI_FIXUP_RESUME ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_8361 , quirk_vialatency ) ;
2005-04-17 02:20:36 +04:00
2018-05-02 16:53:19 +03:00
/* VIA Apollo VP3 needs ETBF on BT848/878 */
2012-11-22 00:35:00 +04:00
static void quirk_viaetbf ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
2014-04-19 04:13:49 +04:00
if ( ( pci_pci_problems & PCIPCI_VIAETBF ) = = 0 ) {
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Limiting direct PCI/PCI transfers \n " ) ;
2005-04-17 02:20:36 +04:00
pci_pci_problems | = PCIPCI_VIAETBF ;
}
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_82C597_0 , quirk_viaetbf ) ;
2005-04-17 02:20:36 +04:00
2012-11-22 00:35:00 +04:00
static void quirk_vsfx ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
2014-04-19 04:13:49 +04:00
if ( ( pci_pci_problems & PCIPCI_VSFX ) = = 0 ) {
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Limiting direct PCI/PCI transfers \n " ) ;
2005-04-17 02:20:36 +04:00
pci_pci_problems | = PCIPCI_VSFX ;
}
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_82C576 , quirk_vsfx ) ;
2005-04-17 02:20:36 +04:00
/*
2018-05-02 16:53:19 +03:00
* ALi Magik requires workarounds to be used by the drivers that DMA to AGP
* space . Latency must be set to 0xA and Triton workaround applied too .
* [ Info kindly provided by ALi ]
2013-11-14 22:28:18 +04:00
*/
2012-11-22 00:35:00 +04:00
static void quirk_alimagik ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
2014-04-19 04:13:49 +04:00
if ( ( pci_pci_problems & PCIPCI_ALIMAGIK ) = = 0 ) {
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Limiting direct PCI/PCI transfers \n " ) ;
2005-04-17 02:20:36 +04:00
pci_pci_problems | = PCIPCI_ALIMAGIK | PCIPCI_TRITON ;
}
}
2013-11-14 22:28:18 +04:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_AL , PCI_DEVICE_ID_AL_M1647 , quirk_alimagik ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_AL , PCI_DEVICE_ID_AL_M1651 , quirk_alimagik ) ;
2005-04-17 02:20:36 +04:00
2018-05-02 16:53:19 +03:00
/* Natoma has some interesting boundary conditions with Zoran stuff at least */
2012-11-22 00:35:00 +04:00
static void quirk_natoma ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
2014-04-19 04:13:49 +04:00
if ( ( pci_pci_problems & PCIPCI_NATOMA ) = = 0 ) {
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Limiting direct PCI/PCI transfers \n " ) ;
2005-04-17 02:20:36 +04:00
pci_pci_problems | = PCIPCI_NATOMA ;
}
}
2013-11-14 22:28:18 +04:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82441 , quirk_natoma ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82443LX_0 , quirk_natoma ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82443LX_1 , quirk_natoma ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82443BX_0 , quirk_natoma ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82443BX_1 , quirk_natoma ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82443BX_2 , quirk_natoma ) ;
2005-04-17 02:20:36 +04:00
/*
2018-05-02 16:53:19 +03:00
* This chip can cause PCI parity errors if config register 0xA0 is read
* while DMAs are occurring .
2005-04-17 02:20:36 +04:00
*/
2012-11-22 00:35:00 +04:00
static void quirk_citrine ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
dev - > cfg_size = 0xA0 ;
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_IBM , PCI_DEVICE_ID_IBM_CITRINE , quirk_citrine ) ;
2005-04-17 02:20:36 +04:00
2015-09-30 09:35:07 +03:00
/*
* This chip can cause bus lockups if config addresses above 0x600
* are read or written .
*/
static void quirk_nfp6000 ( struct pci_dev * dev )
{
dev - > cfg_size = 0x600 ;
}
2015-12-11 05:30:12 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_NETRONOME , PCI_DEVICE_ID_NETRONOME_NFP4000 , quirk_nfp6000 ) ;
2015-09-30 09:35:07 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_NETRONOME , PCI_DEVICE_ID_NETRONOME_NFP6000 , quirk_nfp6000 ) ;
2018-08-15 02:48:50 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_NETRONOME , PCI_DEVICE_ID_NETRONOME_NFP5000 , quirk_nfp6000 ) ;
2015-09-30 09:35:07 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_NETRONOME , PCI_DEVICE_ID_NETRONOME_NFP6000_VF , quirk_nfp6000 ) ;
2014-08-21 03:26:52 +04:00
/* On IBM Crocodile ipr SAS adapters, expand BAR to system page size */
static void quirk_extend_bar_to_page ( struct pci_dev * dev )
{
int i ;
2019-09-28 02:43:08 +03:00
for ( i = 0 ; i < PCI_STD_NUM_BARS ; i + + ) {
2014-08-21 03:26:52 +04:00
struct resource * r = & dev - > resource [ i ] ;
if ( r - > flags & IORESOURCE_MEM & & resource_size ( r ) < PAGE_SIZE ) {
r - > end = PAGE_SIZE - 1 ;
r - > start = 0 ;
r - > flags | = IORESOURCE_UNSET ;
2018-01-18 21:55:24 +03:00
pci_info ( dev , " expanded BAR %d to page size: %pR \n " ,
2014-08-21 03:26:52 +04:00
i , r ) ;
}
}
}
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_IBM , 0x034a , quirk_extend_bar_to_page ) ;
2005-04-17 02:20:36 +04:00
/*
2018-05-02 16:53:19 +03:00
* S3 868 and 968 chips report region size equal to 32 M , but they decode 64 M .
* If it ' s needed , re - allocate the region .
2005-04-17 02:20:36 +04:00
*/
2012-11-22 00:35:00 +04:00
static void quirk_s3_64M ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
struct resource * r = & dev - > resource [ 0 ] ;
if ( ( r - > start & 0x3ffffff ) | | r - > end ! = r - > start + 0x3ffffff ) {
2014-02-26 22:25:58 +04:00
r - > flags | = IORESOURCE_UNSET ;
2005-04-17 02:20:36 +04:00
r - > start = 0 ;
r - > end = 0x3ffffff ;
}
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_S3 , PCI_DEVICE_ID_S3_868 , quirk_s3_64M ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_S3 , PCI_DEVICE_ID_S3_968 , quirk_s3_64M ) ;
2005-04-17 02:20:36 +04:00
2015-02-04 02:01:24 +03:00
static void quirk_io ( struct pci_dev * dev , int pos , unsigned size ,
const char * name )
{
u32 region ;
struct pci_bus_region bus_region ;
struct resource * res = dev - > resource + pos ;
pci_read_config_dword ( dev , PCI_BASE_ADDRESS_0 + ( pos < < 2 ) , & region ) ;
if ( ! region )
return ;
res - > name = pci_name ( dev ) ;
res - > flags = region & ~ PCI_BASE_ADDRESS_IO_MASK ;
res - > flags | =
( IORESOURCE_IO | IORESOURCE_PCI_FIXED | IORESOURCE_SIZEALIGN ) ;
region & = ~ ( size - 1 ) ;
/* Convert from PCI bus to resource space */
bus_region . start = region ;
bus_region . end = region + size - 1 ;
pcibios_bus_to_resource ( dev - > bus , res , & bus_region ) ;
2018-01-18 21:55:24 +03:00
pci_info ( dev , FW_BUG " %s quirk: reg 0x%x: %pR \n " ,
2015-02-04 02:01:24 +03:00
name , PCI_BASE_ADDRESS_0 + ( pos < < 2 ) , res ) ;
}
2010-02-05 09:42:43 +03:00
/*
* Some CS5536 BIOSes ( for example , the Soekris NET5501 board w / comBIOS
* ver . 1.33 20070103 ) don ' t set the correct ISA PCI region header info .
* BAR0 should be 8 bytes ; instead , it may be set to something like 8 k
* ( which conflicts w / BAR1 ' s memory range ) .
2015-02-04 02:01:24 +03:00
*
* CS553x ' s ISA PCI BARs may also be read - only ( ref :
* https : //bugzilla.kernel.org/show_bug.cgi?id=85991 - Comment #4 forward).
2010-02-05 09:42:43 +03:00
*/
2012-11-22 00:35:00 +04:00
static void quirk_cs5536_vsa ( struct pci_dev * dev )
2010-02-05 09:42:43 +03:00
{
2015-02-04 02:01:24 +03:00
static char * name = " CS5536 ISA bridge " ;
2010-02-05 09:42:43 +03:00
if ( pci_resource_len ( dev , 0 ) ! = 8 ) {
2015-02-04 02:01:24 +03:00
quirk_io ( dev , 0 , 8 , name ) ; /* SMB */
quirk_io ( dev , 1 , 256 , name ) ; /* GPIO */
quirk_io ( dev , 2 , 64 , name ) ; /* MFGPT */
2018-01-18 21:55:24 +03:00
pci_info ( dev , " %s bug detected (incorrect header); workaround applied \n " ,
2015-02-04 02:01:24 +03:00
name ) ;
2010-02-05 09:42:43 +03:00
}
}
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_AMD , PCI_DEVICE_ID_AMD_CS5536_ISA , quirk_cs5536_vsa ) ;
2013-04-12 16:44:15 +04:00
static void quirk_io_region ( struct pci_dev * dev , int port ,
unsigned size , int nr , const char * name )
{
u16 region ;
struct pci_bus_region bus_region ;
struct resource * res = dev - > resource + nr ;
pci_read_config_word ( dev , port , & region ) ;
region & = ~ ( size - 1 ) ;
if ( ! region )
return ;
res - > name = pci_name ( dev ) ;
res - > flags = IORESOURCE_IO ;
/* Convert from PCI bus to resource space */
bus_region . start = region ;
bus_region . end = region + size - 1 ;
PCI: Convert pcibios_resource_to_bus() to take a pci_bus, not a pci_dev
These interfaces:
pcibios_resource_to_bus(struct pci_dev *dev, *bus_region, *resource)
pcibios_bus_to_resource(struct pci_dev *dev, *resource, *bus_region)
took a pci_dev, but they really depend only on the pci_bus. And we want to
use them in resource allocation paths where we have the bus but not a
device, so this patch converts them to take the pci_bus instead of the
pci_dev:
pcibios_resource_to_bus(struct pci_bus *bus, *bus_region, *resource)
pcibios_bus_to_resource(struct pci_bus *bus, *resource, *bus_region)
In fact, with standard PCI-PCI bridges, they only depend on the host
bridge, because that's the only place address translation occurs, but
we aren't going that far yet.
[bhelgaas: changelog]
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2013-12-10 10:54:40 +04:00
pcibios_bus_to_resource ( dev - > bus , res , & bus_region ) ;
2013-04-12 16:44:15 +04:00
if ( ! pci_claim_resource ( dev , nr ) )
2018-01-18 21:55:24 +03:00
pci_info ( dev , " quirk: %pR claimed by %s \n " , res , name ) ;
2013-04-12 16:44:15 +04:00
}
2005-04-17 02:20:36 +04:00
/*
2018-05-02 16:53:19 +03:00
* ATI Northbridge setups MCE the processor if you even read somewhere
* between 0x3b0 - > 0x3bb or read 0x3d3
2005-04-17 02:20:36 +04:00
*/
2012-11-22 00:35:00 +04:00
static void quirk_ati_exploding_mce ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
2018-01-18 21:55:24 +03:00
pci_info ( dev , " ATI Northbridge, reserving I/O ports 0x3b0 to 0x3bb \n " ) ;
2005-04-17 02:20:36 +04:00
/* Mae rhaid i ni beidio ag edrych ar y lleoliadiau I/O hyn */
request_region ( 0x3b0 , 0x0C , " RadeonIGP " ) ;
request_region ( 0x3d3 , 0x01 , " RadeonIGP " ) ;
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_ATI , PCI_DEVICE_ID_ATI_RS100 , quirk_ati_exploding_mce ) ;
2005-04-17 02:20:36 +04:00
2014-10-31 06:11:16 +03:00
/*
* In the AMD NL platform , this device ( [ 1022 : 7912 ] ) has a class code of
* PCI_CLASS_SERIAL_USB_XHCI ( 0x0c0330 ) , which means the xhci driver will
* claim it .
2018-05-02 16:53:19 +03:00
*
2014-10-31 06:11:16 +03:00
* But the dwc3 driver is a more specific driver for this device , and we ' d
* prefer to use it instead of xhci . To prevent xhci from claiming the
* device , change the class code to 0x0c03fe , which the PCI r3 .0 spec
* defines as " USB device (not host controller) " . The dwc3 driver can then
* claim it based on its Vendor and Device ID .
*/
static void quirk_amd_nl_class ( struct pci_dev * pdev )
{
2015-06-19 23:28:31 +03:00
u32 class = pdev - > class ;
/* Use "USB Device (not host controller)" class */
2016-03-15 15:06:00 +03:00
pdev - > class = PCI_CLASS_SERIAL_USB_DEVICE ;
2018-01-18 21:55:24 +03:00
pci_info ( pdev , " PCI class overridden (%#08x -> %#08x) so dwc3 driver can claim this instead of xhci \n " ,
2015-06-19 23:28:31 +03:00
class , pdev - > class ) ;
2014-10-31 06:11:16 +03:00
}
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_AMD , PCI_DEVICE_ID_AMD_NL_USB ,
quirk_amd_nl_class ) ;
2018-12-11 01:08:01 +03:00
/*
* Synopsys USB 3. x host HAPS platform has a class code of
* PCI_CLASS_SERIAL_USB_XHCI , and xhci driver can claim it . However , these
* devices should use dwc3 - haps driver . Change these devices ' class code to
* PCI_CLASS_SERIAL_USB_DEVICE to prevent the xhci - pci driver from claiming
* them .
*/
static void quirk_synopsys_haps ( struct pci_dev * pdev )
{
u32 class = pdev - > class ;
switch ( pdev - > device ) {
case PCI_DEVICE_ID_SYNOPSYS_HAPSUSB3 :
case PCI_DEVICE_ID_SYNOPSYS_HAPSUSB3_AXI :
case PCI_DEVICE_ID_SYNOPSYS_HAPSUSB31 :
pdev - > class = PCI_CLASS_SERIAL_USB_DEVICE ;
pci_info ( pdev , " PCI class overridden (%#08x -> %#08x) so dwc3 driver can claim this instead of xhci \n " ,
class , pdev - > class ) ;
break ;
}
}
2019-02-07 02:17:27 +03:00
DECLARE_PCI_FIXUP_CLASS_HEADER ( PCI_VENDOR_ID_SYNOPSYS , PCI_ANY_ID ,
PCI_CLASS_SERIAL_USB_XHCI , 0 ,
quirk_synopsys_haps ) ;
2018-12-11 01:08:01 +03:00
2005-04-17 02:20:36 +04:00
/*
2018-05-02 16:53:19 +03:00
* Let ' s make the southbridge information explicit instead of having to
* worry about people probing the ACPI areas , for example . . ( Yes , it
* happens , and if you read the wrong ACPI register it will put the machine
* to sleep with no way of waking it up again . Bummer ) .
2005-04-17 02:20:36 +04:00
*
* ALI M7101 : Two IO regions pointed to by words at
* 0xE0 ( 64 bytes of ACPI registers )
* 0xE2 ( 32 bytes of SMB registers )
*/
2012-11-22 00:35:00 +04:00
static void quirk_ali7101_acpi ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
2013-04-12 16:44:15 +04:00
quirk_io_region ( dev , 0xE0 , 64 , PCI_BRIDGE_RESOURCES , " ali7101 ACPI " ) ;
quirk_io_region ( dev , 0xE2 , 32 , PCI_BRIDGE_RESOURCES + 1 , " ali7101 SMB " ) ;
2005-04-17 02:20:36 +04:00
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_AL , PCI_DEVICE_ID_AL_M7101 , quirk_ali7101_acpi ) ;
2005-04-17 02:20:36 +04:00
2005-10-26 07:40:09 +04:00
static void piix4_io_quirk ( struct pci_dev * dev , const char * name , unsigned int port , unsigned int enable )
{
u32 devres ;
u32 mask , size , base ;
pci_read_config_dword ( dev , port , & devres ) ;
if ( ( devres & enable ) ! = enable )
return ;
mask = ( devres > > 16 ) & 15 ;
base = devres & 0xffff ;
size = 16 ;
for ( ; ; ) {
unsigned bit = size > > 1 ;
if ( ( bit & mask ) = = bit )
break ;
size = bit ;
}
/*
* For now we only print it out . Eventually we ' ll want to
* reserve it ( at least if it ' s in the 0x1000 + range ) , but
2013-11-14 22:28:18 +04:00
* let ' s get enough confirmation reports first .
2005-10-26 07:40:09 +04:00
*/
base & = - size ;
2018-01-18 21:55:24 +03:00
pci_info ( dev , " %s PIO at %04x-%04x \n " , name , base , base + size - 1 ) ;
2005-10-26 07:40:09 +04:00
}
static void piix4_mem_quirk ( struct pci_dev * dev , const char * name , unsigned int port , unsigned int enable )
{
u32 devres ;
u32 mask , size , base ;
pci_read_config_dword ( dev , port , & devres ) ;
if ( ( devres & enable ) ! = enable )
return ;
base = devres & 0xffff0000 ;
mask = ( devres & 0x3f ) < < 16 ;
size = 128 < < 16 ;
for ( ; ; ) {
unsigned bit = size > > 1 ;
if ( ( bit & mask ) = = bit )
break ;
size = bit ;
}
2018-05-02 16:53:19 +03:00
2005-10-26 07:40:09 +04:00
/*
* For now we only print it out . Eventually we ' ll want to
2013-11-14 22:28:18 +04:00
* reserve it , but let ' s get enough confirmation reports first .
2005-10-26 07:40:09 +04:00
*/
base & = - size ;
2018-01-18 21:55:24 +03:00
pci_info ( dev , " %s MMIO at %04x-%04x \n " , name , base , base + size - 1 ) ;
2005-10-26 07:40:09 +04:00
}
2005-04-17 02:20:36 +04:00
/*
* PIIX4 ACPI : Two IO regions pointed to by longwords at
* 0x40 ( 64 bytes of ACPI registers )
2005-10-31 01:40:07 +03:00
* 0x90 ( 16 bytes of SMB registers )
2005-10-26 07:40:09 +04:00
* and a few strange programmable PIIX4 device resources .
2005-04-17 02:20:36 +04:00
*/
2012-11-22 00:35:00 +04:00
static void quirk_piix4_acpi ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
2013-04-12 16:44:15 +04:00
u32 res_a ;
2005-04-17 02:20:36 +04:00
2013-04-12 16:44:15 +04:00
quirk_io_region ( dev , 0x40 , 64 , PCI_BRIDGE_RESOURCES , " PIIX4 ACPI " ) ;
quirk_io_region ( dev , 0x90 , 16 , PCI_BRIDGE_RESOURCES + 1 , " PIIX4 SMB " ) ;
2005-10-26 07:40:09 +04:00
/* Device resource A has enables for some of the other ones */
pci_read_config_dword ( dev , 0x5c , & res_a ) ;
piix4_io_quirk ( dev , " PIIX4 devres B " , 0x60 , 3 < < 21 ) ;
piix4_io_quirk ( dev , " PIIX4 devres C " , 0x64 , 3 < < 21 ) ;
/* Device resource D is just bitfields for static resources */
/* Device 12 enabled? */
if ( res_a & ( 1 < < 29 ) ) {
piix4_io_quirk ( dev , " PIIX4 devres E " , 0x68 , 1 < < 20 ) ;
piix4_mem_quirk ( dev , " PIIX4 devres F " , 0x6c , 1 < < 7 ) ;
}
/* Device 13 enabled? */
if ( res_a & ( 1 < < 30 ) ) {
piix4_io_quirk ( dev , " PIIX4 devres G " , 0x70 , 1 < < 20 ) ;
piix4_mem_quirk ( dev , " PIIX4 devres H " , 0x74 , 1 < < 7 ) ;
}
piix4_io_quirk ( dev , " PIIX4 devres I " , 0x78 , 1 < < 20 ) ;
piix4_io_quirk ( dev , " PIIX4 devres J " , 0x7c , 1 < < 20 ) ;
2005-04-17 02:20:36 +04:00
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82371AB_3 , quirk_piix4_acpi ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82443MX_3 , quirk_piix4_acpi ) ;
2005-04-17 02:20:36 +04:00
2011-02-28 12:45:09 +03:00
# define ICH_PMBASE 0x40
# define ICH_ACPI_CNTL 0x44
# define ICH4_ACPI_EN 0x10
# define ICH6_ACPI_EN 0x80
# define ICH4_GPIOBASE 0x58
# define ICH4_GPIO_CNTL 0x5c
# define ICH4_GPIO_EN 0x10
# define ICH6_GPIOBASE 0x48
# define ICH6_GPIO_CNTL 0x4c
# define ICH6_GPIO_EN 0x10
2005-04-17 02:20:36 +04:00
/*
* ICH4 , ICH4 - M , ICH5 , ICH5 - M ACPI : Three IO regions pointed to by longwords at
* 0x40 ( 128 bytes of ACPI , GPIO & TCO registers )
* 0x58 ( 64 bytes of GPIO I / O space )
*/
2012-11-22 00:35:00 +04:00
static void quirk_ich4_lpc_acpi ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
2011-02-28 12:45:09 +03:00
u8 enable ;
2005-04-17 02:20:36 +04:00
2011-02-28 12:45:10 +03:00
/*
* The check for PCIBIOS_MIN_IO is to ensure we won ' t create a conflict
* with low legacy ( and fixed ) ports . We don ' t know the decoding
* priority and can ' t tell whether the legacy device or the one created
* here is really at that address . This happens on boards with broken
* BIOSes .
2018-05-02 16:53:19 +03:00
*/
2011-02-28 12:45:09 +03:00
pci_read_config_byte ( dev , ICH_ACPI_CNTL , & enable ) ;
2013-04-12 16:44:15 +04:00
if ( enable & ICH4_ACPI_EN )
quirk_io_region ( dev , ICH_PMBASE , 128 , PCI_BRIDGE_RESOURCES ,
" ICH4 ACPI/GPIO/TCO " ) ;
2005-04-17 02:20:36 +04:00
2011-02-28 12:45:09 +03:00
pci_read_config_byte ( dev , ICH4_GPIO_CNTL , & enable ) ;
2013-04-12 16:44:15 +04:00
if ( enable & ICH4_GPIO_EN )
quirk_io_region ( dev , ICH4_GPIOBASE , 64 , PCI_BRIDGE_RESOURCES + 1 ,
" ICH4 GPIO " ) ;
2005-04-17 02:20:36 +04:00
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82801AA_0 , quirk_ich4_lpc_acpi ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82801AB_0 , quirk_ich4_lpc_acpi ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82801BA_0 , quirk_ich4_lpc_acpi ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82801BA_10 , quirk_ich4_lpc_acpi ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82801CA_0 , quirk_ich4_lpc_acpi ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82801CA_12 , quirk_ich4_lpc_acpi ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82801DB_0 , quirk_ich4_lpc_acpi ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82801DB_12 , quirk_ich4_lpc_acpi ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82801EB_0 , quirk_ich4_lpc_acpi ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_ESB_1 , quirk_ich4_lpc_acpi ) ;
2005-04-17 02:20:36 +04:00
2012-11-22 00:35:00 +04:00
static void ich6_lpc_acpi_gpio ( struct pci_dev * dev )
2005-09-28 01:54:51 +04:00
{
2011-02-28 12:45:09 +03:00
u8 enable ;
2005-09-28 01:54:51 +04:00
2011-02-28 12:45:09 +03:00
pci_read_config_byte ( dev , ICH_ACPI_CNTL , & enable ) ;
2013-04-12 16:44:15 +04:00
if ( enable & ICH6_ACPI_EN )
quirk_io_region ( dev , ICH_PMBASE , 128 , PCI_BRIDGE_RESOURCES ,
" ICH6 ACPI/GPIO/TCO " ) ;
2005-09-28 01:54:51 +04:00
2011-02-28 12:45:09 +03:00
pci_read_config_byte ( dev , ICH6_GPIO_CNTL , & enable ) ;
2013-04-12 16:44:15 +04:00
if ( enable & ICH6_GPIO_EN )
quirk_io_region ( dev , ICH6_GPIOBASE , 64 , PCI_BRIDGE_RESOURCES + 1 ,
" ICH6 GPIO " ) ;
2005-09-28 01:54:51 +04:00
}
2008-12-06 21:10:10 +03:00
2018-05-02 16:53:19 +03:00
static void ich6_lpc_generic_decode ( struct pci_dev * dev , unsigned reg ,
const char * name , int dynsize )
2008-12-06 21:10:10 +03:00
{
u32 val ;
u32 size , base ;
pci_read_config_dword ( dev , reg , & val ) ;
/* Enabled? */
if ( ! ( val & 1 ) )
return ;
base = val & 0xfffc ;
if ( dynsize ) {
/*
* This is not correct . It is 16 , 32 or 64 bytes depending on
* register D31 : F0 : ADh bits 5 : 4.
*
* But this gets us at least _part_ of it .
*/
size = 16 ;
} else {
size = 128 ;
}
base & = ~ ( size - 1 ) ;
2018-05-02 16:53:19 +03:00
/*
* Just print it out for now . We should reserve it after more
* debugging .
*/
2018-01-18 21:55:24 +03:00
pci_info ( dev , " %s PIO at %04x-%04x \n " , name , base , base + size - 1 ) ;
2008-12-06 21:10:10 +03:00
}
2012-11-22 00:35:00 +04:00
static void quirk_ich6_lpc ( struct pci_dev * dev )
2008-12-06 21:10:10 +03:00
{
/* Shared ACPI/GPIO decode with all ICH6+ */
ich6_lpc_acpi_gpio ( dev ) ;
/* ICH6-specific generic IO decode */
ich6_lpc_generic_decode ( dev , 0x84 , " LPC Generic IO decode 1 " , 0 ) ;
ich6_lpc_generic_decode ( dev , 0x88 , " LPC Generic IO decode 2 " , 1 ) ;
}
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_ICH6_0 , quirk_ich6_lpc ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_ICH6_1 , quirk_ich6_lpc ) ;
2018-05-02 16:53:19 +03:00
static void ich7_lpc_generic_decode ( struct pci_dev * dev , unsigned reg ,
const char * name )
2008-12-06 21:10:10 +03:00
{
u32 val ;
u32 mask , base ;
pci_read_config_dword ( dev , reg , & val ) ;
/* Enabled? */
if ( ! ( val & 1 ) )
return ;
2018-05-02 16:53:19 +03:00
/* IO base in bits 15:2, mask in bits 23:18, both are dword-based */
2008-12-06 21:10:10 +03:00
base = val & 0xfffc ;
mask = ( val > > 16 ) & 0xfc ;
mask | = 3 ;
2018-05-02 16:53:19 +03:00
/*
* Just print it out for now . We should reserve it after more
* debugging .
*/
2018-01-18 21:55:24 +03:00
pci_info ( dev , " %s PIO at %04x (mask %04x) \n " , name , base , mask ) ;
2008-12-06 21:10:10 +03:00
}
/* ICH7-10 has the same common LPC generic IO decode registers */
2012-11-22 00:35:00 +04:00
static void quirk_ich7_lpc ( struct pci_dev * dev )
2008-12-06 21:10:10 +03:00
{
2011-04-15 12:03:53 +04:00
/* We share the common ACPI/GPIO decode with ICH6 */
2008-12-06 21:10:10 +03:00
ich6_lpc_acpi_gpio ( dev ) ;
/* And have 4 ICH7+ generic decodes */
ich7_lpc_generic_decode ( dev , 0x84 , " ICH7 LPC Generic IO decode 1 " ) ;
ich7_lpc_generic_decode ( dev , 0x88 , " ICH7 LPC Generic IO decode 2 " ) ;
ich7_lpc_generic_decode ( dev , 0x8c , " ICH7 LPC Generic IO decode 3 " ) ;
ich7_lpc_generic_decode ( dev , 0x90 , " ICH7 LPC Generic IO decode 4 " ) ;
}
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_ICH7_0 , quirk_ich7_lpc ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_ICH7_1 , quirk_ich7_lpc ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_ICH7_31 , quirk_ich7_lpc ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_ICH8_0 , quirk_ich7_lpc ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_ICH8_2 , quirk_ich7_lpc ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_ICH8_3 , quirk_ich7_lpc ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_ICH8_1 , quirk_ich7_lpc ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_ICH8_4 , quirk_ich7_lpc ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_ICH9_2 , quirk_ich7_lpc ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_ICH9_4 , quirk_ich7_lpc ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_ICH9_7 , quirk_ich7_lpc ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_ICH9_8 , quirk_ich7_lpc ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_ICH10_1 , quirk_ich7_lpc ) ;
2005-09-28 01:54:51 +04:00
2005-04-17 02:20:36 +04:00
/*
* VIA ACPI : One IO region pointed to by longword at
* 0x48 or 0x20 ( 256 bytes of ACPI registers )
*/
2012-11-22 00:35:00 +04:00
static void quirk_vt82c586_acpi ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
2013-04-12 16:44:15 +04:00
if ( dev - > revision & 0x10 )
quirk_io_region ( dev , 0x48 , 256 , PCI_BRIDGE_RESOURCES ,
" vt82c586 ACPI " ) ;
2005-04-17 02:20:36 +04:00
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_82C586_3 , quirk_vt82c586_acpi ) ;
2005-04-17 02:20:36 +04:00
/*
* VIA VT82C686 ACPI : Three IO region pointed to by ( long ) words at
* 0x48 ( 256 bytes of ACPI registers )
* 0x70 ( 128 bytes of hardware monitoring register )
* 0x90 ( 16 bytes of SMB registers )
*/
2012-11-22 00:35:00 +04:00
static void quirk_vt82c686_acpi ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
quirk_vt82c586_acpi ( dev ) ;
2013-04-12 16:44:15 +04:00
quirk_io_region ( dev , 0x70 , 128 , PCI_BRIDGE_RESOURCES + 1 ,
" vt82c686 HW-mon " ) ;
2005-04-17 02:20:36 +04:00
2013-04-12 16:44:15 +04:00
quirk_io_region ( dev , 0x90 , 16 , PCI_BRIDGE_RESOURCES + 2 , " vt82c686 SMB " ) ;
2005-04-17 02:20:36 +04:00
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_82C686_4 , quirk_vt82c686_acpi ) ;
2005-04-17 02:20:36 +04:00
2005-08-08 12:55:54 +04:00
/*
* VIA VT8235 ISA Bridge : Two IO regions pointed to by words at
* 0x88 ( 128 bytes of power management registers )
* 0xd0 ( 16 bytes of SMB registers )
*/
2012-11-22 00:35:00 +04:00
static void quirk_vt8235_acpi ( struct pci_dev * dev )
2005-08-08 12:55:54 +04:00
{
2013-04-12 16:44:15 +04:00
quirk_io_region ( dev , 0x88 , 128 , PCI_BRIDGE_RESOURCES , " vt8235 PM " ) ;
quirk_io_region ( dev , 0xd0 , 16 , PCI_BRIDGE_RESOURCES + 1 , " vt8235 SMB " ) ;
2005-08-08 12:55:54 +04:00
}
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_8235 , quirk_vt8235_acpi ) ;
2009-10-06 18:19:45 +04:00
/*
2018-05-02 16:53:19 +03:00
* TI XIO2000a PCIe - PCI Bridge erroneously reports it supports fast
* back - to - back : Disable fast back - to - back on the secondary bus segment
2009-10-06 18:19:45 +04:00
*/
2012-11-22 00:35:00 +04:00
static void quirk_xio2000a ( struct pci_dev * dev )
2009-10-06 18:19:45 +04:00
{
struct pci_dev * pdev ;
u16 command ;
2018-01-18 21:55:24 +03:00
pci_warn ( dev , " TI XIO2000a quirk detected; secondary bus fast back-to-back transfers disabled \n " ) ;
2009-10-06 18:19:45 +04:00
list_for_each_entry ( pdev , & dev - > subordinate - > devices , bus_list ) {
pci_read_config_word ( pdev , PCI_COMMAND , & command ) ;
if ( command & PCI_COMMAND_FAST_BACK )
pci_write_config_word ( pdev , PCI_COMMAND , command & ~ PCI_COMMAND_FAST_BACK ) ;
}
}
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_TI , PCI_DEVICE_ID_TI_XIO2000A ,
quirk_xio2000a ) ;
2005-04-17 02:20:36 +04:00
2013-11-14 22:28:18 +04:00
# ifdef CONFIG_X86_IO_APIC
2005-04-17 02:20:36 +04:00
# include <asm/io_apic.h>
/*
* VIA 686 A / B : If an IO - APIC is active , we need to route all on - chip
* devices to the external APIC .
*
2018-05-02 16:53:19 +03:00
* TODO : When we have device - specific interrupt routers , this code will go
* away from quirks .
2005-04-17 02:20:36 +04:00
*/
2006-12-05 02:14:45 +03:00
static void quirk_via_ioapic ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
u8 tmp ;
2013-11-14 22:28:18 +04:00
2005-04-17 02:20:36 +04:00
if ( nr_ioapics < 1 )
tmp = 0 ; /* nothing routed to external APIC */
else
tmp = 0x1f ; /* all known bits (4-0) routed to external APIC */
2013-11-14 22:28:18 +04:00
2018-01-18 21:55:24 +03:00
pci_info ( dev , " %sbling VIA external APIC routing \n " ,
2005-04-17 02:20:36 +04:00
tmp = = 0 ? " Disa " : " Ena " ) ;
/* Offset 0x58: External APIC IRQ output control */
2014-04-19 04:13:49 +04:00
pci_write_config_byte ( dev , 0x58 , tmp ) ;
2005-04-17 02:20:36 +04:00
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_82C686 , quirk_via_ioapic ) ;
2008-05-15 23:51:31 +04:00
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_82C686 , quirk_via_ioapic ) ;
2005-04-17 02:20:36 +04:00
2005-09-04 02:56:33 +04:00
/*
2013-11-14 22:28:18 +04:00
* VIA 8237 : Some BIOSes don ' t set the ' Bypass APIC De - Assert Message ' Bit .
2005-09-04 02:56:33 +04:00
* This leads to doubled level interrupt rates .
* Set this bit to get rid of cycle wastage .
* Otherwise uncritical .
*/
2006-12-05 02:14:45 +03:00
static void quirk_via_vt8237_bypass_apic_deassert ( struct pci_dev * dev )
2005-09-04 02:56:33 +04:00
{
u8 misc_control2 ;
# define BYPASS_APIC_DEASSERT 8
pci_read_config_byte ( dev , 0x5B , & misc_control2 ) ;
if ( ! ( misc_control2 & BYPASS_APIC_DEASSERT ) ) {
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Bypassing VIA 8237 APIC De-Assert Message \n " ) ;
2005-09-04 02:56:33 +04:00
pci_write_config_byte ( dev , 0x5B , misc_control2 | BYPASS_APIC_DEASSERT ) ;
}
}
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_8237 , quirk_via_vt8237_bypass_apic_deassert ) ;
2008-05-15 23:51:31 +04:00
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_8237 , quirk_via_vt8237_bypass_apic_deassert ) ;
2005-09-04 02:56:33 +04:00
2005-04-17 02:20:36 +04:00
/*
2018-05-02 16:53:19 +03:00
* The AMD IO - APIC can hang the box when an APIC IRQ is masked .
2005-04-17 02:20:36 +04:00
* We check all revs > = B0 ( yet not in the pre production ! ) as the bug
* is currently marked NoFix
*
* We have multiple reports of hangs with this chipset that went away with
2006-10-01 10:27:03 +04:00
* noapic specified . For the moment we assume it ' s the erratum . We may be wrong
2018-05-02 16:53:19 +03:00
* of course . However the advice is demonstrably good even if so .
2005-04-17 02:20:36 +04:00
*/
2012-11-22 00:35:00 +04:00
static void quirk_amd_ioapic ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
2007-06-09 02:46:36 +04:00
if ( dev - > revision > = 0x02 ) {
2018-01-18 21:55:24 +03:00
pci_warn ( dev , " I/O APIC: AMD Erratum #22 may be present. In the event of instability try \n " ) ;
pci_warn ( dev , " : booting with the \" noapic \" option \n " ) ;
2005-04-17 02:20:36 +04:00
}
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_AMD , PCI_DEVICE_ID_AMD_VIPER_7410 , quirk_amd_ioapic ) ;
2005-04-17 02:20:36 +04:00
# endif /* CONFIG_X86_IO_APIC */
2016-09-05 12:12:57 +03:00
# if defined(CONFIG_ARM64) && defined(CONFIG_PCI_ATS)
2016-08-24 02:27:14 +03:00
static void quirk_cavium_sriov_rnm_link ( struct pci_dev * dev )
{
2018-05-02 16:53:19 +03:00
/* Fix for improper SR-IOV configuration on Cavium cn88xx RNM device */
2016-08-24 02:27:14 +03:00
if ( dev - > subsystem_device = = 0xa118 )
dev - > sriov - > link = dev - > devfn ;
}
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_CAVIUM , 0xa018 , quirk_cavium_sriov_rnm_link ) ;
# endif
2007-05-15 15:59:13 +04:00
/*
* Some settings of MMRBC can lead to data corruption so block changes .
* See AMD 8131 HyperTransport PCI - X Tunnel Revision Guide
*/
2012-11-22 00:35:00 +04:00
static void quirk_amd_8131_mmrbc ( struct pci_dev * dev )
2007-05-15 15:59:13 +04:00
{
2007-08-28 03:17:47 +04:00
if ( dev - > subordinate & & dev - > revision < = 0x12 ) {
2018-01-18 21:55:24 +03:00
pci_info ( dev , " AMD8131 rev %x detected; disabling PCI-X MMRBC \n " ,
2014-04-19 04:13:50 +04:00
dev - > revision ) ;
2007-05-15 15:59:13 +04:00
dev - > subordinate - > bus_flags | = PCI_BUS_FLAGS_NO_MMRBC ;
}
}
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_AMD , PCI_DEVICE_ID_AMD_8131_BRIDGE , quirk_amd_8131_mmrbc ) ;
2005-04-17 02:20:36 +04:00
/*
2018-05-02 16:53:19 +03:00
* FIXME : it is questionable that quirk_via_acpi ( ) is needed . It shows up
* as an ISA bridge , and does not support the PCI_INTERRUPT_LINE register
* at all . Therefore it seems like setting the pci_dev ' s IRQ to the value
* of the ACPI SCI interrupt is only done for convenience .
2005-04-17 02:20:36 +04:00
* - jgarzik
*/
2012-11-22 00:35:00 +04:00
static void quirk_via_acpi ( struct pci_dev * d )
2005-04-17 02:20:36 +04:00
{
u8 irq ;
2018-05-02 16:53:19 +03:00
/* VIA ACPI device: SCI IRQ line in PCI config byte 0x42 */
2005-04-17 02:20:36 +04:00
pci_read_config_byte ( d , 0x42 , & irq ) ;
irq & = 0xf ;
if ( irq & & ( irq ! = 2 ) )
d - > irq = irq ;
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_82C586_3 , quirk_via_acpi ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_82C686_4 , quirk_via_acpi ) ;
2005-04-17 02:20:36 +04:00
2018-05-02 16:53:19 +03:00
/* VIA bridges which have VLink */
2007-01-31 01:36:09 +03:00
static int via_vlink_dev_lo = - 1 , via_vlink_dev_hi = 18 ;
static void quirk_via_bridge ( struct pci_dev * dev )
{
/* See what bridge we have and find the device ranges */
switch ( dev - > device ) {
case PCI_DEVICE_ID_VIA_82C686 :
2018-05-02 16:53:19 +03:00
/*
* The VT82C686 is special ; it attaches to PCI and can have
* any device number . All its subdevices are functions of
* that single device .
*/
2007-02-01 10:48:12 +03:00
via_vlink_dev_lo = PCI_SLOT ( dev - > devfn ) ;
via_vlink_dev_hi = PCI_SLOT ( dev - > devfn ) ;
2007-01-31 01:36:09 +03:00
break ;
case PCI_DEVICE_ID_VIA_8237 :
case PCI_DEVICE_ID_VIA_8237A :
via_vlink_dev_lo = 15 ;
break ;
case PCI_DEVICE_ID_VIA_8235 :
via_vlink_dev_lo = 16 ;
break ;
case PCI_DEVICE_ID_VIA_8231 :
case PCI_DEVICE_ID_VIA_8233_0 :
case PCI_DEVICE_ID_VIA_8233A :
case PCI_DEVICE_ID_VIA_8233C_0 :
via_vlink_dev_lo = 17 ;
break ;
}
}
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_82C686 , quirk_via_bridge ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_8231 , quirk_via_bridge ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_8233_0 , quirk_via_bridge ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_8233A , quirk_via_bridge ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_8233C_0 , quirk_via_bridge ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_8235 , quirk_via_bridge ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_8237 , quirk_via_bridge ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_8237A , quirk_via_bridge ) ;
2006-09-26 03:52:19 +04:00
2018-05-02 16:53:19 +03:00
/*
* quirk_via_vlink - VIA VLink IRQ number update
* @ dev : PCI device
2006-12-05 02:14:45 +03:00
*
2018-05-02 16:53:19 +03:00
* If the device we are dealing with is on a PIC IRQ we need to ensure that
* the IRQ line register which usually is not relevant for PCI cards , is
* actually written so that interrupts get sent to the right place .
*
* We only do this on systems where a VIA south bridge was detected , and
* only for VIA devices on the motherboard ( see quirk_via_bridge above ) .
2006-12-05 02:14:45 +03:00
*/
static void quirk_via_vlink ( struct pci_dev * dev )
2005-05-27 12:21:50 +04:00
{
u8 irq , new_irq ;
2007-01-31 01:36:09 +03:00
/* Check if we have VLink at all */
if ( via_vlink_dev_lo = = - 1 )
2006-09-26 03:52:19 +04:00
return ;
new_irq = dev - > irq ;
/* Don't quirk interrupts outside the legacy IRQ range */
if ( ! new_irq | | new_irq > 15 )
return ;
2006-12-05 02:14:45 +03:00
/* Internal device ? */
2007-01-31 01:36:09 +03:00
if ( dev - > bus - > number ! = 0 | | PCI_SLOT ( dev - > devfn ) > via_vlink_dev_hi | |
PCI_SLOT ( dev - > devfn ) < via_vlink_dev_lo )
2006-12-05 02:14:45 +03:00
return ;
2018-05-02 16:53:19 +03:00
/*
* This is an internal VLink device on a PIC interrupt . The BIOS
* ought to have set this but may not have , so we redo it .
*/
2005-05-27 12:21:50 +04:00
pci_read_config_byte ( dev , PCI_INTERRUPT_LINE , & irq ) ;
if ( new_irq ! = irq ) {
2018-01-18 21:55:24 +03:00
pci_info ( dev , " VIA VLink IRQ fixup, from %d to %d \n " ,
2007-12-18 00:09:39 +03:00
irq , new_irq ) ;
2005-05-27 12:21:50 +04:00
udelay ( 15 ) ; /* unknown if delay really needed */
pci_write_config_byte ( dev , PCI_INTERRUPT_LINE , new_irq ) ;
}
}
2006-12-05 02:14:45 +03:00
DECLARE_PCI_FIXUP_ENABLE ( PCI_VENDOR_ID_VIA , PCI_ANY_ID , quirk_via_vlink ) ;
2005-05-27 12:21:50 +04:00
2005-04-17 02:20:36 +04:00
/*
2018-05-02 16:53:19 +03:00
* VIA VT82C598 has its device ID settable and many BIOSes set it to the ID
* of VT82C597 for backward compatibility . We need to switch it off to be
* able to recognize the real type of the chip .
2005-04-17 02:20:36 +04:00
*/
2012-11-22 00:35:00 +04:00
static void quirk_vt82c598_id ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
pci_write_config_byte ( dev , 0xfc , 0 ) ;
pci_read_config_word ( dev , PCI_DEVICE_ID , & dev - > device ) ;
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_82C597_0 , quirk_vt82c598_id ) ;
2005-04-17 02:20:36 +04:00
/*
2018-05-02 16:53:19 +03:00
* CardBus controllers have a legacy base address that enables them to
* respond as i82365 pcmcia controllers . We don ' t want them to do this
* even if the Linux CardBus driver is not loaded , because the Linux i82365
* driver does not ( and should not ) handle CardBus .
2005-04-17 02:20:36 +04:00
*/
2006-12-05 02:14:45 +03:00
static void quirk_cardbus_legacy ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
pci_write_config_dword ( dev , PCI_CB_LEGACY_MODE_BASE , 0 ) ;
}
2012-02-24 11:46:54 +04:00
DECLARE_PCI_FIXUP_CLASS_FINAL ( PCI_ANY_ID , PCI_ANY_ID ,
PCI_CLASS_BRIDGE_CARDBUS , 8 , quirk_cardbus_legacy ) ;
DECLARE_PCI_FIXUP_CLASS_RESUME_EARLY ( PCI_ANY_ID , PCI_ANY_ID ,
PCI_CLASS_BRIDGE_CARDBUS , 8 , quirk_cardbus_legacy ) ;
2005-04-17 02:20:36 +04:00
/*
2018-05-02 16:53:19 +03:00
* Following the PCI ordering rules is optional on the AMD762 . I ' m not sure
* what the designers were smoking but let ' s not inhale . . .
2005-04-17 02:20:36 +04:00
*
2018-05-02 16:53:19 +03:00
* To be fair to AMD , it follows the spec by default , it ' s BIOS people who
* turn it off !
2005-04-17 02:20:36 +04:00
*/
2006-12-05 02:14:45 +03:00
static void quirk_amd_ordering ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
u32 pcic ;
pci_read_config_dword ( dev , 0x4C , & pcic ) ;
2014-04-19 04:13:49 +04:00
if ( ( pcic & 6 ) ! = 6 ) {
2005-04-17 02:20:36 +04:00
pcic | = 6 ;
2018-01-18 21:55:24 +03:00
pci_warn ( dev , " BIOS failed to enable PCI standards compliance; fixing this error \n " ) ;
2005-04-17 02:20:36 +04:00
pci_write_config_dword ( dev , 0x4C , pcic ) ;
pci_read_config_dword ( dev , 0x84 , & pcic ) ;
2014-04-19 04:13:49 +04:00
pcic | = ( 1 < < 23 ) ; /* Required in this mode */
2005-04-17 02:20:36 +04:00
pci_write_config_dword ( dev , 0x84 , pcic ) ;
}
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_AMD , PCI_DEVICE_ID_AMD_FE_GATE_700C , quirk_amd_ordering ) ;
2008-05-15 23:51:31 +04:00
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_AMD , PCI_DEVICE_ID_AMD_FE_GATE_700C , quirk_amd_ordering ) ;
2005-04-17 02:20:36 +04:00
/*
2018-05-02 16:53:19 +03:00
* DreamWorks - provided workaround for Dunord I - 3000 problem
2005-04-17 02:20:36 +04:00
*
2018-05-02 16:53:19 +03:00
* This card decodes and responds to addresses not apparently assigned to
* it . We force a larger allocation to ensure that nothing gets put too
* close to it .
2005-04-17 02:20:36 +04:00
*/
2012-11-22 00:35:00 +04:00
static void quirk_dunord ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
2014-04-19 04:13:49 +04:00
struct resource * r = & dev - > resource [ 1 ] ;
2014-02-26 22:25:58 +04:00
r - > flags | = IORESOURCE_UNSET ;
2005-04-17 02:20:36 +04:00
r - > start = 0 ;
r - > end = 0xffffff ;
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_DUNORD , PCI_DEVICE_ID_DUNORD_I3000 , quirk_dunord ) ;
2005-04-17 02:20:36 +04:00
/*
2018-05-02 16:53:19 +03:00
* i82380FB mobile docking controller : its PCI - to - PCI bridge is subtractive
* decoding ( transparent ) , and does indicate this in the ProgIf .
* Unfortunately , the ProgIf value is wrong - 0x80 instead of 0x01 .
2005-04-17 02:20:36 +04:00
*/
2012-11-22 00:35:00 +04:00
static void quirk_transparent_bridge ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
dev - > transparent = 1 ;
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82380FB , quirk_transparent_bridge ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_TOSHIBA , 0x605 , quirk_transparent_bridge ) ;
2005-04-17 02:20:36 +04:00
/*
2018-05-02 16:53:19 +03:00
* Common misconfiguration of the MediaGX / Geode PCI master that will reduce
* PCI bandwidth from 70 MB / s to 25 MB / s . See the GXM / GXLV / GX1 datasheets
* found at http : //www.national.com/analog for info on what these bits do.
* < christer @ weinigel . se >
2005-04-17 02:20:36 +04:00
*/
2006-12-05 02:14:45 +03:00
static void quirk_mediagx_master ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
u8 reg ;
2014-04-19 04:13:49 +04:00
2005-04-17 02:20:36 +04:00
pci_read_config_byte ( dev , 0x41 , & reg ) ;
if ( reg & 2 ) {
reg & = ~ 2 ;
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Fixup for MediaGX/Geode Slave Disconnect Boundary (0x41=0x%02x) \n " ,
2014-04-19 04:13:50 +04:00
reg ) ;
2014-04-19 04:13:49 +04:00
pci_write_config_byte ( dev , 0x41 , reg ) ;
2005-04-17 02:20:36 +04:00
}
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_CYRIX , PCI_DEVICE_ID_CYRIX_PCI_MASTER , quirk_mediagx_master ) ;
DECLARE_PCI_FIXUP_RESUME ( PCI_VENDOR_ID_CYRIX , PCI_DEVICE_ID_CYRIX_PCI_MASTER , quirk_mediagx_master ) ;
2005-04-17 02:20:36 +04:00
/*
2018-05-02 16:53:19 +03:00
* Ensure C0 rev restreaming is off . This is normally done by the BIOS but
* in the odd case it is not the results are corruption hence the presence
* of a Linux check .
2005-04-17 02:20:36 +04:00
*/
2006-12-05 02:14:45 +03:00
static void quirk_disable_pxb ( struct pci_dev * pdev )
2005-04-17 02:20:36 +04:00
{
u16 config ;
2013-11-14 22:28:18 +04:00
2007-06-09 02:46:36 +04:00
if ( pdev - > revision ! = 0x04 ) /* Only C0 requires this */
2005-04-17 02:20:36 +04:00
return ;
pci_read_config_word ( pdev , 0x40 , & config ) ;
if ( config & ( 1 < < 6 ) ) {
config & = ~ ( 1 < < 6 ) ;
pci_write_config_word ( pdev , 0x40 , config ) ;
2018-01-18 21:55:24 +03:00
pci_info ( pdev , " C0 revision 450NX. Disabling PCI restreaming \n " ) ;
2005-04-17 02:20:36 +04:00
}
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82454NX , quirk_disable_pxb ) ;
2008-05-15 23:51:31 +04:00
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82454NX , quirk_disable_pxb ) ;
2005-04-17 02:20:36 +04:00
2012-07-10 01:36:14 +04:00
static void quirk_amd_ide_mode ( struct pci_dev * pdev )
2006-12-20 00:11:37 +03:00
{
2009-10-13 07:14:00 +04:00
/* set SBX00/Hudson-2 SATA in IDE mode to AHCI mode */
2008-02-02 08:56:56 +03:00
u8 tmp ;
2006-12-20 00:11:37 +03:00
2008-02-02 08:56:56 +03:00
pci_read_config_byte ( pdev , PCI_CLASS_DEVICE , & tmp ) ;
if ( tmp = = 0x01 ) {
2006-12-20 00:11:37 +03:00
pci_read_config_byte ( pdev , 0x40 , & tmp ) ;
pci_write_config_byte ( pdev , 0x40 , tmp | 1 ) ;
pci_write_config_byte ( pdev , 0x9 , 1 ) ;
pci_write_config_byte ( pdev , 0xa , 6 ) ;
pci_write_config_byte ( pdev , 0x40 , tmp ) ;
2007-01-09 13:32:51 +03:00
pdev - > class = PCI_CLASS_STORAGE_SATA_AHCI ;
2018-01-18 21:55:24 +03:00
pci_info ( pdev , " set SATA to AHCI mode \n " ) ;
2006-12-20 00:11:37 +03:00
}
}
2008-02-02 08:56:56 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_ATI , PCI_DEVICE_ID_ATI_IXP600_SATA , quirk_amd_ide_mode ) ;
2008-05-15 23:51:31 +04:00
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_ATI , PCI_DEVICE_ID_ATI_IXP600_SATA , quirk_amd_ide_mode ) ;
2008-02-02 08:56:56 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_ATI , PCI_DEVICE_ID_ATI_IXP700_SATA , quirk_amd_ide_mode ) ;
2008-05-15 23:51:31 +04:00
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_ATI , PCI_DEVICE_ID_ATI_IXP700_SATA , quirk_amd_ide_mode ) ;
2009-10-13 07:14:00 +04:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_AMD , PCI_DEVICE_ID_AMD_HUDSON2_SATA_IDE , quirk_amd_ide_mode ) ;
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_AMD , PCI_DEVICE_ID_AMD_HUDSON2_SATA_IDE , quirk_amd_ide_mode ) ;
2013-06-03 14:24:10 +04:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_AMD , 0x7900 , quirk_amd_ide_mode ) ;
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_AMD , 0x7900 , quirk_amd_ide_mode ) ;
2006-12-20 00:11:37 +03:00
2018-05-02 16:53:19 +03:00
/* Serverworks CSB5 IDE does not fully support native mode */
2012-11-22 00:35:00 +04:00
static void quirk_svwks_csb5ide ( struct pci_dev * pdev )
2005-04-17 02:20:36 +04:00
{
u8 prog ;
pci_read_config_byte ( pdev , PCI_CLASS_PROG , & prog ) ;
if ( prog & 5 ) {
prog & = ~ 5 ;
pdev - > class & = ~ 5 ;
pci_write_config_byte ( pdev , PCI_CLASS_PROG , prog ) ;
2006-10-04 03:41:26 +04:00
/* PCI layer will sort out resources */
2005-04-17 02:20:36 +04:00
}
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_SERVERWORKS , PCI_DEVICE_ID_SERVERWORKS_CSB5IDE , quirk_svwks_csb5ide ) ;
2005-04-17 02:20:36 +04:00
2018-05-02 16:53:19 +03:00
/* Intel 82801CAM ICH3-M datasheet says IDE modes must be the same */
2012-11-22 00:35:00 +04:00
static void quirk_ide_samemode ( struct pci_dev * pdev )
2005-04-17 02:20:36 +04:00
{
u8 prog ;
pci_read_config_byte ( pdev , PCI_CLASS_PROG , & prog ) ;
if ( ( ( prog & 1 ) & & ! ( prog & 4 ) ) | | ( ( prog & 4 ) & & ! ( prog & 1 ) ) ) {
2018-01-18 21:55:24 +03:00
pci_info ( pdev , " IDE mode mismatch; forcing legacy mode \n " ) ;
2005-04-17 02:20:36 +04:00
prog & = ~ 5 ;
pdev - > class & = ~ 5 ;
pci_write_config_byte ( pdev , PCI_CLASS_PROG , prog ) ;
}
}
2006-10-04 03:41:26 +04:00
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82801CA_10 , quirk_ide_samemode ) ;
2005-04-17 02:20:36 +04:00
2018-05-02 16:53:19 +03:00
/* Some ATA devices break if put into D3 */
2012-11-22 00:35:00 +04:00
static void quirk_no_ata_d3 ( struct pci_dev * pdev )
2008-07-24 20:18:38 +04:00
{
2012-02-24 11:46:55 +04:00
pdev - > dev_flags | = PCI_DEV_FLAGS_NO_D3 ;
2008-07-24 20:18:38 +04:00
}
2012-02-24 11:46:55 +04:00
/* Quirk the legacy ATA devices only. The AHCI ones are ok */
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_SERVERWORKS , PCI_ANY_ID ,
PCI_CLASS_STORAGE_IDE , 8 , quirk_no_ata_d3 ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_ATI , PCI_ANY_ID ,
PCI_CLASS_STORAGE_IDE , 8 , quirk_no_ata_d3 ) ;
2009-06-24 19:02:27 +04:00
/* ALi loses some register settings that we cannot then restore */
2012-02-24 11:46:55 +04:00
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_AL , PCI_ANY_ID ,
PCI_CLASS_STORAGE_IDE , 8 , quirk_no_ata_d3 ) ;
2009-06-24 19:02:27 +04:00
/* VIA comes back fine but we need to keep it alive or ACPI GTM failures
occur when mode detecting */
2012-02-24 11:46:55 +04:00
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_VIA , PCI_ANY_ID ,
PCI_CLASS_STORAGE_IDE , 8 , quirk_no_ata_d3 ) ;
2008-07-24 20:18:38 +04:00
2018-05-02 16:53:19 +03:00
/*
* This was originally an Alpha - specific thing , but it really fits here .
2005-04-17 02:20:36 +04:00
* The i82375 PCI / EISA bridge appears as non - classified . Fix that .
*/
2012-11-22 00:35:00 +04:00
static void quirk_eisa_bridge ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
dev - > class = PCI_CLASS_BRIDGE_EISA < < 8 ;
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82375 , quirk_eisa_bridge ) ;
2005-04-17 02:20:36 +04:00
/*
* On ASUS P4B boards , the SMBus PCI Device within the ICH2 / 4 southbridge
* is not activated . The myth is that Asus said that they do not want the
* users to be irritated by just another PCI Device in the Win98 device
2013-11-14 22:28:18 +04:00
* manager . ( see the file prog / hotplug / README . p4b in the lm_sensors
2005-04-17 02:20:36 +04:00
* package 2.7 .0 for details )
*
2013-11-14 22:28:18 +04:00
* The SMBus PCI Device can be activated by setting a bit in the ICH LPC
* bridge . Unfortunately , this device has no subvendor / subdevice ID . So it
2007-08-23 23:22:04 +04:00
* becomes necessary to do this tweak in two steps - - the chosen trigger
* is either the Host bridge ( preferred ) or on - board VGA controller .
2007-03-24 18:56:44 +03:00
*
* Note that we used to unhide the SMBus that way on Toshiba laptops
* ( Satellite A40 and Tecra M2 ) but then found that the thermal management
* was done by SMM code , which could cause unsynchronized concurrent
* accesses to the SMBus registers , with potentially bad effects . Thus you
* should be very careful when adding new entries : if SMM is accessing the
* Intel SMBus , this is a very good reason to leave it hidden .
2008-03-29 00:16:04 +03:00
*
* Likewise , many recent laptops use ACPI for thermal management . If the
* ACPI DSDT code accesses the SMBus , then Linux should not access it
* natively , and keeping the SMBus hidden is the right thing to do . If you
* are about to add an entry in the table below , please first disassemble
* the DSDT and double - check that there is no code accessing the SMBus .
2005-04-17 02:20:36 +04:00
*/
2007-01-11 03:52:44 +03:00
static int asus_hides_smbus ;
2005-04-17 02:20:36 +04:00
2012-11-22 00:35:00 +04:00
static void asus_hides_smbus_hostbridge ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
if ( unlikely ( dev - > subsystem_vendor = = PCI_VENDOR_ID_ASUSTEK ) ) {
if ( dev - > device = = PCI_DEVICE_ID_INTEL_82845_HB )
2014-04-19 04:13:49 +04:00
switch ( dev - > subsystem_device ) {
2005-06-29 19:04:06 +04:00
case 0x8025 : /* P4B-LX */
2005-04-17 02:20:36 +04:00
case 0x8070 : /* P4B */
case 0x8088 : /* P4B533 */
case 0x1626 : /* L3C notebook */
asus_hides_smbus = 1 ;
}
2007-01-05 13:23:15 +03:00
else if ( dev - > device = = PCI_DEVICE_ID_INTEL_82845G_HB )
2014-04-19 04:13:49 +04:00
switch ( dev - > subsystem_device ) {
2005-04-17 02:20:36 +04:00
case 0x80b1 : /* P4GE-V */
case 0x80b2 : /* P4PE */
case 0x8093 : /* P4B533-V */
asus_hides_smbus = 1 ;
}
2007-01-05 13:23:15 +03:00
else if ( dev - > device = = PCI_DEVICE_ID_INTEL_82850_HB )
2014-04-19 04:13:49 +04:00
switch ( dev - > subsystem_device ) {
2005-04-17 02:20:36 +04:00
case 0x8030 : /* P4T533 */
asus_hides_smbus = 1 ;
}
2007-01-05 13:23:15 +03:00
else if ( dev - > device = = PCI_DEVICE_ID_INTEL_7205_0 )
2005-04-17 02:20:36 +04:00
switch ( dev - > subsystem_device ) {
case 0x8070 : /* P4G8X Deluxe */
asus_hides_smbus = 1 ;
}
2007-01-05 13:23:15 +03:00
else if ( dev - > device = = PCI_DEVICE_ID_INTEL_E7501_MCH )
2006-07-31 10:53:15 +04:00
switch ( dev - > subsystem_device ) {
case 0x80c9 : /* PU-DLS */
asus_hides_smbus = 1 ;
}
2007-01-05 13:23:15 +03:00
else if ( dev - > device = = PCI_DEVICE_ID_INTEL_82855GM_HB )
2005-04-17 02:20:36 +04:00
switch ( dev - > subsystem_device ) {
case 0x1751 : /* M2N notebook */
case 0x1821 : /* M5N notebook */
2009-05-12 14:05:23 +04:00
case 0x1897 : /* A6L notebook */
2005-04-17 02:20:36 +04:00
asus_hides_smbus = 1 ;
}
2007-01-05 13:23:15 +03:00
else if ( dev - > device = = PCI_DEVICE_ID_INTEL_82855PM_HB )
2005-04-17 02:20:36 +04:00
switch ( dev - > subsystem_device ) {
case 0x184b : /* W1N notebook */
case 0x186a : /* M6Ne notebook */
asus_hides_smbus = 1 ;
}
2007-01-05 13:23:15 +03:00
else if ( dev - > device = = PCI_DEVICE_ID_INTEL_82865_HB )
2007-01-05 11:17:56 +03:00
switch ( dev - > subsystem_device ) {
case 0x80f2 : /* P4P800-X */
asus_hides_smbus = 1 ;
}
2007-01-05 13:23:15 +03:00
else if ( dev - > device = = PCI_DEVICE_ID_INTEL_82915GM_HB )
2005-09-29 12:35:41 +04:00
switch ( dev - > subsystem_device ) {
case 0x1882 : /* M6V notebook */
2006-04-01 18:46:35 +04:00
case 0x1977 : /* A6VA notebook */
2005-09-29 12:35:41 +04:00
asus_hides_smbus = 1 ;
}
2005-04-17 02:20:36 +04:00
} else if ( unlikely ( dev - > subsystem_vendor = = PCI_VENDOR_ID_HP ) ) {
if ( dev - > device = = PCI_DEVICE_ID_INTEL_82855PM_HB )
2014-04-19 04:13:49 +04:00
switch ( dev - > subsystem_device ) {
2005-04-17 02:20:36 +04:00
case 0x088C : /* HP Compaq nc8000 */
case 0x0890 : /* HP Compaq nc6000 */
asus_hides_smbus = 1 ;
}
2007-01-05 13:23:15 +03:00
else if ( dev - > device = = PCI_DEVICE_ID_INTEL_82865_HB )
2005-04-17 02:20:36 +04:00
switch ( dev - > subsystem_device ) {
case 0x12bc : /* HP D330L */
2005-09-22 00:26:31 +04:00
case 0x12bd : /* HP D530 */
2009-05-13 00:49:25 +04:00
case 0x006a : /* HP Compaq nx9500 */
2005-04-17 02:20:36 +04:00
asus_hides_smbus = 1 ;
}
2007-11-21 20:29:06 +03:00
else if ( dev - > device = = PCI_DEVICE_ID_INTEL_82875_HB )
switch ( dev - > subsystem_device ) {
case 0x12bf : /* HP xw4100 */
asus_hides_smbus = 1 ;
}
2014-04-19 04:13:49 +04:00
} else if ( unlikely ( dev - > subsystem_vendor = = PCI_VENDOR_ID_SAMSUNG ) ) {
if ( dev - > device = = PCI_DEVICE_ID_INTEL_82855PM_HB )
switch ( dev - > subsystem_device ) {
case 0xC00C : /* Samsung P35 notebook */
asus_hides_smbus = 1 ;
}
2005-09-07 00:39:32 +04:00
} else if ( unlikely ( dev - > subsystem_vendor = = PCI_VENDOR_ID_COMPAQ ) ) {
if ( dev - > device = = PCI_DEVICE_ID_INTEL_82855PM_HB )
2014-04-19 04:13:49 +04:00
switch ( dev - > subsystem_device ) {
2005-09-07 00:39:32 +04:00
case 0x0058 : /* Compaq Evo N620c */
asus_hides_smbus = 1 ;
}
2007-08-23 23:22:04 +04:00
else if ( dev - > device = = PCI_DEVICE_ID_INTEL_82810_IG3 )
2014-04-19 04:13:49 +04:00
switch ( dev - > subsystem_device ) {
2007-08-23 23:22:04 +04:00
case 0xB16C : /* Compaq Deskpro EP 401963-001 (PCA# 010174) */
/* Motherboard doesn't have Host bridge
* subvendor / subdevice IDs , therefore checking
* its on - board VGA controller */
asus_hides_smbus = 1 ;
}
2009-03-02 11:51:13 +03:00
else if ( dev - > device = = PCI_DEVICE_ID_INTEL_82801DB_2 )
2014-04-19 04:13:49 +04:00
switch ( dev - > subsystem_device ) {
2008-06-04 15:53:31 +04:00
case 0x00b8 : /* Compaq Evo D510 CMT */
case 0x00b9 : /* Compaq Evo D510 SFF */
2009-07-28 13:49:19 +04:00
case 0x00ba : /* Compaq Evo D510 USDT */
2009-03-02 11:51:13 +03:00
/* Motherboard doesn't have Host bridge
* subvendor / subdevice IDs and on - board VGA
* controller is disabled if an AGP card is
* inserted , therefore checking USB UHCI
* Controller # 1 */
2008-06-04 15:53:31 +04:00
asus_hides_smbus = 1 ;
}
2008-06-08 15:47:02 +04:00
else if ( dev - > device = = PCI_DEVICE_ID_INTEL_82815_CGC )
switch ( dev - > subsystem_device ) {
case 0x001A : /* Compaq Deskpro EN SSF P667 815E */
/* Motherboard doesn't have host bridge
* subvendor / subdevice IDs , therefore checking
* its on - board VGA controller */
asus_hides_smbus = 1 ;
}
2005-04-17 02:20:36 +04:00
}
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82845_HB , asus_hides_smbus_hostbridge ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82845G_HB , asus_hides_smbus_hostbridge ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82850_HB , asus_hides_smbus_hostbridge ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82865_HB , asus_hides_smbus_hostbridge ) ;
2007-11-21 20:29:06 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82875_HB , asus_hides_smbus_hostbridge ) ;
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_7205_0 , asus_hides_smbus_hostbridge ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_E7501_MCH , asus_hides_smbus_hostbridge ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82855PM_HB , asus_hides_smbus_hostbridge ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82855GM_HB , asus_hides_smbus_hostbridge ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82915GM_HB , asus_hides_smbus_hostbridge ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82810_IG3 , asus_hides_smbus_hostbridge ) ;
2009-03-02 11:51:13 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82801DB_2 , asus_hides_smbus_hostbridge ) ;
2008-06-08 15:47:02 +04:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82815_CGC , asus_hides_smbus_hostbridge ) ;
2007-08-23 23:22:04 +04:00
2006-12-05 02:14:45 +03:00
static void asus_hides_smbus_lpc ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
u16 val ;
2013-11-14 22:28:18 +04:00
2005-04-17 02:20:36 +04:00
if ( likely ( ! asus_hides_smbus ) )
return ;
pci_read_config_word ( dev , 0xF2 , & val ) ;
if ( val & 0x8 ) {
pci_write_config_word ( dev , 0xF2 , val & ( ~ 0x8 ) ) ;
pci_read_config_word ( dev , 0xF2 , & val ) ;
if ( val & 0x8 )
2018-01-18 21:55:24 +03:00
pci_info ( dev , " i801 SMBus device continues to play 'hide and seek'! 0x%x \n " ,
2014-04-19 04:13:50 +04:00
val ) ;
2005-04-17 02:20:36 +04:00
else
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Enabled i801 SMBus device \n " ) ;
2005-04-17 02:20:36 +04:00
}
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82801AA_0 , asus_hides_smbus_lpc ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82801DB_0 , asus_hides_smbus_lpc ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82801BA_0 , asus_hides_smbus_lpc ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82801CA_0 , asus_hides_smbus_lpc ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82801CA_12 , asus_hides_smbus_lpc ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82801DB_12 , asus_hides_smbus_lpc ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82801EB_0 , asus_hides_smbus_lpc ) ;
2008-05-15 23:51:31 +04:00
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82801AA_0 , asus_hides_smbus_lpc ) ;
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82801DB_0 , asus_hides_smbus_lpc ) ;
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82801BA_0 , asus_hides_smbus_lpc ) ;
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82801CA_0 , asus_hides_smbus_lpc ) ;
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82801CA_12 , asus_hides_smbus_lpc ) ;
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82801DB_12 , asus_hides_smbus_lpc ) ;
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82801EB_0 , asus_hides_smbus_lpc ) ;
2006-12-05 02:14:45 +03:00
2008-05-15 23:51:31 +04:00
/* It appears we just have one such device. If not, we have a warning */
static void __iomem * asus_rcba_base ;
static void asus_hides_smbus_lpc_ich6_suspend ( struct pci_dev * dev )
2005-09-29 12:35:41 +04:00
{
2008-05-15 23:51:31 +04:00
u32 rcba ;
2005-09-29 12:35:41 +04:00
if ( likely ( ! asus_hides_smbus ) )
return ;
2008-05-15 23:51:31 +04:00
WARN_ON ( asus_rcba_base ) ;
2005-09-29 12:35:41 +04:00
pci_read_config_dword ( dev , 0xF0 , & rcba ) ;
2008-05-15 23:51:31 +04:00
/* use bits 31:14, 16 kB aligned */
2020-01-06 11:43:50 +03:00
asus_rcba_base = ioremap ( rcba & 0xFFFFC000 , 0x4000 ) ;
2008-05-15 23:51:31 +04:00
if ( asus_rcba_base = = NULL )
return ;
}
static void asus_hides_smbus_lpc_ich6_resume_early ( struct pci_dev * dev )
{
u32 val ;
if ( likely ( ! asus_hides_smbus | | ! asus_rcba_base ) )
return ;
2018-05-02 16:53:19 +03:00
2008-05-15 23:51:31 +04:00
/* read the Function Disable register, dword mode only */
val = readl ( asus_rcba_base + 0x3418 ) ;
2018-05-02 16:53:19 +03:00
/* enable the SMBus device */
writel ( val & 0xFFFFFFF7 , asus_rcba_base + 0x3418 ) ;
2008-05-15 23:51:31 +04:00
}
static void asus_hides_smbus_lpc_ich6_resume ( struct pci_dev * dev )
{
if ( likely ( ! asus_hides_smbus | | ! asus_rcba_base ) )
return ;
2018-05-02 16:53:19 +03:00
2008-05-15 23:51:31 +04:00
iounmap ( asus_rcba_base ) ;
asus_rcba_base = NULL ;
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Enabled ICH6/i801 SMBus device \n " ) ;
2005-09-29 12:35:41 +04:00
}
2008-05-15 23:51:31 +04:00
static void asus_hides_smbus_lpc_ich6 ( struct pci_dev * dev )
{
asus_hides_smbus_lpc_ich6_suspend ( dev ) ;
asus_hides_smbus_lpc_ich6_resume_early ( dev ) ;
asus_hides_smbus_lpc_ich6_resume ( dev ) ;
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_ICH6_1 , asus_hides_smbus_lpc_ich6 ) ;
2008-05-15 23:51:31 +04:00
DECLARE_PCI_FIXUP_SUSPEND ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_ICH6_1 , asus_hides_smbus_lpc_ich6_suspend ) ;
DECLARE_PCI_FIXUP_RESUME ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_ICH6_1 , asus_hides_smbus_lpc_ich6_resume ) ;
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_ICH6_1 , asus_hides_smbus_lpc_ich6_resume_early ) ;
2006-05-15 20:44:33 +04:00
2018-05-02 16:53:19 +03:00
/* SiS 96x south bridge: BIOS typically hides SMBus device... */
2006-12-05 02:14:45 +03:00
static void quirk_sis_96x_smbus ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
u8 val = 0 ;
pci_read_config_byte ( dev , 0x77 , & val ) ;
2007-01-09 06:11:29 +03:00
if ( val & 0x10 ) {
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Enabling SiS 96x SMBus \n " ) ;
2007-01-09 06:11:29 +03:00
pci_write_config_byte ( dev , 0x77 , val & ~ 0x10 ) ;
}
2005-04-17 02:20:36 +04:00
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_SI , PCI_DEVICE_ID_SI_961 , quirk_sis_96x_smbus ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_SI , PCI_DEVICE_ID_SI_962 , quirk_sis_96x_smbus ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_SI , PCI_DEVICE_ID_SI_963 , quirk_sis_96x_smbus ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_SI , PCI_DEVICE_ID_SI_LPC , quirk_sis_96x_smbus ) ;
2008-05-15 23:51:31 +04:00
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_SI , PCI_DEVICE_ID_SI_961 , quirk_sis_96x_smbus ) ;
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_SI , PCI_DEVICE_ID_SI_962 , quirk_sis_96x_smbus ) ;
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_SI , PCI_DEVICE_ID_SI_963 , quirk_sis_96x_smbus ) ;
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_SI , PCI_DEVICE_ID_SI_LPC , quirk_sis_96x_smbus ) ;
2005-04-17 02:20:36 +04:00
/*
* . . . This is further complicated by the fact that some SiS96x south
* bridges pretend to be 85 C503 / 5513 instead . In that case see if we
* spotted a compatible north bridge to make sure .
2018-05-02 16:53:19 +03:00
* ( pci_find_device ( ) doesn ' t work yet )
2005-04-17 02:20:36 +04:00
*
* We can also enable the sis96x bit in the discovery register . .
*/
# define SIS_DETECT_REGISTER 0x40
2006-12-05 02:14:45 +03:00
static void quirk_sis_503 ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
u8 reg ;
u16 devid ;
pci_read_config_byte ( dev , SIS_DETECT_REGISTER , & reg ) ;
pci_write_config_byte ( dev , SIS_DETECT_REGISTER , reg | ( 1 < < 6 ) ) ;
pci_read_config_word ( dev , PCI_DEVICE_ID , & devid ) ;
if ( ( ( devid & 0xfff0 ) ! = 0x0960 ) & & ( devid ! = 0x0018 ) ) {
pci_write_config_byte ( dev , SIS_DETECT_REGISTER , reg ) ;
return ;
}
/*
2018-05-02 16:53:19 +03:00
* Ok , it now shows up as a 96 x . Run the 96 x quirk by hand in case
* it has already been processed . ( Depends on link order , which is
* apparently not guaranteed )
2005-04-17 02:20:36 +04:00
*/
dev - > device = devid ;
2007-01-09 06:11:29 +03:00
quirk_sis_96x_smbus ( dev ) ;
2005-04-17 02:20:36 +04:00
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_SI , PCI_DEVICE_ID_SI_503 , quirk_sis_503 ) ;
2008-05-15 23:51:31 +04:00
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_SI , PCI_DEVICE_ID_SI_503 , quirk_sis_503 ) ;
2005-04-17 02:20:36 +04:00
2006-02-28 23:44:36 +03:00
/*
* On ASUS A8V and A8V Deluxe boards , the onboard AC97 audio controller
* and MC97 modem controller are disabled when a second PCI soundcard is
* present . This patch , tweaking the VT8237 ISA bridge , enables them .
* - - bjd
*/
2006-12-05 02:14:45 +03:00
static void asus_hides_ac97_lpc ( struct pci_dev * dev )
2006-02-28 23:44:36 +03:00
{
u8 val ;
int asus_hides_ac97 = 0 ;
if ( likely ( dev - > subsystem_vendor = = PCI_VENDOR_ID_ASUSTEK ) ) {
if ( dev - > device = = PCI_DEVICE_ID_VIA_8237 )
asus_hides_ac97 = 1 ;
}
if ( ! asus_hides_ac97 )
return ;
pci_read_config_byte ( dev , 0x50 , & val ) ;
if ( val & 0xc0 ) {
pci_write_config_byte ( dev , 0x50 , val & ( ~ 0xc0 ) ) ;
pci_read_config_byte ( dev , 0x50 , & val ) ;
if ( val & 0xc0 )
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Onboard AC97/MC97 devices continue to play 'hide and seek'! 0x%x \n " ,
2014-04-19 04:13:50 +04:00
val ) ;
2006-02-28 23:44:36 +03:00
else
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Enabled onboard AC97/MC97 devices \n " ) ;
2006-02-28 23:44:36 +03:00
}
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_8237 , asus_hides_ac97_lpc ) ;
2008-05-15 23:51:31 +04:00
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_8237 , asus_hides_ac97_lpc ) ;
2006-12-05 02:14:45 +03:00
2006-08-18 22:54:39 +04:00
# if defined(CONFIG_ATA) || defined(CONFIG_ATA_MODULE)
2006-07-12 18:05:41 +04:00
/*
2018-05-02 16:53:19 +03:00
* If we are using libata we can drive this chip properly but must do this
* early on to make the additional device appear during the PCI scanning .
2006-07-12 18:05:41 +04:00
*/
2007-02-26 14:16:13 +03:00
static void quirk_jmicron_ata ( struct pci_dev * pdev )
2006-07-12 18:05:41 +04:00
{
2007-02-26 14:24:03 +03:00
u32 conf1 , conf5 , class ;
2006-07-12 18:05:41 +04:00
u8 hdr ;
/* Only poke fn 0 */
if ( PCI_FUNC ( pdev - > devfn ) )
return ;
2007-02-26 14:16:13 +03:00
pci_read_config_dword ( pdev , 0x40 , & conf1 ) ;
pci_read_config_dword ( pdev , 0x80 , & conf5 ) ;
2006-07-12 18:05:41 +04:00
2007-02-26 14:16:13 +03:00
conf1 & = ~ 0x00CFF302 ; /* Clear bit 1, 8, 9, 12-19, 22, 23 */
conf5 & = ~ ( 1 < < 24 ) ; /* Clear bit 24 */
switch ( pdev - > device ) {
2010-06-03 13:57:04 +04:00
case PCI_DEVICE_ID_JMICRON_JMB360 : /* SATA single port */
case PCI_DEVICE_ID_JMICRON_JMB362 : /* SATA dual ports */
2010-07-30 13:42:42 +04:00
case PCI_DEVICE_ID_JMICRON_JMB364 : /* SATA dual ports */
2007-02-26 14:16:13 +03:00
/* The controller should be in single function ahci mode */
conf1 | = 0x0002A100 ; /* Set 8, 13, 15, 17 */
break ;
case PCI_DEVICE_ID_JMICRON_JMB365 :
case PCI_DEVICE_ID_JMICRON_JMB366 :
/* Redirect IDE second PATA port to the right spot */
conf5 | = ( 1 < < 24 ) ;
2020-08-24 01:36:59 +03:00
fallthrough ;
2007-02-26 14:16:13 +03:00
case PCI_DEVICE_ID_JMICRON_JMB361 :
case PCI_DEVICE_ID_JMICRON_JMB363 :
2010-07-30 13:42:42 +04:00
case PCI_DEVICE_ID_JMICRON_JMB369 :
2007-02-26 14:16:13 +03:00
/* Enable dual function mode, AHCI on fn 0, IDE fn1 */
/* Set the class codes correctly and then direct IDE 0 */
2007-10-23 10:27:31 +04:00
conf1 | = 0x00C2A1B3 ; /* Set 0, 1, 4, 5, 7, 8, 13, 15, 17, 22, 23 */
2007-02-26 14:16:13 +03:00
break ;
case PCI_DEVICE_ID_JMICRON_JMB368 :
/* The controller should be in single function IDE mode */
conf1 | = 0x00C00000 ; /* Set 22, 23 */
break ;
2006-07-12 18:05:41 +04:00
}
2007-02-26 14:16:13 +03:00
pci_write_config_dword ( pdev , 0x40 , conf1 ) ;
pci_write_config_dword ( pdev , 0x80 , conf5 ) ;
/* Update pdev accordingly */
pci_read_config_byte ( pdev , PCI_HEADER_TYPE , & hdr ) ;
pdev - > hdr_type = hdr & 0x7f ;
pdev - > multifunction = ! ! ( hdr & 0x80 ) ;
2007-02-26 14:24:03 +03:00
pci_read_config_dword ( pdev , PCI_CLASS_REVISION , & class ) ;
pdev - > class = class > > 8 ;
2006-07-12 18:05:41 +04:00
}
2007-02-26 14:16:13 +03:00
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_JMICRON , PCI_DEVICE_ID_JMICRON_JMB360 , quirk_jmicron_ata ) ;
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_JMICRON , PCI_DEVICE_ID_JMICRON_JMB361 , quirk_jmicron_ata ) ;
2010-06-03 13:57:04 +04:00
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_JMICRON , PCI_DEVICE_ID_JMICRON_JMB362 , quirk_jmicron_ata ) ;
2007-02-26 14:16:13 +03:00
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_JMICRON , PCI_DEVICE_ID_JMICRON_JMB363 , quirk_jmicron_ata ) ;
2010-07-30 13:42:42 +04:00
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_JMICRON , PCI_DEVICE_ID_JMICRON_JMB364 , quirk_jmicron_ata ) ;
2007-02-26 14:16:13 +03:00
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_JMICRON , PCI_DEVICE_ID_JMICRON_JMB365 , quirk_jmicron_ata ) ;
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_JMICRON , PCI_DEVICE_ID_JMICRON_JMB366 , quirk_jmicron_ata ) ;
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_JMICRON , PCI_DEVICE_ID_JMICRON_JMB368 , quirk_jmicron_ata ) ;
2010-07-30 13:42:42 +04:00
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_JMICRON , PCI_DEVICE_ID_JMICRON_JMB369 , quirk_jmicron_ata ) ;
2008-05-15 23:51:31 +04:00
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_JMICRON , PCI_DEVICE_ID_JMICRON_JMB360 , quirk_jmicron_ata ) ;
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_JMICRON , PCI_DEVICE_ID_JMICRON_JMB361 , quirk_jmicron_ata ) ;
2010-06-03 13:57:04 +04:00
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_JMICRON , PCI_DEVICE_ID_JMICRON_JMB362 , quirk_jmicron_ata ) ;
2008-05-15 23:51:31 +04:00
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_JMICRON , PCI_DEVICE_ID_JMICRON_JMB363 , quirk_jmicron_ata ) ;
2010-07-30 13:42:42 +04:00
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_JMICRON , PCI_DEVICE_ID_JMICRON_JMB364 , quirk_jmicron_ata ) ;
2008-05-15 23:51:31 +04:00
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_JMICRON , PCI_DEVICE_ID_JMICRON_JMB365 , quirk_jmicron_ata ) ;
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_JMICRON , PCI_DEVICE_ID_JMICRON_JMB366 , quirk_jmicron_ata ) ;
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_JMICRON , PCI_DEVICE_ID_JMICRON_JMB368 , quirk_jmicron_ata ) ;
2010-07-30 13:42:42 +04:00
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_JMICRON , PCI_DEVICE_ID_JMICRON_JMB369 , quirk_jmicron_ata ) ;
2006-07-12 18:05:41 +04:00
# endif
PCI: Disable async suspend/resume for JMicron multi-function SATA/AHCI
On multi-function JMicron SATA/PATA/AHCI devices, the PATA controller at
function 1 doesn't work if it is powered on before the SATA controller at
function 0. The result is that PATA doesn't work after resume, and we
print messages like this:
pata_jmicron 0000:02:00.1: Refused to change power state, currently in D3
irq 17: nobody cared (try booting with the "irqpoll" option)
Async resume was introduced in v3.15 by 76569faa62c4 ("PM / sleep:
Asynchronous threads for resume_noirq"). Prior to that, we powered on
the functions in order, so this problem shouldn't happen.
e6b7e41cdd8c ("ata: Disabling the async PM for JMicron chip 363/361")
solved the problem for JMicron 361 and 363 devices. With async suspend
disabled, we always power on function 0 before function 1.
Barto then reported the same problem with a JMicron 368 (see comment #57 in
the bugzilla).
Rather than extending the blacklist piecemeal, disable async suspend for
all JMicron multi-function SATA/PATA/AHCI devices.
This quirk could stay in the ahci and pata_jmicron drivers, but it's likely
the problem will occur even if pata_jmicron isn't loaded until after the
suspend/resume. Making it a PCI quirk ensures that we'll preserve the
power-on order even if the drivers aren't loaded.
[bhelgaas: changelog, limit to multi-function, limit to IDE/ATA]
Link: https://bugzilla.kernel.org/show_bug.cgi?id=81551
Reported-and-tested-by: Barto <mister.freeman@laposte.net>
Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
CC: stable@vger.kernel.org # v3.15+
2015-08-24 23:27:11 +03:00
static void quirk_jmicron_async_suspend ( struct pci_dev * dev )
{
if ( dev - > multifunction ) {
device_disable_async_suspend ( & dev - > dev ) ;
2018-01-18 21:55:24 +03:00
pci_info ( dev , " async suspend disabled to avoid multi-function power-on ordering issue \n " ) ;
PCI: Disable async suspend/resume for JMicron multi-function SATA/AHCI
On multi-function JMicron SATA/PATA/AHCI devices, the PATA controller at
function 1 doesn't work if it is powered on before the SATA controller at
function 0. The result is that PATA doesn't work after resume, and we
print messages like this:
pata_jmicron 0000:02:00.1: Refused to change power state, currently in D3
irq 17: nobody cared (try booting with the "irqpoll" option)
Async resume was introduced in v3.15 by 76569faa62c4 ("PM / sleep:
Asynchronous threads for resume_noirq"). Prior to that, we powered on
the functions in order, so this problem shouldn't happen.
e6b7e41cdd8c ("ata: Disabling the async PM for JMicron chip 363/361")
solved the problem for JMicron 361 and 363 devices. With async suspend
disabled, we always power on function 0 before function 1.
Barto then reported the same problem with a JMicron 368 (see comment #57 in
the bugzilla).
Rather than extending the blacklist piecemeal, disable async suspend for
all JMicron multi-function SATA/PATA/AHCI devices.
This quirk could stay in the ahci and pata_jmicron drivers, but it's likely
the problem will occur even if pata_jmicron isn't loaded until after the
suspend/resume. Making it a PCI quirk ensures that we'll preserve the
power-on order even if the drivers aren't loaded.
[bhelgaas: changelog, limit to multi-function, limit to IDE/ATA]
Link: https://bugzilla.kernel.org/show_bug.cgi?id=81551
Reported-and-tested-by: Barto <mister.freeman@laposte.net>
Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
CC: stable@vger.kernel.org # v3.15+
2015-08-24 23:27:11 +03:00
}
}
DECLARE_PCI_FIXUP_CLASS_FINAL ( PCI_VENDOR_ID_JMICRON , PCI_ANY_ID , PCI_CLASS_STORAGE_IDE , 8 , quirk_jmicron_async_suspend ) ;
DECLARE_PCI_FIXUP_CLASS_FINAL ( PCI_VENDOR_ID_JMICRON , PCI_ANY_ID , PCI_CLASS_STORAGE_SATA_AHCI , 0 , quirk_jmicron_async_suspend ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_JMICRON , 0x2362 , quirk_jmicron_async_suspend ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_JMICRON , 0x236f , quirk_jmicron_async_suspend ) ;
2005-04-17 02:20:36 +04:00
# ifdef CONFIG_X86_IO_APIC
2012-11-22 00:35:00 +04:00
static void quirk_alder_ioapic ( struct pci_dev * pdev )
2005-04-17 02:20:36 +04:00
{
int i ;
if ( ( pdev - > class > > 8 ) ! = 0xff00 )
return ;
2018-05-02 16:53:19 +03:00
/*
* The first BAR is the location of the IO - APIC . . . we must
2005-04-17 02:20:36 +04:00
* not touch this ( and it ' s already covered by the fixmap ) , so
2018-05-02 16:53:19 +03:00
* forcibly insert it into the resource tree .
*/
2005-04-17 02:20:36 +04:00
if ( pci_resource_start ( pdev , 0 ) & & pci_resource_len ( pdev , 0 ) )
insert_resource ( & iomem_resource , & pdev - > resource [ 0 ] ) ;
2018-05-02 16:53:19 +03:00
/*
* The next five BARs all seem to be rubbish , so just clean
* them out .
*/
2019-09-28 02:43:08 +03:00
for ( i = 1 ; i < PCI_STD_NUM_BARS ; i + + )
2005-04-17 02:20:36 +04:00
memset ( & pdev - > resource [ i ] , 0 , sizeof ( pdev - > resource [ i ] ) ) ;
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_EESSC , quirk_alder_ioapic ) ;
2005-04-17 02:20:36 +04:00
# endif
2012-11-22 00:35:00 +04:00
static void quirk_pcie_mch ( struct pci_dev * pdev )
2005-04-17 02:20:36 +04:00
{
2009-09-07 08:48:35 +04:00
pdev - > no_msi = 1 ;
2005-04-17 02:20:36 +04:00
}
2007-11-22 02:07:13 +03:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_E7520_MCH , quirk_pcie_mch ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_E7320_MCH , quirk_pcie_mch ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_E7525_MCH , quirk_pcie_mch ) ;
2005-04-17 02:20:36 +04:00
2017-12-28 12:53:32 +03:00
DECLARE_PCI_FIXUP_CLASS_FINAL ( PCI_VENDOR_ID_HUAWEI , 0x1610 , PCI_CLASS_BRIDGE_PCI , 8 , quirk_pcie_mch ) ;
2005-08-17 02:15:58 +04:00
/*
2018-05-02 16:53:19 +03:00
* It ' s possible for the MSI to get corrupted if SHPC and ACPI are used
* together on certain PXH - based systems .
2005-08-17 02:15:58 +04:00
*/
2012-11-22 00:35:00 +04:00
static void quirk_pcie_pxh ( struct pci_dev * dev )
2005-08-17 02:15:58 +04:00
{
dev - > no_msi = 1 ;
2018-01-18 21:55:24 +03:00
pci_warn ( dev , " PXH quirk detected; SHPC device MSI disabled \n " ) ;
2005-08-17 02:15:58 +04:00
}
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_PXHD_0 , quirk_pcie_pxh ) ;
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_PXHD_1 , quirk_pcie_pxh ) ;
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_PXH_0 , quirk_pcie_pxh ) ;
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_PXH_1 , quirk_pcie_pxh ) ;
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_PXHV , quirk_pcie_pxh ) ;
2006-07-12 19:59:00 +04:00
/*
2018-05-02 16:53:19 +03:00
* Some Intel PCI Express chipsets have trouble with downstream device
* power management .
2006-07-12 19:59:00 +04:00
*/
2014-04-19 04:13:49 +04:00
static void quirk_intel_pcie_pm ( struct pci_dev * dev )
2006-07-12 19:59:00 +04:00
{
2020-07-31 00:08:48 +03:00
pci_pm_d3hot_delay = 120 ;
2006-07-12 19:59:00 +04:00
dev - > no_d1d2 = 1 ;
}
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x25e2 , quirk_intel_pcie_pm ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x25e3 , quirk_intel_pcie_pm ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x25e4 , quirk_intel_pcie_pm ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x25e5 , quirk_intel_pcie_pm ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x25e6 , quirk_intel_pcie_pm ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x25e7 , quirk_intel_pcie_pm ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x25f7 , quirk_intel_pcie_pm ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x25f8 , quirk_intel_pcie_pm ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x25f9 , quirk_intel_pcie_pm ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x25fa , quirk_intel_pcie_pm ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x2601 , quirk_intel_pcie_pm ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x2602 , quirk_intel_pcie_pm ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x2603 , quirk_intel_pcie_pm ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x2604 , quirk_intel_pcie_pm ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x2605 , quirk_intel_pcie_pm ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x2606 , quirk_intel_pcie_pm ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x2607 , quirk_intel_pcie_pm ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x2608 , quirk_intel_pcie_pm ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x2609 , quirk_intel_pcie_pm ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x260a , quirk_intel_pcie_pm ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x260b , quirk_intel_pcie_pm ) ;
2005-08-17 02:15:58 +04:00
2019-11-27 08:38:35 +03:00
static void quirk_d3hot_delay ( struct pci_dev * dev , unsigned int delay )
{
2020-07-31 00:08:48 +03:00
if ( dev - > d3hot_delay > = delay )
2019-11-27 08:38:35 +03:00
return ;
2020-07-31 00:08:48 +03:00
dev - > d3hot_delay = delay ;
2019-11-27 08:38:35 +03:00
pci_info ( dev , " extending delay after power-on from D3hot to %d msec \n " ,
2020-07-31 00:08:48 +03:00
dev - > d3hot_delay ) ;
2019-11-27 08:38:35 +03:00
}
2017-05-09 18:10:18 +03:00
static void quirk_radeon_pm ( struct pci_dev * dev )
{
if ( dev - > subsystem_vendor = = PCI_VENDOR_ID_APPLE & &
2019-11-27 08:38:35 +03:00
dev - > subsystem_device = = 0x00e2 )
quirk_d3hot_delay ( dev , 20 ) ;
2017-05-09 18:10:18 +03:00
}
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_ATI , 0x6741 , quirk_radeon_pm ) ;
2019-11-27 08:38:36 +03:00
/*
* Ryzen5 / 7 XHCI controllers fail upon resume from runtime suspend or s2idle .
* https : //bugzilla.kernel.org/show_bug.cgi?id=205587
*
* The kernel attempts to transition these devices to D3cold , but that seems
* to be ineffective on the platforms in question ; the PCI device appears to
* remain on in D3hot state . The D3hot - to - D0 transition then requires an
* extended delay in order to succeed .
*/
static void quirk_ryzen_xhci_d3hot ( struct pci_dev * dev )
{
quirk_d3hot_delay ( dev , 20 ) ;
}
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_AMD , 0x15e0 , quirk_ryzen_xhci_d3hot ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_AMD , 0x15e1 , quirk_ryzen_xhci_d3hot ) ;
2008-06-11 18:35:16 +04:00
# ifdef CONFIG_X86_IO_APIC
2017-04-19 10:22:45 +03:00
static int dmi_disable_ioapicreroute ( const struct dmi_system_id * d )
{
noioapicreroute = 1 ;
pr_info ( " %s detected: disable boot interrupt reroute \n " , d - > ident ) ;
return 0 ;
}
2017-09-14 12:59:30 +03:00
static const struct dmi_system_id boot_interrupt_dmi_table [ ] = {
2017-04-19 10:22:45 +03:00
/*
* Systems to exclude from boot interrupt reroute quirks
*/
{
. callback = dmi_disable_ioapicreroute ,
. ident = " ASUSTek Computer INC. M2N-LR " ,
. matches = {
DMI_MATCH ( DMI_SYS_VENDOR , " ASUSTek Computer INC. " ) ,
DMI_MATCH ( DMI_PRODUCT_NAME , " M2N-LR " ) ,
} ,
} ,
{ }
} ;
2008-06-11 18:35:17 +04:00
/*
* Boot interrupts on some chipsets cannot be turned off . For these chipsets ,
2018-05-02 16:53:19 +03:00
* remap the original interrupt in the Linux kernel to the boot interrupt , so
2008-06-11 18:35:17 +04:00
* that a PCI device ' s interrupt handler is installed on the boot interrupt
* line instead .
*/
static void quirk_reroute_to_boot_interrupts_intel ( struct pci_dev * dev )
{
2017-04-19 10:22:45 +03:00
dmi_check_system ( boot_interrupt_dmi_table ) ;
2008-07-15 15:48:55 +04:00
if ( noioapicquirk | | noioapicreroute )
2008-06-11 18:35:17 +04:00
return ;
dev - > irq_reroute_variant = INTEL_IRQ_REROUTE_VARIANT ;
2018-01-18 21:55:24 +03:00
pci_info ( dev , " rerouting interrupts for [%04x:%04x] \n " ,
2009-09-15 00:36:41 +04:00
dev - > vendor , dev - > device ) ;
2008-06-11 18:35:17 +04:00
}
2008-07-08 17:59:48 +04:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_80333_0 , quirk_reroute_to_boot_interrupts_intel ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_80333_1 , quirk_reroute_to_boot_interrupts_intel ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_ESB2_0 , quirk_reroute_to_boot_interrupts_intel ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_PXH_0 , quirk_reroute_to_boot_interrupts_intel ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_PXH_1 , quirk_reroute_to_boot_interrupts_intel ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_PXHV , quirk_reroute_to_boot_interrupts_intel ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_80332_0 , quirk_reroute_to_boot_interrupts_intel ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_80332_1 , quirk_reroute_to_boot_interrupts_intel ) ;
DECLARE_PCI_FIXUP_RESUME ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_80333_0 , quirk_reroute_to_boot_interrupts_intel ) ;
DECLARE_PCI_FIXUP_RESUME ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_80333_1 , quirk_reroute_to_boot_interrupts_intel ) ;
DECLARE_PCI_FIXUP_RESUME ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_ESB2_0 , quirk_reroute_to_boot_interrupts_intel ) ;
DECLARE_PCI_FIXUP_RESUME ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_PXH_0 , quirk_reroute_to_boot_interrupts_intel ) ;
DECLARE_PCI_FIXUP_RESUME ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_PXH_1 , quirk_reroute_to_boot_interrupts_intel ) ;
DECLARE_PCI_FIXUP_RESUME ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_PXHV , quirk_reroute_to_boot_interrupts_intel ) ;
DECLARE_PCI_FIXUP_RESUME ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_80332_0 , quirk_reroute_to_boot_interrupts_intel ) ;
DECLARE_PCI_FIXUP_RESUME ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_80332_1 , quirk_reroute_to_boot_interrupts_intel ) ;
2008-06-11 18:35:17 +04:00
2008-06-11 18:35:16 +04:00
/*
* On some chipsets we can disable the generation of legacy INTx boot
* interrupts .
*/
/*
2018-05-02 16:53:19 +03:00
* IO - APIC1 on 6300 ESB generates boot interrupts , see Intel order no
2008-06-11 18:35:16 +04:00
* 300641 - 004U S , section 5.7 .3 .
2020-02-20 22:29:29 +03:00
*
* Core IO on Xeon E5 1600 / 2600 / 4600 , see Intel order no 326509 - 003.
* Core IO on Xeon E5 v2 , see Intel order no 329188 - 003.
* Core IO on Xeon E7 v2 , see Intel order no 329595 - 002.
* Core IO on Xeon E5 v3 , see Intel order no 330784 - 003.
* Core IO on Xeon E7 v3 , see Intel order no 332315 - 001U S .
* Core IO on Xeon E5 v4 , see Intel order no 333810 - 002U S .
* Core IO on Xeon E7 v4 , see Intel order no 332315 - 001U S .
* Core IO on Xeon D - 1500 , see Intel order no 332051 - 001.
* Core IO on Xeon Scalable , see Intel order no 610950.
2008-06-11 18:35:16 +04:00
*/
2020-02-20 22:29:29 +03:00
# define INTEL_6300_IOAPIC_ABAR 0x40 /* Bus 0, Dev 29, Func 5 */
2008-06-11 18:35:16 +04:00
# define INTEL_6300_DISABLE_BOOT_IRQ (1<<14)
2020-02-20 22:29:29 +03:00
# define INTEL_CIPINTRC_CFG_OFFSET 0x14C /* Bus 0, Dev 5, Func 0 */
# define INTEL_CIPINTRC_DIS_INTX_ICH (1<<25)
2008-06-11 18:35:16 +04:00
static void quirk_disable_intel_boot_interrupt ( struct pci_dev * dev )
{
u16 pci_config_word ;
2020-02-20 22:29:29 +03:00
u32 pci_config_dword ;
2008-06-11 18:35:16 +04:00
if ( noioapicquirk )
return ;
2020-02-20 22:29:29 +03:00
switch ( dev - > device ) {
case PCI_DEVICE_ID_INTEL_ESB_10 :
pci_read_config_word ( dev , INTEL_6300_IOAPIC_ABAR ,
& pci_config_word ) ;
pci_config_word | = INTEL_6300_DISABLE_BOOT_IRQ ;
pci_write_config_word ( dev , INTEL_6300_IOAPIC_ABAR ,
pci_config_word ) ;
break ;
case 0x3c28 : /* Xeon E5 1600/2600/4600 */
case 0x0e28 : /* Xeon E5/E7 V2 */
case 0x2f28 : /* Xeon E5/E7 V3,V4 */
case 0x6f28 : /* Xeon D-1500 */
case 0x2034 : /* Xeon Scalable Family */
pci_read_config_dword ( dev , INTEL_CIPINTRC_CFG_OFFSET ,
& pci_config_dword ) ;
pci_config_dword | = INTEL_CIPINTRC_DIS_INTX_ICH ;
pci_write_config_dword ( dev , INTEL_CIPINTRC_CFG_OFFSET ,
pci_config_dword ) ;
break ;
default :
return ;
}
2018-01-18 21:55:24 +03:00
pci_info ( dev , " disabled boot interrupts on device [%04x:%04x] \n " ,
2009-09-15 00:36:41 +04:00
dev - > vendor , dev - > device ) ;
2008-06-11 18:35:16 +04:00
}
2020-02-20 22:29:29 +03:00
/*
* Device 29 Func 5 Device IDs of IO - APIC
* containing ABAR — APIC1 Alternate Base Address Register
*/
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_ESB_10 ,
quirk_disable_intel_boot_interrupt ) ;
DECLARE_PCI_FIXUP_RESUME ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_ESB_10 ,
quirk_disable_intel_boot_interrupt ) ;
/*
* Device 5 Func 0 Device IDs of Core IO modules / hubs
* containing Coherent Interface Protocol Interrupt Control
*
* Device IDs obtained from volume 2 datasheets of commented
* families above .
*/
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x3c28 ,
quirk_disable_intel_boot_interrupt ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x0e28 ,
quirk_disable_intel_boot_interrupt ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x2f28 ,
quirk_disable_intel_boot_interrupt ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x6f28 ,
quirk_disable_intel_boot_interrupt ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x2034 ,
quirk_disable_intel_boot_interrupt ) ;
DECLARE_PCI_FIXUP_RESUME ( PCI_VENDOR_ID_INTEL , 0x3c28 ,
quirk_disable_intel_boot_interrupt ) ;
DECLARE_PCI_FIXUP_RESUME ( PCI_VENDOR_ID_INTEL , 0x0e28 ,
quirk_disable_intel_boot_interrupt ) ;
DECLARE_PCI_FIXUP_RESUME ( PCI_VENDOR_ID_INTEL , 0x2f28 ,
quirk_disable_intel_boot_interrupt ) ;
DECLARE_PCI_FIXUP_RESUME ( PCI_VENDOR_ID_INTEL , 0x6f28 ,
quirk_disable_intel_boot_interrupt ) ;
DECLARE_PCI_FIXUP_RESUME ( PCI_VENDOR_ID_INTEL , 0x2034 ,
quirk_disable_intel_boot_interrupt ) ;
2008-07-08 17:59:47 +04:00
2018-05-02 16:53:19 +03:00
/* Disable boot interrupts on HT-1000 */
2008-07-08 17:59:47 +04:00
# define BC_HT1000_FEATURE_REG 0x64
# define BC_HT1000_PIC_REGS_ENABLE (1<<0)
# define BC_HT1000_MAP_IDX 0xC00
# define BC_HT1000_MAP_DATA 0xC01
static void quirk_disable_broadcom_boot_interrupt ( struct pci_dev * dev )
{
u32 pci_config_dword ;
u8 irq ;
if ( noioapicquirk )
return ;
pci_read_config_dword ( dev , BC_HT1000_FEATURE_REG , & pci_config_dword ) ;
pci_write_config_dword ( dev , BC_HT1000_FEATURE_REG , pci_config_dword |
BC_HT1000_PIC_REGS_ENABLE ) ;
for ( irq = 0x10 ; irq < 0x10 + 32 ; irq + + ) {
outb ( irq , BC_HT1000_MAP_IDX ) ;
outb ( 0x00 , BC_HT1000_MAP_DATA ) ;
}
pci_write_config_dword ( dev , BC_HT1000_FEATURE_REG , pci_config_dword ) ;
2018-01-18 21:55:24 +03:00
pci_info ( dev , " disabled boot interrupts on device [%04x:%04x] \n " ,
2009-09-15 00:36:41 +04:00
dev - > vendor , dev - > device ) ;
2008-07-08 17:59:47 +04:00
}
2013-11-14 22:28:18 +04:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_SERVERWORKS , PCI_DEVICE_ID_SERVERWORKS_HT1000SB , quirk_disable_broadcom_boot_interrupt ) ;
DECLARE_PCI_FIXUP_RESUME ( PCI_VENDOR_ID_SERVERWORKS , PCI_DEVICE_ID_SERVERWORKS_HT1000SB , quirk_disable_broadcom_boot_interrupt ) ;
2008-07-08 17:59:48 +04:00
2018-05-02 16:53:19 +03:00
/* Disable boot interrupts on AMD and ATI chipsets */
2008-07-08 17:59:48 +04:00
/*
* NOIOAMODE needs to be disabled to disable " boot interrupts " . For AMD 8131
* rev . A0 and B0 , NOIOAMODE needs to be disabled anyway to fix IO - APIC mode
* ( due to an erratum ) .
*/
# define AMD_813X_MISC 0x40
# define AMD_813X_NOIOAMODE (1<<0)
2009-10-27 10:57:42 +03:00
# define AMD_813X_REV_B1 0x12
2009-02-26 21:46:48 +03:00
# define AMD_813X_REV_B2 0x13
2008-07-08 17:59:48 +04:00
static void quirk_disable_amd_813x_boot_interrupt ( struct pci_dev * dev )
{
u32 pci_config_dword ;
if ( noioapicquirk )
return ;
2009-10-27 10:57:42 +03:00
if ( ( dev - > revision = = AMD_813X_REV_B1 ) | |
( dev - > revision = = AMD_813X_REV_B2 ) )
2009-02-26 21:46:48 +03:00
return ;
2008-07-08 17:59:48 +04:00
pci_read_config_dword ( dev , AMD_813X_MISC , & pci_config_dword ) ;
pci_config_dword & = ~ AMD_813X_NOIOAMODE ;
pci_write_config_dword ( dev , AMD_813X_MISC , pci_config_dword ) ;
2018-01-18 21:55:24 +03:00
pci_info ( dev , " disabled boot interrupts on device [%04x:%04x] \n " ,
2009-09-15 00:36:41 +04:00
dev - > vendor , dev - > device ) ;
2008-07-08 17:59:48 +04:00
}
2009-10-27 10:57:42 +03:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_AMD , PCI_DEVICE_ID_AMD_8131_BRIDGE , quirk_disable_amd_813x_boot_interrupt ) ;
DECLARE_PCI_FIXUP_RESUME ( PCI_VENDOR_ID_AMD , PCI_DEVICE_ID_AMD_8131_BRIDGE , quirk_disable_amd_813x_boot_interrupt ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_AMD , PCI_DEVICE_ID_AMD_8132_BRIDGE , quirk_disable_amd_813x_boot_interrupt ) ;
DECLARE_PCI_FIXUP_RESUME ( PCI_VENDOR_ID_AMD , PCI_DEVICE_ID_AMD_8132_BRIDGE , quirk_disable_amd_813x_boot_interrupt ) ;
2008-07-08 17:59:48 +04:00
# define AMD_8111_PCI_IRQ_ROUTING 0x56
static void quirk_disable_amd_8111_boot_interrupt ( struct pci_dev * dev )
{
u16 pci_config_word ;
if ( noioapicquirk )
return ;
pci_read_config_word ( dev , AMD_8111_PCI_IRQ_ROUTING , & pci_config_word ) ;
if ( ! pci_config_word ) {
2018-01-18 21:55:24 +03:00
pci_info ( dev , " boot interrupts on device [%04x:%04x] already disabled \n " ,
2014-04-19 04:13:50 +04:00
dev - > vendor , dev - > device ) ;
2008-07-08 17:59:48 +04:00
return ;
}
pci_write_config_word ( dev , AMD_8111_PCI_IRQ_ROUTING , 0 ) ;
2018-01-18 21:55:24 +03:00
pci_info ( dev , " disabled boot interrupts on device [%04x:%04x] \n " ,
2009-09-15 00:36:41 +04:00
dev - > vendor , dev - > device ) ;
2008-07-08 17:59:48 +04:00
}
2013-11-14 22:28:18 +04:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_AMD , PCI_DEVICE_ID_AMD_8111_SMBUS , quirk_disable_amd_8111_boot_interrupt ) ;
DECLARE_PCI_FIXUP_RESUME ( PCI_VENDOR_ID_AMD , PCI_DEVICE_ID_AMD_8111_SMBUS , quirk_disable_amd_8111_boot_interrupt ) ;
2008-06-11 18:35:16 +04:00
# endif /* CONFIG_X86_IO_APIC */
2007-02-07 20:18:45 +03:00
/*
* Toshiba TC86C001 IDE controller reports the standard 8 - byte BAR0 size
* but the PIO transfers won ' t work if BAR0 falls at the odd 8 bytes .
* Re - allocate the region if needed . . .
*/
2012-11-22 00:35:00 +04:00
static void quirk_tc86c001_ide ( struct pci_dev * dev )
2007-02-07 20:18:45 +03:00
{
struct resource * r = & dev - > resource [ 0 ] ;
if ( r - > start & 0x8 ) {
2014-02-26 22:25:58 +04:00
r - > flags | = IORESOURCE_UNSET ;
2007-02-07 20:18:45 +03:00
r - > start = 0 ;
r - > end = 0xf ;
}
}
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_TOSHIBA_2 ,
PCI_DEVICE_ID_TOSHIBA_TC86C001_IDE ,
quirk_tc86c001_ide ) ;
2012-10-30 21:25:53 +04:00
/*
2018-05-02 16:53:19 +03:00
* PLX PCI 9050 PCI Target bridge controller has an erratum that prevents the
2012-10-30 21:25:53 +04:00
* local configuration registers accessible via BAR0 ( memory ) or BAR1 ( i / o )
* being read correctly if bit 7 of the base address is set .
* The BAR0 or BAR1 region may be disabled ( size 0 ) or enabled ( size 128 ) .
* Re - allocate the regions to a 256 - byte boundary if necessary .
*/
2012-12-14 00:14:47 +04:00
static void quirk_plx_pci9050 ( struct pci_dev * dev )
2012-10-30 21:25:53 +04:00
{
unsigned int bar ;
/* Fixed in revision 2 (PCI 9052). */
if ( dev - > revision > = 2 )
return ;
for ( bar = 0 ; bar < = 1 ; bar + + )
if ( pci_resource_len ( dev , bar ) = = 0x80 & &
( pci_resource_start ( dev , bar ) & 0x80 ) ) {
struct resource * r = & dev - > resource [ bar ] ;
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Re-allocating PLX PCI 9050 BAR %u to length 256 to avoid bit 7 bug \n " ,
2012-10-30 21:25:53 +04:00
bar ) ;
2014-02-26 22:25:58 +04:00
r - > flags | = IORESOURCE_UNSET ;
2012-10-30 21:25:53 +04:00
r - > start = 0 ;
r - > end = 0xff ;
}
}
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_PLX , PCI_DEVICE_ID_PLX_9050 ,
quirk_plx_pci9050 ) ;
2012-10-29 18:40:18 +04:00
/*
* The following Meilhaus ( vendor ID 0x1402 ) device IDs ( amongst others )
* may be using the PLX PCI 9050 : 0x0630 , 0x0940 , 0x0950 , 0x0960 , 0x100b ,
* 0x1400 , 0x140a , 0x140b , 0x14e0 , 0x14ea , 0x14eb , 0x1604 , 0x1608 , 0x160c ,
* 0x168f , 0x2000 , 0x2600 , 0x3000 , 0x810a , 0x810b .
*
* Currently , device IDs 0x2000 and 0x2600 are used by the Comedi " me_daq "
* driver .
*/
DECLARE_PCI_FIXUP_HEADER ( 0x1402 , 0x2000 , quirk_plx_pci9050 ) ;
DECLARE_PCI_FIXUP_HEADER ( 0x1402 , 0x2600 , quirk_plx_pci9050 ) ;
2012-10-30 21:25:53 +04:00
2012-11-22 00:35:00 +04:00
static void quirk_netmos ( struct pci_dev * dev )
2005-04-17 02:20:36 +04:00
{
unsigned int num_parallel = ( dev - > subsystem_device & 0xf0 ) > > 4 ;
unsigned int num_serial = dev - > subsystem_device & 0xf ;
/*
* These Netmos parts are multiport serial devices with optional
* parallel ports . Even when parallel ports are present , they
* are identified as class SERIAL , which means the serial driver
* will claim them . To prevent this , mark them as class OTHER .
* These combo devices should be claimed by parport_serial .
*
* The subdevice ID is of the form 0x00 PS , where < P > is the number
* of parallel ports and < S > is the number of serial ports .
*/
switch ( dev - > device ) {
2008-12-08 18:19:14 +03:00
case PCI_DEVICE_ID_NETMOS_9835 :
/* Well, this rule doesn't hold for the following 9835 device */
if ( dev - > subsystem_vendor = = PCI_VENDOR_ID_IBM & &
dev - > subsystem_device = = 0x0299 )
return ;
2020-08-24 01:36:59 +03:00
fallthrough ;
2005-04-17 02:20:36 +04:00
case PCI_DEVICE_ID_NETMOS_9735 :
case PCI_DEVICE_ID_NETMOS_9745 :
case PCI_DEVICE_ID_NETMOS_9845 :
case PCI_DEVICE_ID_NETMOS_9855 :
2012-02-24 11:46:56 +04:00
if ( num_parallel ) {
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Netmos %04x (%u parallel, %u serial); changing class SERIAL to OTHER (use parport_serial) \n " ,
2005-04-17 02:20:36 +04:00
dev - > device , num_parallel , num_serial ) ;
dev - > class = ( PCI_CLASS_COMMUNICATION_OTHER < < 8 ) |
( dev - > class & 0xff ) ;
}
}
}
2012-02-24 11:46:56 +04:00
DECLARE_PCI_FIXUP_CLASS_HEADER ( PCI_VENDOR_ID_NETMOS , PCI_ANY_ID ,
PCI_CLASS_COMMUNICATION_SERIAL , 8 , quirk_netmos ) ;
2005-04-17 02:20:36 +04:00
2012-11-22 00:35:00 +04:00
static void quirk_e100_interrupt ( struct pci_dev * dev )
2006-04-05 16:47:00 +04:00
{
2007-12-18 00:39:27 +03:00
u16 command , pmcsr ;
2006-04-05 16:47:00 +04:00
u8 __iomem * csr ;
u8 cmd_hi ;
switch ( dev - > device ) {
/* PCI IDs taken from drivers/net/e100.c */
case 0x1029 :
case 0x1030 . . . 0x1034 :
case 0x1038 . . . 0x103E :
case 0x1050 . . . 0x1057 :
case 0x1059 :
case 0x1064 . . . 0x106B :
case 0x1091 . . . 0x1095 :
case 0x1209 :
case 0x1229 :
case 0x2449 :
case 0x2459 :
case 0x245D :
case 0x27DC :
break ;
default :
return ;
}
/*
* Some firmware hands off the e100 with interrupts enabled ,
* which can cause a flood of interrupts if packets are
* received before the driver attaches to the device . So
* disable all e100 interrupts here . The driver will
* re - enable them when it ' s ready .
*/
pci_read_config_word ( dev , PCI_COMMAND , & command ) ;
2007-09-29 03:06:21 +04:00
if ( ! ( command & PCI_COMMAND_MEMORY ) | | ! pci_resource_start ( dev , 0 ) )
2006-04-05 16:47:00 +04:00
return ;
2007-12-18 00:39:27 +03:00
/*
* Check that the device is in the D0 power state . If it ' s not ,
* there is no point to look any further .
*/
2013-06-18 12:22:14 +04:00
if ( dev - > pm_cap ) {
pci_read_config_word ( dev , dev - > pm_cap + PCI_PM_CTRL , & pmcsr ) ;
2007-12-18 00:39:27 +03:00
if ( ( pmcsr & PCI_PM_CTRL_STATE_MASK ) ! = PCI_D0 )
return ;
}
2007-09-29 03:06:21 +04:00
/* Convert from PCI bus to resource space. */
csr = ioremap ( pci_resource_start ( dev , 0 ) , 8 ) ;
2006-04-05 16:47:00 +04:00
if ( ! csr ) {
2018-01-18 21:55:24 +03:00
pci_warn ( dev , " Can't map e100 registers \n " ) ;
2006-04-05 16:47:00 +04:00
return ;
}
cmd_hi = readb ( csr + 3 ) ;
if ( cmd_hi = = 0 ) {
2018-01-18 21:55:24 +03:00
pci_warn ( dev , " Firmware left e100 interrupts enabled; disabling \n " ) ;
2006-04-05 16:47:00 +04:00
writeb ( 1 , csr + 3 ) ;
}
iounmap ( csr ) ;
}
2012-02-24 11:46:57 +04:00
DECLARE_PCI_FIXUP_CLASS_FINAL ( PCI_VENDOR_ID_INTEL , PCI_ANY_ID ,
PCI_CLASS_NETWORK_ETHERNET , 8 , quirk_e100_interrupt ) ;
2005-11-01 01:43:56 +03:00
2009-03-05 21:57:28 +03:00
/*
* The 82575 and 82598 may experience data corruption issues when transitioning
2017-09-02 00:35:50 +03:00
* out of L0S . To prevent this we need to disable L0S on the PCIe link .
2009-03-05 21:57:28 +03:00
*/
2012-11-22 00:35:00 +04:00
static void quirk_disable_aspm_l0s ( struct pci_dev * dev )
2009-03-05 21:57:28 +03:00
{
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Disabling L0s \n " ) ;
2009-03-05 21:57:28 +03:00
pci_disable_link_state ( dev , PCIE_LINK_STATE_L0S ) ;
}
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x10a7 , quirk_disable_aspm_l0s ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x10a9 , quirk_disable_aspm_l0s ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x10b6 , quirk_disable_aspm_l0s ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x10c6 , quirk_disable_aspm_l0s ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x10c7 , quirk_disable_aspm_l0s ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x10c8 , quirk_disable_aspm_l0s ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x10d6 , quirk_disable_aspm_l0s ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x10db , quirk_disable_aspm_l0s ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x10dd , quirk_disable_aspm_l0s ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x10e1 , quirk_disable_aspm_l0s ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x10ec , quirk_disable_aspm_l0s ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x10f1 , quirk_disable_aspm_l0s ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x10f4 , quirk_disable_aspm_l0s ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x1508 , quirk_disable_aspm_l0s ) ;
2020-07-22 05:18:03 +03:00
static void quirk_disable_aspm_l0s_l1 ( struct pci_dev * dev )
{
pci_info ( dev , " Disabling ASPM L0s/L1 \n " ) ;
pci_disable_link_state ( dev , PCIE_LINK_STATE_L0S | PCIE_LINK_STATE_L1 ) ;
}
/*
* ASM1083 / 1085 PCIe - PCI bridge devices cause AER timeout errors on the
* upstream PCIe root port when ASPM is enabled . At least L0s mode is affected ;
* disable both L0s and L1 for now to be safe .
*/
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_ASMEDIA , 0x1080 , quirk_disable_aspm_l0s_l1 ) ;
2019-03-29 20:07:35 +03:00
/*
* Some Pericom PCIe - to - PCI bridges in reverse mode need the PCIe Retrain
* Link bit cleared after starting the link retrain process to allow this
* process to finish .
*
* Affected devices : PI7C9X110 , PI7C9X111SL , PI7C9X130 . See also the
* Pericom Errata Sheet PI7C9X111SLB_errata_rev1 .2 _102711 . pdf .
*/
static void quirk_enable_clear_retrain_link ( struct pci_dev * dev )
{
dev - > clear_retrain_link = 1 ;
pci_info ( dev , " Enable PCIe Retrain Link quirk \n " ) ;
}
DECLARE_PCI_FIXUP_HEADER ( 0x12d8 , 0xe110 , quirk_enable_clear_retrain_link ) ;
DECLARE_PCI_FIXUP_HEADER ( 0x12d8 , 0xe111 , quirk_enable_clear_retrain_link ) ;
DECLARE_PCI_FIXUP_HEADER ( 0x12d8 , 0xe130 , quirk_enable_clear_retrain_link ) ;
2012-11-22 00:35:00 +04:00
static void fixup_rev1_53c810 ( struct pci_dev * dev )
2005-11-01 01:43:56 +03:00
{
PCI: Fix generic NCR 53c810 class code quirk
In the generic quirk fixup_rev1_53c810(), added by a5312e28c195 ("[PATCH]
PCI: NCR 53c810 quirk"), we assigned "class = PCI_CLASS_STORAGE_SCSI". But
PCI_CLASS_STORAGE_SCSI is only the two-byte base class/sub-class and needs
to be shifted to make space for the low-order interface byte.
Furthermore, we had a similar quirk, pci_fixup_ncr53c810(), for arch/x86,
which assigned class correctly. The arch code is linked before the PCI
core, so arch quirks run before generic quirks. Therefore, on x86, the x86
arch quirk ran first, and the generic quirk did nothing because it saw that
dev->class was already set. But on other arches, the generic quirk set the
wrong class code.
Fix the generic quirk to set the correct class code and remove the
now-unnecessary x86-specific quirk.
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
CC: Matthew Wilcox <matthew@wil.cx>
2015-06-19 23:36:45 +03:00
u32 class = dev - > class ;
/*
* rev 1 ncr53c810 chips don ' t set the class at all which means
2005-11-01 01:43:56 +03:00
* they don ' t get their resources remapped . Fix that here .
*/
PCI: Fix generic NCR 53c810 class code quirk
In the generic quirk fixup_rev1_53c810(), added by a5312e28c195 ("[PATCH]
PCI: NCR 53c810 quirk"), we assigned "class = PCI_CLASS_STORAGE_SCSI". But
PCI_CLASS_STORAGE_SCSI is only the two-byte base class/sub-class and needs
to be shifted to make space for the low-order interface byte.
Furthermore, we had a similar quirk, pci_fixup_ncr53c810(), for arch/x86,
which assigned class correctly. The arch code is linked before the PCI
core, so arch quirks run before generic quirks. Therefore, on x86, the x86
arch quirk ran first, and the generic quirk did nothing because it saw that
dev->class was already set. But on other arches, the generic quirk set the
wrong class code.
Fix the generic quirk to set the correct class code and remove the
now-unnecessary x86-specific quirk.
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
CC: Matthew Wilcox <matthew@wil.cx>
2015-06-19 23:36:45 +03:00
if ( class )
return ;
2005-11-01 01:43:56 +03:00
PCI: Fix generic NCR 53c810 class code quirk
In the generic quirk fixup_rev1_53c810(), added by a5312e28c195 ("[PATCH]
PCI: NCR 53c810 quirk"), we assigned "class = PCI_CLASS_STORAGE_SCSI". But
PCI_CLASS_STORAGE_SCSI is only the two-byte base class/sub-class and needs
to be shifted to make space for the low-order interface byte.
Furthermore, we had a similar quirk, pci_fixup_ncr53c810(), for arch/x86,
which assigned class correctly. The arch code is linked before the PCI
core, so arch quirks run before generic quirks. Therefore, on x86, the x86
arch quirk ran first, and the generic quirk did nothing because it saw that
dev->class was already set. But on other arches, the generic quirk set the
wrong class code.
Fix the generic quirk to set the correct class code and remove the
now-unnecessary x86-specific quirk.
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
CC: Matthew Wilcox <matthew@wil.cx>
2015-06-19 23:36:45 +03:00
dev - > class = PCI_CLASS_STORAGE_SCSI < < 8 ;
2018-01-18 21:55:24 +03:00
pci_info ( dev , " NCR 53c810 rev 1 PCI class overridden (%#08x -> %#08x) \n " ,
PCI: Fix generic NCR 53c810 class code quirk
In the generic quirk fixup_rev1_53c810(), added by a5312e28c195 ("[PATCH]
PCI: NCR 53c810 quirk"), we assigned "class = PCI_CLASS_STORAGE_SCSI". But
PCI_CLASS_STORAGE_SCSI is only the two-byte base class/sub-class and needs
to be shifted to make space for the low-order interface byte.
Furthermore, we had a similar quirk, pci_fixup_ncr53c810(), for arch/x86,
which assigned class correctly. The arch code is linked before the PCI
core, so arch quirks run before generic quirks. Therefore, on x86, the x86
arch quirk ran first, and the generic quirk did nothing because it saw that
dev->class was already set. But on other arches, the generic quirk set the
wrong class code.
Fix the generic quirk to set the correct class code and remove the
now-unnecessary x86-specific quirk.
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
CC: Matthew Wilcox <matthew@wil.cx>
2015-06-19 23:36:45 +03:00
class , dev - > class ) ;
2005-11-01 01:43:56 +03:00
}
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_NCR , PCI_DEVICE_ID_NCR_53C810 , fixup_rev1_53c810 ) ;
2005-12-05 15:06:43 +03:00
/* Enable 1k I/O space granularity on the Intel P64H2 */
2012-11-22 00:35:00 +04:00
static void quirk_p64h2_1k_io ( struct pci_dev * dev )
2005-12-05 15:06:43 +03:00
{
u16 en1k ;
pci_read_config_word ( dev , 0x40 , & en1k ) ;
if ( en1k & 0x200 ) {
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Enable I/O Space to 1KB granularity \n " ) ;
2012-07-09 23:38:57 +04:00
dev - > io_window_1k = 1 ;
2005-12-05 15:06:43 +03:00
}
}
2018-05-02 16:53:19 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x1460 , quirk_p64h2_1k_io ) ;
2005-12-05 15:06:43 +03:00
2018-05-02 16:53:19 +03:00
/*
* Under some circumstances , AER is not linked with extended capabilities .
2006-06-13 22:35:42 +04:00
* Force it to be linked by setting the corresponding control bit in the
* config space .
*/
2006-12-05 02:14:45 +03:00
static void quirk_nvidia_ck804_pcie_aer_ext_cap ( struct pci_dev * dev )
2006-06-13 22:35:42 +04:00
{
uint8_t b ;
2018-05-02 16:53:19 +03:00
2006-06-13 22:35:42 +04:00
if ( pci_read_config_byte ( dev , 0xf41 , & b ) = = 0 ) {
if ( ! ( b & 0x20 ) ) {
pci_write_config_byte ( dev , 0xf41 , b | 0x20 ) ;
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Linking AER extended capability \n " ) ;
2006-06-13 22:35:42 +04:00
}
}
}
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_NVIDIA , PCI_DEVICE_ID_NVIDIA_CK804_PCIE ,
quirk_nvidia_ck804_pcie_aer_ext_cap ) ;
2008-05-15 23:51:31 +04:00
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_NVIDIA , PCI_DEVICE_ID_NVIDIA_CK804_PCIE ,
2006-12-05 02:14:45 +03:00
quirk_nvidia_ck804_pcie_aer_ext_cap ) ;
2006-06-13 22:35:42 +04:00
2012-11-22 00:35:00 +04:00
static void quirk_via_cx700_pci_parking_caching ( struct pci_dev * dev )
2007-11-02 02:14:54 +03:00
{
/*
* Disable PCI Bus Parking and PCI Master read caching on CX700
* which causes unspecified timing errors with a VT6212L on the PCI
2010-03-20 00:22:58 +03:00
* bus leading to USB2 .0 packet loss .
*
* This quirk is only enabled if a second ( on the external PCI bus )
* VT6212L is found - - the CX700 core itself also contains a USB
* host controller with the same PCI ID as the VT6212L .
2007-11-02 02:14:54 +03:00
*/
2010-03-20 00:22:58 +03:00
/* Count VT6212L instances */
struct pci_dev * p = pci_get_device ( PCI_VENDOR_ID_VIA ,
PCI_DEVICE_ID_VIA_8235_USB_2 , NULL ) ;
2007-11-02 02:14:54 +03:00
uint8_t b ;
2010-03-20 00:22:58 +03:00
2018-05-02 16:53:19 +03:00
/*
* p should contain the first ( internal ) VT6212L - - see if we have
* an external one by searching again .
*/
2010-03-20 00:22:58 +03:00
p = pci_get_device ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_8235_USB_2 , p ) ;
if ( ! p )
return ;
pci_dev_put ( p ) ;
2007-11-02 02:14:54 +03:00
if ( pci_read_config_byte ( dev , 0x76 , & b ) = = 0 ) {
if ( b & 0x40 ) {
/* Turn off PCI Bus Parking */
pci_write_config_byte ( dev , 0x76 , b ^ 0x40 ) ;
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Disabling VIA CX700 PCI parking \n " ) ;
2008-03-30 23:58:59 +04:00
}
}
if ( pci_read_config_byte ( dev , 0x72 , & b ) = = 0 ) {
if ( b ! = 0 ) {
2007-11-02 02:14:54 +03:00
/* Turn off PCI Master read caching */
pci_write_config_byte ( dev , 0x72 , 0x0 ) ;
2008-03-30 23:58:59 +04:00
/* Set PCI Master Bus time-out to "1x16 PCLK" */
2007-11-02 02:14:54 +03:00
pci_write_config_byte ( dev , 0x75 , 0x1 ) ;
2008-03-30 23:58:59 +04:00
/* Disable "Read FIFO Timer" */
2007-11-02 02:14:54 +03:00
pci_write_config_byte ( dev , 0x77 , 0x0 ) ;
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Disabling VIA CX700 PCI caching \n " ) ;
2007-11-02 02:14:54 +03:00
}
}
}
2010-03-20 00:22:58 +03:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_VIA , 0x324e , quirk_via_cx700_pci_parking_caching ) ;
2007-11-02 02:14:54 +03:00
2012-07-10 01:36:14 +04:00
static void quirk_brcm_5719_limit_mrrs ( struct pci_dev * dev )
2012-02-27 13:44:48 +04:00
{
u32 rev ;
pci_read_config_dword ( dev , 0xf4 , & rev ) ;
/* Only CAP the MRRS if the device is a 5719 A0 */
if ( rev = = 0x05719000 ) {
int readrq = pcie_get_readrq ( dev ) ;
if ( readrq > 2048 )
pcie_set_readrq ( dev , 2048 ) ;
}
}
DECLARE_PCI_FIXUP_ENABLE ( PCI_VENDOR_ID_BROADCOM ,
PCI_DEVICE_ID_TIGON3_5719 ,
quirk_brcm_5719_limit_mrrs ) ;
2018-05-02 16:53:19 +03:00
/*
* Originally in EDAC sources for i82875P : Intel tells BIOS developers to
* hide device 6 which configures the overflow device access containing the
* DRBs - this is where we expose device 6.
2009-05-13 00:49:26 +04:00
* http : //www.x86-secret.com/articles/tweak/pat/patsecrets-2.htm
*/
2012-11-22 00:35:00 +04:00
static void quirk_unhide_mch_dev6 ( struct pci_dev * dev )
2009-05-13 00:49:26 +04:00
{
u8 reg ;
if ( pci_read_config_byte ( dev , 0xF4 , & reg ) = = 0 & & ! ( reg & 0x02 ) ) {
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Enabling MCH 'Overflow' Device \n " ) ;
2009-05-13 00:49:26 +04:00
pci_write_config_byte ( dev , 0xF4 , reg | 0x02 ) ;
}
}
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82865_HB ,
quirk_unhide_mch_dev6 ) ;
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82875_HB ,
quirk_unhide_mch_dev6 ) ;
2006-08-31 09:54:56 +04:00
# ifdef CONFIG_PCI_MSI
2018-05-02 16:53:19 +03:00
/*
* Some chipsets do not support MSI . We cannot easily rely on setting
* PCI_BUS_FLAGS_NO_MSI in its bus flags because there are actually some
* other buses controlled by the chipset even if Linux is not aware of it .
* Instead of setting the flag on all buses in the machine , simply disable
* MSI globally .
2006-08-31 09:54:56 +04:00
*/
2012-11-22 00:35:00 +04:00
static void quirk_disable_all_msi ( struct pci_dev * dev )
2006-08-31 09:54:56 +04:00
{
2007-01-25 11:34:07 +03:00
pci_no_msi ( ) ;
2018-01-18 21:55:24 +03:00
pci_warn ( dev , " MSI quirk detected; MSI disabled \n " ) ;
2006-08-31 09:54:56 +04:00
}
2007-05-31 11:40:48 +04:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_SERVERWORKS , PCI_DEVICE_ID_SERVERWORKS_GCNB_LE , quirk_disable_all_msi ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_ATI , PCI_DEVICE_ID_ATI_RS400_200 , quirk_disable_all_msi ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_ATI , PCI_DEVICE_ID_ATI_RS480 , quirk_disable_all_msi ) ;
2008-07-04 20:59:32 +04:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_VT3336 , quirk_disable_all_msi ) ;
2007-05-27 02:01:04 +04:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_VT3351 , quirk_disable_all_msi ) ;
2009-04-03 17:34:00 +04:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_VT3364 , quirk_disable_all_msi ) ;
2010-05-23 12:22:55 +04:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_VIA , PCI_DEVICE_ID_VIA_8380_0 , quirk_disable_all_msi ) ;
2015-09-25 01:02:07 +03:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_SI , 0x0761 , quirk_disable_all_msi ) ;
2006-08-31 09:54:56 +04:00
/* Disable MSI on chipsets that are known to not support it */
2012-11-22 00:35:00 +04:00
static void quirk_disable_msi ( struct pci_dev * dev )
2006-08-31 09:54:56 +04:00
{
if ( dev - > subordinate ) {
2018-01-18 21:55:24 +03:00
pci_warn ( dev , " MSI quirk detected; subordinate MSI disabled \n " ) ;
2006-08-31 09:54:56 +04:00
dev - > subordinate - > bus_flags | = PCI_BUS_FLAGS_NO_MSI ;
}
}
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_AMD , PCI_DEVICE_ID_AMD_8131_BRIDGE , quirk_disable_msi ) ;
2010-03-24 16:11:01 +03:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_VIA , 0xa238 , quirk_disable_msi ) ;
2010-05-18 18:42:53 +04:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_ATI , 0x5a3f , quirk_disable_msi ) ;
2006-08-31 09:55:32 +04:00
2010-05-26 14:21:10 +04:00
/*
* The APC bridge device in AMD 780 family northbridges has some random
* OEM subsystem ID in its vendor ID register ( erratum 18 ) , so instead
* we use the possible vendor / device IDs of the host bridge for the
* declared quirk , and search for the APC bridge by slot number .
*/
2012-11-22 00:35:00 +04:00
static void quirk_amd_780_apc_msi ( struct pci_dev * host_bridge )
2010-05-26 14:21:10 +04:00
{
struct pci_dev * apc_bridge ;
apc_bridge = pci_get_slot ( host_bridge - > bus , PCI_DEVFN ( 1 , 0 ) ) ;
if ( apc_bridge ) {
if ( apc_bridge - > device = = 0x9602 )
quirk_disable_msi ( apc_bridge ) ;
pci_dev_put ( apc_bridge ) ;
}
}
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_AMD , 0x9600 , quirk_amd_780_apc_msi ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_AMD , 0x9601 , quirk_amd_780_apc_msi ) ;
2018-05-02 16:53:19 +03:00
/*
* Go through the list of HyperTransport capabilities and return 1 if a HT
* MSI capability is found and enabled .
*/
2012-07-10 01:36:14 +04:00
static int msi_ht_cap_enabled ( struct pci_dev * dev )
2006-08-31 09:55:32 +04:00
{
2015-06-30 04:16:41 +03:00
int pos , ttl = PCI_FIND_CAP_TTL ;
2006-11-22 10:26:21 +03:00
pos = pci_find_ht_capability ( dev , HT_CAPTYPE_MSI_MAPPING ) ;
while ( pos & & ttl - - ) {
u8 flags ;
if ( pci_read_config_byte ( dev , pos + HT_MSI_FLAGS ,
2014-04-19 04:13:49 +04:00
& flags ) = = 0 ) {
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Found %s HT MSI Mapping \n " ,
2006-11-22 10:26:21 +03:00
flags & HT_MSI_FLAGS_ENABLE ?
2007-12-18 00:09:39 +03:00
" enabled " : " disabled " ) ;
2006-11-22 10:26:21 +03:00
return ( flags & HT_MSI_FLAGS_ENABLE ) ! = 0 ;
2006-08-31 09:55:32 +04:00
}
2006-11-22 10:26:21 +03:00
pos = pci_find_next_ht_capability ( dev , pos ,
HT_CAPTYPE_MSI_MAPPING ) ;
2006-08-31 09:55:32 +04:00
}
return 0 ;
}
2018-05-02 16:53:19 +03:00
/* Check the HyperTransport MSI mapping to know whether MSI is enabled or not */
2012-07-10 01:36:14 +04:00
static void quirk_msi_ht_cap ( struct pci_dev * dev )
2006-08-31 09:55:32 +04:00
{
if ( dev - > subordinate & & ! msi_ht_cap_enabled ( dev ) ) {
2018-01-18 21:55:24 +03:00
pci_warn ( dev , " MSI quirk detected; subordinate MSI disabled \n " ) ;
2006-08-31 09:55:32 +04:00
dev - > subordinate - > bus_flags | = PCI_BUS_FLAGS_NO_MSI ;
}
}
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_SERVERWORKS , PCI_DEVICE_ID_SERVERWORKS_HT2000_PCIE ,
quirk_msi_ht_cap ) ;
2007-12-14 03:09:25 +03:00
2018-05-02 16:53:19 +03:00
/*
* The nVidia CK804 chipset may have 2 HT MSI mappings . MSI is supported
* if the MSI capability is set in any of these mappings .
2006-08-31 09:55:32 +04:00
*/
2012-07-10 01:36:14 +04:00
static void quirk_nvidia_ck804_msi_ht_cap ( struct pci_dev * dev )
2006-08-31 09:55:32 +04:00
{
struct pci_dev * pdev ;
if ( ! dev - > subordinate )
return ;
2018-05-02 16:53:19 +03:00
/*
* Check HT MSI cap on this chipset and the root one . A single one
* having MSI is enough to be sure that MSI is supported .
2006-08-31 09:55:32 +04:00
*/
2006-10-11 01:39:00 +04:00
pdev = pci_get_slot ( dev - > bus , 0 ) ;
2006-12-05 02:14:48 +03:00
if ( ! pdev )
return ;
2006-12-03 22:55:34 +03:00
if ( ! msi_ht_cap_enabled ( dev ) & & ! msi_ht_cap_enabled ( pdev ) ) {
2018-01-18 21:55:24 +03:00
pci_warn ( dev , " MSI quirk detected; subordinate MSI disabled \n " ) ;
2006-08-31 09:55:32 +04:00
dev - > subordinate - > bus_flags | = PCI_BUS_FLAGS_NO_MSI ;
}
2006-10-11 01:39:00 +04:00
pci_dev_put ( pdev ) ;
2006-08-31 09:55:32 +04:00
}
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_NVIDIA , PCI_DEVICE_ID_NVIDIA_CK804_PCIE ,
quirk_nvidia_ck804_msi_ht_cap ) ;
2007-10-25 12:16:30 +04:00
2008-03-01 02:04:39 +03:00
/* Force enable MSI mapping capability on HT bridges */
2012-07-10 01:36:14 +04:00
static void ht_enable_msi_mapping ( struct pci_dev * dev )
2008-02-05 10:50:13 +03:00
{
2015-06-30 04:16:41 +03:00
int pos , ttl = PCI_FIND_CAP_TTL ;
2008-02-05 10:50:13 +03:00
pos = pci_find_ht_capability ( dev , HT_CAPTYPE_MSI_MAPPING ) ;
while ( pos & & ttl - - ) {
u8 flags ;
if ( pci_read_config_byte ( dev , pos + HT_MSI_FLAGS ,
& flags ) = = 0 ) {
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Enabling HT MSI Mapping \n " ) ;
2008-02-05 10:50:13 +03:00
pci_write_config_byte ( dev , pos + HT_MSI_FLAGS ,
flags | HT_MSI_FLAGS_ENABLE ) ;
}
pos = pci_find_next_ht_capability ( dev , pos ,
HT_CAPTYPE_MSI_MAPPING ) ;
}
}
2008-03-01 02:04:39 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_SERVERWORKS ,
PCI_DEVICE_ID_SERVERWORKS_HT1000_PXB ,
ht_enable_msi_mapping ) ;
2009-02-18 07:40:09 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_AMD , PCI_DEVICE_ID_AMD_8132_BRIDGE ,
ht_enable_msi_mapping ) ;
2018-05-02 16:53:19 +03:00
/*
* The P5N32 - SLI motherboards from Asus have a problem with MSI
* for the MCP55 NIC . It is not yet determined whether the MSI problem
* also affects other devices . As for now , turn off MSI for this device .
2008-11-21 07:42:25 +03:00
*/
2012-11-22 00:35:00 +04:00
static void nvenet_msi_disable ( struct pci_dev * dev )
2008-11-21 07:42:25 +03:00
{
2011-05-15 20:13:46 +04:00
const char * board_name = dmi_get_system_info ( DMI_BOARD_NAME ) ;
if ( board_name & &
( strstr ( board_name , " P5N32-SLI PREMIUM " ) | |
strstr ( board_name , " P5N32-E SLI " ) ) ) {
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Disabling MSI for MCP55 NIC on P5N32-SLI \n " ) ;
2008-11-21 07:42:25 +03:00
dev - > no_msi = 1 ;
}
}
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_NVIDIA ,
PCI_DEVICE_ID_NVIDIA_NVENET_15 ,
nvenet_msi_disable ) ;
2019-08-13 14:36:16 +03:00
/*
* PCIe spec r4 .0 sec 7.7 .1 .2 and sec 7.7 .2 .2 say that if MSI / MSI - X is enabled ,
* then the device can ' t use INTx interrupts . Tegra ' s PCIe root ports don ' t
* generate MSI interrupts for PME and AER events instead only INTx interrupts
* are generated . Though Tegra ' s PCIe root ports can generate MSI interrupts
* for other events , since PCIe specificiation doesn ' t support using a mix of
* INTx and MSI / MSI - X , it is required to disable MSI interrupts to avoid port
* service drivers registering their respective ISRs for MSIs .
*/
static void pci_quirk_nvidia_tegra_disable_rp_msi ( struct pci_dev * dev )
{
dev - > no_msi = 1 ;
}
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_NVIDIA , 0x1ad0 ,
PCI_CLASS_BRIDGE_PCI , 8 ,
pci_quirk_nvidia_tegra_disable_rp_msi ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_NVIDIA , 0x1ad1 ,
PCI_CLASS_BRIDGE_PCI , 8 ,
pci_quirk_nvidia_tegra_disable_rp_msi ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_NVIDIA , 0x1ad2 ,
PCI_CLASS_BRIDGE_PCI , 8 ,
pci_quirk_nvidia_tegra_disable_rp_msi ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_NVIDIA , 0x0bf0 ,
PCI_CLASS_BRIDGE_PCI , 8 ,
pci_quirk_nvidia_tegra_disable_rp_msi ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_NVIDIA , 0x0bf1 ,
PCI_CLASS_BRIDGE_PCI , 8 ,
pci_quirk_nvidia_tegra_disable_rp_msi ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_NVIDIA , 0x0e1c ,
PCI_CLASS_BRIDGE_PCI , 8 ,
pci_quirk_nvidia_tegra_disable_rp_msi ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_NVIDIA , 0x0e1d ,
PCI_CLASS_BRIDGE_PCI , 8 ,
pci_quirk_nvidia_tegra_disable_rp_msi ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_NVIDIA , 0x0e12 ,
PCI_CLASS_BRIDGE_PCI , 8 ,
pci_quirk_nvidia_tegra_disable_rp_msi ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_NVIDIA , 0x0e13 ,
PCI_CLASS_BRIDGE_PCI , 8 ,
pci_quirk_nvidia_tegra_disable_rp_msi ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_NVIDIA , 0x0fae ,
PCI_CLASS_BRIDGE_PCI , 8 ,
pci_quirk_nvidia_tegra_disable_rp_msi ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_NVIDIA , 0x0faf ,
PCI_CLASS_BRIDGE_PCI , 8 ,
pci_quirk_nvidia_tegra_disable_rp_msi ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_NVIDIA , 0x10e5 ,
PCI_CLASS_BRIDGE_PCI , 8 ,
pci_quirk_nvidia_tegra_disable_rp_msi ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_NVIDIA , 0x10e6 ,
PCI_CLASS_BRIDGE_PCI , 8 ,
pci_quirk_nvidia_tegra_disable_rp_msi ) ;
PCI: add quirk for non-symmetric-mode irq routing to versions 0 and 4 of the MCP55 northbridge
A long time ago I worked on a RHEL5 bug in which kdump hung during boot
on a set of systems. The systems hung because they never received timer
interrupts during calibrate_delay. These systems also all had Opteron
processors on a hypertransport bus, bridged to a pci bus via an Nvidia
MCP55 northbridge chip. After much wrangling I managed to learn from
Nvidia that they have an undocumented register in some versions of that
chip which control how legacy interrupts are send to the cpu complex
when the ioapic isn't active. Nvidia defaults this register to only
send legacy interrupts to the BSP, so if kdump happens to boot on an AP,
we never get timer interrupts and boom. I had initially used this quirk
as a workaround, with my intent being to move apic initalization to an
earlier point in the boot process, so the setting of the register would
be irrelevant. Given the work involved in doing that however, the
fragile nature of the apic initalization code, and the fact that, over
the 2 years since we found this bug, the MCP55 is the only chip which
seems to have this issue, I've figure at this point its likely safer to
just carry the quirk around. By setting the referenced bits in this
hidden register, interrupts will be broadcast to all cpus when the
ioapic isn't active on the above described systems.
Acked-by: Simon Horman <horms@verge.net.au>
Acked-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
2010-09-21 21:54:39 +04:00
/*
2013-11-14 22:28:18 +04:00
* Some versions of the MCP55 bridge from Nvidia have a legacy IRQ routing
* config register . This register controls the routing of legacy
* interrupts from devices that route through the MCP55 . If this register
* is misprogrammed , interrupts are only sent to the BSP , unlike
* conventional systems where the IRQ is broadcast to all online CPUs . Not
* having this register set properly prevents kdump from booting up
* properly , so let ' s make sure that we have it set correctly .
* Note that this is an undocumented register .
PCI: add quirk for non-symmetric-mode irq routing to versions 0 and 4 of the MCP55 northbridge
A long time ago I worked on a RHEL5 bug in which kdump hung during boot
on a set of systems. The systems hung because they never received timer
interrupts during calibrate_delay. These systems also all had Opteron
processors on a hypertransport bus, bridged to a pci bus via an Nvidia
MCP55 northbridge chip. After much wrangling I managed to learn from
Nvidia that they have an undocumented register in some versions of that
chip which control how legacy interrupts are send to the cpu complex
when the ioapic isn't active. Nvidia defaults this register to only
send legacy interrupts to the BSP, so if kdump happens to boot on an AP,
we never get timer interrupts and boom. I had initially used this quirk
as a workaround, with my intent being to move apic initalization to an
earlier point in the boot process, so the setting of the register would
be irrelevant. Given the work involved in doing that however, the
fragile nature of the apic initalization code, and the fact that, over
the 2 years since we found this bug, the MCP55 is the only chip which
seems to have this issue, I've figure at this point its likely safer to
just carry the quirk around. By setting the referenced bits in this
hidden register, interrupts will be broadcast to all cpus when the
ioapic isn't active on the above described systems.
Acked-by: Simon Horman <horms@verge.net.au>
Acked-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
2010-09-21 21:54:39 +04:00
*/
2012-11-22 00:35:00 +04:00
static void nvbridge_check_legacy_irq_routing ( struct pci_dev * dev )
PCI: add quirk for non-symmetric-mode irq routing to versions 0 and 4 of the MCP55 northbridge
A long time ago I worked on a RHEL5 bug in which kdump hung during boot
on a set of systems. The systems hung because they never received timer
interrupts during calibrate_delay. These systems also all had Opteron
processors on a hypertransport bus, bridged to a pci bus via an Nvidia
MCP55 northbridge chip. After much wrangling I managed to learn from
Nvidia that they have an undocumented register in some versions of that
chip which control how legacy interrupts are send to the cpu complex
when the ioapic isn't active. Nvidia defaults this register to only
send legacy interrupts to the BSP, so if kdump happens to boot on an AP,
we never get timer interrupts and boom. I had initially used this quirk
as a workaround, with my intent being to move apic initalization to an
earlier point in the boot process, so the setting of the register would
be irrelevant. Given the work involved in doing that however, the
fragile nature of the apic initalization code, and the fact that, over
the 2 years since we found this bug, the MCP55 is the only chip which
seems to have this issue, I've figure at this point its likely safer to
just carry the quirk around. By setting the referenced bits in this
hidden register, interrupts will be broadcast to all cpus when the
ioapic isn't active on the above described systems.
Acked-by: Simon Horman <horms@verge.net.au>
Acked-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
2010-09-21 21:54:39 +04:00
{
u32 cfg ;
2010-12-08 17:47:48 +03:00
if ( ! pci_find_capability ( dev , PCI_CAP_ID_HT ) )
return ;
PCI: add quirk for non-symmetric-mode irq routing to versions 0 and 4 of the MCP55 northbridge
A long time ago I worked on a RHEL5 bug in which kdump hung during boot
on a set of systems. The systems hung because they never received timer
interrupts during calibrate_delay. These systems also all had Opteron
processors on a hypertransport bus, bridged to a pci bus via an Nvidia
MCP55 northbridge chip. After much wrangling I managed to learn from
Nvidia that they have an undocumented register in some versions of that
chip which control how legacy interrupts are send to the cpu complex
when the ioapic isn't active. Nvidia defaults this register to only
send legacy interrupts to the BSP, so if kdump happens to boot on an AP,
we never get timer interrupts and boom. I had initially used this quirk
as a workaround, with my intent being to move apic initalization to an
earlier point in the boot process, so the setting of the register would
be irrelevant. Given the work involved in doing that however, the
fragile nature of the apic initalization code, and the fact that, over
the 2 years since we found this bug, the MCP55 is the only chip which
seems to have this issue, I've figure at this point its likely safer to
just carry the quirk around. By setting the referenced bits in this
hidden register, interrupts will be broadcast to all cpus when the
ioapic isn't active on the above described systems.
Acked-by: Simon Horman <horms@verge.net.au>
Acked-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
2010-09-21 21:54:39 +04:00
pci_read_config_dword ( dev , 0x74 , & cfg ) ;
if ( cfg & ( ( 1 < < 2 ) | ( 1 < < 15 ) ) ) {
2019-04-20 07:03:46 +03:00
pr_info ( " Rewriting IRQ routing register on MCP55 \n " ) ;
PCI: add quirk for non-symmetric-mode irq routing to versions 0 and 4 of the MCP55 northbridge
A long time ago I worked on a RHEL5 bug in which kdump hung during boot
on a set of systems. The systems hung because they never received timer
interrupts during calibrate_delay. These systems also all had Opteron
processors on a hypertransport bus, bridged to a pci bus via an Nvidia
MCP55 northbridge chip. After much wrangling I managed to learn from
Nvidia that they have an undocumented register in some versions of that
chip which control how legacy interrupts are send to the cpu complex
when the ioapic isn't active. Nvidia defaults this register to only
send legacy interrupts to the BSP, so if kdump happens to boot on an AP,
we never get timer interrupts and boom. I had initially used this quirk
as a workaround, with my intent being to move apic initalization to an
earlier point in the boot process, so the setting of the register would
be irrelevant. Given the work involved in doing that however, the
fragile nature of the apic initalization code, and the fact that, over
the 2 years since we found this bug, the MCP55 is the only chip which
seems to have this issue, I've figure at this point its likely safer to
just carry the quirk around. By setting the referenced bits in this
hidden register, interrupts will be broadcast to all cpus when the
ioapic isn't active on the above described systems.
Acked-by: Simon Horman <horms@verge.net.au>
Acked-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
2010-09-21 21:54:39 +04:00
cfg & = ~ ( ( 1 < < 2 ) | ( 1 < < 15 ) ) ;
pci_write_config_dword ( dev , 0x74 , cfg ) ;
}
}
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_NVIDIA ,
PCI_DEVICE_ID_NVIDIA_MCP55_BRIDGE_V0 ,
nvbridge_check_legacy_irq_routing ) ;
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_NVIDIA ,
PCI_DEVICE_ID_NVIDIA_MCP55_BRIDGE_V4 ,
nvbridge_check_legacy_irq_routing ) ;
2012-07-10 01:36:14 +04:00
static int ht_check_msi_mapping ( struct pci_dev * dev )
2009-03-21 05:29:41 +03:00
{
2015-06-30 04:16:41 +03:00
int pos , ttl = PCI_FIND_CAP_TTL ;
2009-03-21 05:29:41 +03:00
int found = 0 ;
2018-05-02 16:53:19 +03:00
/* Check if there is HT MSI cap or enabled on this device */
2009-03-21 05:29:41 +03:00
pos = pci_find_ht_capability ( dev , HT_CAPTYPE_MSI_MAPPING ) ;
while ( pos & & ttl - - ) {
u8 flags ;
if ( found < 1 )
found = 1 ;
if ( pci_read_config_byte ( dev , pos + HT_MSI_FLAGS ,
& flags ) = = 0 ) {
if ( flags & HT_MSI_FLAGS_ENABLE ) {
if ( found < 2 ) {
found = 2 ;
break ;
}
}
}
pos = pci_find_next_ht_capability ( dev , pos ,
HT_CAPTYPE_MSI_MAPPING ) ;
}
return found ;
}
2012-07-10 01:36:14 +04:00
static int host_bridge_with_leaf ( struct pci_dev * host_bridge )
2009-03-21 05:29:41 +03:00
{
struct pci_dev * dev ;
int pos ;
int i , dev_no ;
int found = 0 ;
dev_no = host_bridge - > devfn > > 3 ;
for ( i = dev_no + 1 ; i < 0x20 ; i + + ) {
dev = pci_get_slot ( host_bridge - > bus , PCI_DEVFN ( i , 0 ) ) ;
if ( ! dev )
continue ;
2018-05-02 16:53:19 +03:00
/* found next host bridge? */
2009-03-21 05:29:41 +03:00
pos = pci_find_ht_capability ( dev , HT_CAPTYPE_SLAVE ) ;
if ( pos ! = 0 ) {
pci_dev_put ( dev ) ;
break ;
}
if ( ht_check_msi_mapping ( dev ) ) {
found = 1 ;
pci_dev_put ( dev ) ;
break ;
}
pci_dev_put ( dev ) ;
}
return found ;
}
2009-03-29 23:30:05 +04:00
# define PCI_HT_CAP_SLAVE_CTRL0 4 /* link control */
# define PCI_HT_CAP_SLAVE_CTRL1 8 /* link control to */
2012-07-10 01:36:14 +04:00
static int is_end_of_ht_chain ( struct pci_dev * dev )
2009-03-29 23:30:05 +04:00
{
int pos , ctrl_off ;
int end = 0 ;
u16 flags , ctrl ;
pos = pci_find_ht_capability ( dev , HT_CAPTYPE_SLAVE ) ;
if ( ! pos )
goto out ;
pci_read_config_word ( dev , pos + PCI_CAP_FLAGS , & flags ) ;
ctrl_off = ( ( flags > > 10 ) & 1 ) ?
PCI_HT_CAP_SLAVE_CTRL0 : PCI_HT_CAP_SLAVE_CTRL1 ;
pci_read_config_word ( dev , pos + ctrl_off , & ctrl ) ;
if ( ctrl & ( 1 < < 6 ) )
end = 1 ;
out :
return end ;
}
2012-07-10 01:36:14 +04:00
static void nv_ht_enable_msi_mapping ( struct pci_dev * dev )
2008-02-05 10:50:13 +03:00
{
struct pci_dev * host_bridge ;
2009-02-23 22:51:59 +03:00
int pos ;
int i , dev_no ;
int found = 0 ;
dev_no = dev - > devfn > > 3 ;
for ( i = dev_no ; i > = 0 ; i - - ) {
host_bridge = pci_get_slot ( dev - > bus , PCI_DEVFN ( i , 0 ) ) ;
if ( ! host_bridge )
continue ;
pos = pci_find_ht_capability ( host_bridge , HT_CAPTYPE_SLAVE ) ;
if ( pos ! = 0 ) {
found = 1 ;
break ;
}
pci_dev_put ( host_bridge ) ;
}
if ( ! found )
return ;
2009-03-29 23:30:05 +04:00
/* don't enable end_device/host_bridge with leaf directly here */
if ( host_bridge = = dev & & is_end_of_ht_chain ( host_bridge ) & &
host_bridge_with_leaf ( host_bridge ) )
2009-03-21 05:29:41 +03:00
goto out ;
2009-02-23 22:51:59 +03:00
/* root did that ! */
if ( msi_ht_cap_enabled ( host_bridge ) )
goto out ;
ht_enable_msi_mapping ( dev ) ;
out :
pci_dev_put ( host_bridge ) ;
}
2012-07-10 01:36:14 +04:00
static void ht_disable_msi_mapping ( struct pci_dev * dev )
2009-02-23 22:51:59 +03:00
{
2015-06-30 04:16:41 +03:00
int pos , ttl = PCI_FIND_CAP_TTL ;
2009-02-23 22:51:59 +03:00
pos = pci_find_ht_capability ( dev , HT_CAPTYPE_MSI_MAPPING ) ;
while ( pos & & ttl - - ) {
u8 flags ;
if ( pci_read_config_byte ( dev , pos + HT_MSI_FLAGS ,
& flags ) = = 0 ) {
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Disabling HT MSI Mapping \n " ) ;
2009-02-23 22:51:59 +03:00
pci_write_config_byte ( dev , pos + HT_MSI_FLAGS ,
flags & ~ HT_MSI_FLAGS_ENABLE ) ;
}
pos = pci_find_next_ht_capability ( dev , pos ,
HT_CAPTYPE_MSI_MAPPING ) ;
}
}
2012-07-10 01:36:14 +04:00
static void __nv_msi_ht_cap_quirk ( struct pci_dev * dev , int all )
2009-02-23 22:51:59 +03:00
{
struct pci_dev * host_bridge ;
int pos ;
int found ;
2010-07-24 00:19:55 +04:00
if ( ! pci_msi_enabled ( ) )
return ;
2009-02-23 22:51:59 +03:00
/* check if there is HT MSI cap or enabled on this device */
found = ht_check_msi_mapping ( dev ) ;
/* no HT MSI CAP */
if ( found = = 0 )
return ;
2008-02-05 10:50:13 +03:00
/*
* HT MSI mapping should be disabled on devices that are below
* a non - Hypertransport host bridge . Locate the host bridge . . .
*/
2017-12-19 08:37:53 +03:00
host_bridge = pci_get_domain_bus_and_slot ( pci_domain_nr ( dev - > bus ) , 0 ,
PCI_DEVFN ( 0 , 0 ) ) ;
2008-02-05 10:50:13 +03:00
if ( host_bridge = = NULL ) {
2018-01-18 21:55:24 +03:00
pci_warn ( dev , " nv_msi_ht_cap_quirk didn't locate host bridge \n " ) ;
2008-02-05 10:50:13 +03:00
return ;
}
pos = pci_find_ht_capability ( host_bridge , HT_CAPTYPE_SLAVE ) ;
if ( pos ! = 0 ) {
/* Host bridge is to HT */
2009-02-23 22:51:59 +03:00
if ( found = = 1 ) {
/* it is not enabled, try to enable it */
2009-03-21 05:29:41 +03:00
if ( all )
ht_enable_msi_mapping ( dev ) ;
else
nv_ht_enable_msi_mapping ( dev ) ;
2009-02-23 22:51:59 +03:00
}
2012-07-10 01:36:08 +04:00
goto out ;
2008-02-05 10:50:13 +03:00
}
2009-02-23 22:51:59 +03:00
/* HT MSI is not enabled */
if ( found = = 1 )
2012-07-10 01:36:08 +04:00
goto out ;
2008-02-05 10:50:13 +03:00
2009-02-23 22:51:59 +03:00
/* Host bridge is not to HT, disable HT MSI mapping on this device */
ht_disable_msi_mapping ( dev ) ;
2012-07-10 01:36:08 +04:00
out :
pci_dev_put ( host_bridge ) ;
2008-02-05 10:50:13 +03:00
}
2009-03-21 05:29:41 +03:00
2012-07-10 01:36:14 +04:00
static void nv_msi_ht_cap_quirk_all ( struct pci_dev * dev )
2009-03-21 05:29:41 +03:00
{
return __nv_msi_ht_cap_quirk ( dev , 1 ) ;
}
2018-05-02 16:53:19 +03:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_AL , PCI_ANY_ID , nv_msi_ht_cap_quirk_all ) ;
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_AL , PCI_ANY_ID , nv_msi_ht_cap_quirk_all ) ;
2009-03-21 05:29:41 +03:00
2012-07-10 01:36:14 +04:00
static void nv_msi_ht_cap_quirk_leaf ( struct pci_dev * dev )
2009-03-21 05:29:41 +03:00
{
return __nv_msi_ht_cap_quirk ( dev , 0 ) ;
}
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_NVIDIA , PCI_ANY_ID , nv_msi_ht_cap_quirk_leaf ) ;
2009-07-22 03:08:43 +04:00
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_NVIDIA , PCI_ANY_ID , nv_msi_ht_cap_quirk_leaf ) ;
2009-03-21 05:29:41 +03:00
2012-11-22 00:35:00 +04:00
static void quirk_msi_intx_disable_bug ( struct pci_dev * dev )
2007-10-25 12:16:30 +04:00
{
dev - > dev_flags | = PCI_DEV_FLAGS_MSI_INTX_DISABLE_BUG ;
}
2018-05-02 16:53:19 +03:00
2012-11-22 00:35:00 +04:00
static void quirk_msi_intx_disable_ati_bug ( struct pci_dev * dev )
2008-01-25 09:46:24 +03:00
{
struct pci_dev * p ;
2018-05-02 16:53:19 +03:00
/*
* SB700 MSI issue will be fixed at HW level from revision A21 ;
2008-01-25 09:46:24 +03:00
* we need check PCI REVISION ID of SMBus controller to get SB700
* revision .
*/
p = pci_get_device ( PCI_VENDOR_ID_ATI , PCI_DEVICE_ID_ATI_SBX00_SMBUS ,
NULL ) ;
if ( ! p )
return ;
if ( ( p - > revision < 0x3B ) & & ( p - > revision > = 0x30 ) )
dev - > dev_flags | = PCI_DEV_FLAGS_MSI_INTX_DISABLE_BUG ;
pci_dev_put ( p ) ;
}
2018-05-02 16:53:19 +03:00
2013-03-07 12:55:16 +04:00
static void quirk_msi_intx_disable_qca_bug ( struct pci_dev * dev )
{
/* AR816X/AR817X/E210X MSI is fixed at HW level from revision 0x18 */
if ( dev - > revision < 0x18 ) {
2018-01-18 21:55:24 +03:00
pci_info ( dev , " set MSI_INTX_DISABLE_BUG flag \n " ) ;
2013-03-07 12:55:16 +04:00
dev - > dev_flags | = PCI_DEV_FLAGS_MSI_INTX_DISABLE_BUG ;
}
}
2007-10-25 12:16:30 +04:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_BROADCOM ,
PCI_DEVICE_ID_TIGON3_5780 ,
quirk_msi_intx_disable_bug ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_BROADCOM ,
PCI_DEVICE_ID_TIGON3_5780S ,
quirk_msi_intx_disable_bug ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_BROADCOM ,
PCI_DEVICE_ID_TIGON3_5714 ,
quirk_msi_intx_disable_bug ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_BROADCOM ,
PCI_DEVICE_ID_TIGON3_5714S ,
quirk_msi_intx_disable_bug ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_BROADCOM ,
PCI_DEVICE_ID_TIGON3_5715 ,
quirk_msi_intx_disable_bug ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_BROADCOM ,
PCI_DEVICE_ID_TIGON3_5715S ,
quirk_msi_intx_disable_bug ) ;
2007-10-25 12:16:52 +04:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_ATI , 0x4390 ,
2008-01-25 09:46:24 +03:00
quirk_msi_intx_disable_ati_bug ) ;
2007-10-25 12:16:52 +04:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_ATI , 0x4391 ,
2008-01-25 09:46:24 +03:00
quirk_msi_intx_disable_ati_bug ) ;
2007-10-25 12:16:52 +04:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_ATI , 0x4392 ,
2008-01-25 09:46:24 +03:00
quirk_msi_intx_disable_ati_bug ) ;
2007-10-25 12:16:52 +04:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_ATI , 0x4393 ,
2008-01-25 09:46:24 +03:00
quirk_msi_intx_disable_ati_bug ) ;
2007-10-25 12:16:52 +04:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_ATI , 0x4394 ,
2008-01-25 09:46:24 +03:00
quirk_msi_intx_disable_ati_bug ) ;
2007-10-25 12:16:52 +04:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_ATI , 0x4373 ,
quirk_msi_intx_disable_bug ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_ATI , 0x4374 ,
quirk_msi_intx_disable_bug ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_ATI , 0x4375 ,
quirk_msi_intx_disable_bug ) ;
2012-04-30 19:38:49 +04:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_ATTANSIC , 0x1062 ,
quirk_msi_intx_disable_bug ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_ATTANSIC , 0x1063 ,
quirk_msi_intx_disable_bug ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_ATTANSIC , 0x2060 ,
quirk_msi_intx_disable_bug ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_ATTANSIC , 0x2062 ,
quirk_msi_intx_disable_bug ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_ATTANSIC , 0x1073 ,
quirk_msi_intx_disable_bug ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_ATTANSIC , 0x1083 ,
quirk_msi_intx_disable_bug ) ;
2013-03-07 12:55:16 +04:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_ATTANSIC , 0x1090 ,
quirk_msi_intx_disable_qca_bug ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_ATTANSIC , 0x1091 ,
quirk_msi_intx_disable_qca_bug ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_ATTANSIC , 0x10a0 ,
quirk_msi_intx_disable_qca_bug ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_ATTANSIC , 0x10a1 ,
quirk_msi_intx_disable_qca_bug ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_ATTANSIC , 0xe091 ,
quirk_msi_intx_disable_qca_bug ) ;
2019-09-12 16:00:42 +03:00
/*
* Amazon ' s Annapurna Labs 1 c36 : 0031 Root Ports don ' t support MSI - X , so it
* should be disabled on platforms where the device ( mistakenly ) advertises it .
*
* Notice that this quirk also disables MSI ( which may work , but hasn ' t been
* tested ) , since currently there is no standard way to disable only MSI - X .
*
* The 0031 device id is reused for other non Root Port device types ,
* therefore the quirk is registered for the PCI_CLASS_BRIDGE_PCI class .
*/
static void quirk_al_msi_disable ( struct pci_dev * dev )
{
dev - > no_msi = 1 ;
pci_warn ( dev , " Disabling MSI/MSI-X \n " ) ;
}
DECLARE_PCI_FIXUP_CLASS_FINAL ( PCI_VENDOR_ID_AMAZON_ANNAPURNA_LABS , 0x0031 ,
PCI_CLASS_BRIDGE_PCI , 8 , quirk_al_msi_disable ) ;
2006-08-31 09:54:56 +04:00
# endif /* CONFIG_PCI_MSI */
2008-08-19 12:28:24 +04:00
2018-05-02 16:53:19 +03:00
/*
* Allow manual resource allocation for PCI hotplug bridges via
* pci = hpmemsize = nnM and pci = hpiosize = nnM parameters . For some PCI - PCI
* hotplug bridges , like PLX 6254 ( former HINT HB6 ) , kernel fails to
* allocate resources when hotplug device is inserted and PCI bus is
* rescanned .
2010-03-28 17:02:02 +04:00
*/
2012-11-22 00:35:00 +04:00
static void quirk_hotplug_bridge ( struct pci_dev * dev )
2010-03-28 17:02:02 +04:00
{
dev - > is_hotplug_bridge = 1 ;
}
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_HINT , 0x0020 , quirk_hotplug_bridge ) ;
ricoh_mmc: port from driver to pci quirk
This patch solves nasty problem original driver has.
Original goal of the ricoh_mmc was to disable this device because then,
mmc cards can be read using standard SDHCI controller, thus avoiding
writing of yet another driver.
However, the act of disablement, makes other pci functions that belong to
this controller (xD and memstick) shift up one level, thus pci core has
now wrong idea about these devices.
To fix this issue, this patch moves the driver into the pci quirk section,
thus it is executes before the pci is enumerated, and therefore solving
that issue, also same sequence of commands is performed on resume for same
reasons.
Also regardless of the above, this way is cleaner. You still need to set
CONFIG_MMC_RICOH_MMC to enable this quirk
Signed-off-by: Maxim Levitsky <maximlevitsky@gmail.com>
Acked-by: Philip Langdale <philipl@overt.org>
Acked-by: Wolfram Sang <w.sang@pengutronix.de>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 00:43:20 +03:00
/*
2018-05-02 16:53:19 +03:00
* This is a quirk for the Ricoh MMC controller found as a part of some
* multifunction chips .
*
2011-03-31 05:57:33 +04:00
* This is very similar and based on the ricoh_mmc driver written by
ricoh_mmc: port from driver to pci quirk
This patch solves nasty problem original driver has.
Original goal of the ricoh_mmc was to disable this device because then,
mmc cards can be read using standard SDHCI controller, thus avoiding
writing of yet another driver.
However, the act of disablement, makes other pci functions that belong to
this controller (xD and memstick) shift up one level, thus pci core has
now wrong idea about these devices.
To fix this issue, this patch moves the driver into the pci quirk section,
thus it is executes before the pci is enumerated, and therefore solving
that issue, also same sequence of commands is performed on resume for same
reasons.
Also regardless of the above, this way is cleaner. You still need to set
CONFIG_MMC_RICOH_MMC to enable this quirk
Signed-off-by: Maxim Levitsky <maximlevitsky@gmail.com>
Acked-by: Philip Langdale <philipl@overt.org>
Acked-by: Wolfram Sang <w.sang@pengutronix.de>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 00:43:20 +03:00
* Philip Langdale . Thank you for these magic sequences .
*
2018-05-02 16:53:19 +03:00
* These chips implement the four main memory card controllers ( SD , MMC ,
* MS , xD ) and one or both of CardBus or FireWire .
ricoh_mmc: port from driver to pci quirk
This patch solves nasty problem original driver has.
Original goal of the ricoh_mmc was to disable this device because then,
mmc cards can be read using standard SDHCI controller, thus avoiding
writing of yet another driver.
However, the act of disablement, makes other pci functions that belong to
this controller (xD and memstick) shift up one level, thus pci core has
now wrong idea about these devices.
To fix this issue, this patch moves the driver into the pci quirk section,
thus it is executes before the pci is enumerated, and therefore solving
that issue, also same sequence of commands is performed on resume for same
reasons.
Also regardless of the above, this way is cleaner. You still need to set
CONFIG_MMC_RICOH_MMC to enable this quirk
Signed-off-by: Maxim Levitsky <maximlevitsky@gmail.com>
Acked-by: Philip Langdale <philipl@overt.org>
Acked-by: Wolfram Sang <w.sang@pengutronix.de>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 00:43:20 +03:00
*
2018-05-02 16:53:19 +03:00
* It happens that they implement SD and MMC support as separate
* controllers ( and PCI functions ) . The Linux SDHCI driver supports MMC
* cards but the chip detects MMC cards in hardware and directs them to the
* MMC controller - so the SDHCI driver never sees them .
ricoh_mmc: port from driver to pci quirk
This patch solves nasty problem original driver has.
Original goal of the ricoh_mmc was to disable this device because then,
mmc cards can be read using standard SDHCI controller, thus avoiding
writing of yet another driver.
However, the act of disablement, makes other pci functions that belong to
this controller (xD and memstick) shift up one level, thus pci core has
now wrong idea about these devices.
To fix this issue, this patch moves the driver into the pci quirk section,
thus it is executes before the pci is enumerated, and therefore solving
that issue, also same sequence of commands is performed on resume for same
reasons.
Also regardless of the above, this way is cleaner. You still need to set
CONFIG_MMC_RICOH_MMC to enable this quirk
Signed-off-by: Maxim Levitsky <maximlevitsky@gmail.com>
Acked-by: Philip Langdale <philipl@overt.org>
Acked-by: Wolfram Sang <w.sang@pengutronix.de>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 00:43:20 +03:00
*
2018-05-02 16:53:19 +03:00
* To get around this , we must disable the useless MMC controller . At that
* point , the SDHCI controller will start seeing them . It seems to be the
* case that the relevant PCI registers to deactivate the MMC controller
* live on PCI function 0 , which might be the CardBus controller or the
* FireWire controller , depending on the particular chip in question
ricoh_mmc: port from driver to pci quirk
This patch solves nasty problem original driver has.
Original goal of the ricoh_mmc was to disable this device because then,
mmc cards can be read using standard SDHCI controller, thus avoiding
writing of yet another driver.
However, the act of disablement, makes other pci functions that belong to
this controller (xD and memstick) shift up one level, thus pci core has
now wrong idea about these devices.
To fix this issue, this patch moves the driver into the pci quirk section,
thus it is executes before the pci is enumerated, and therefore solving
that issue, also same sequence of commands is performed on resume for same
reasons.
Also regardless of the above, this way is cleaner. You still need to set
CONFIG_MMC_RICOH_MMC to enable this quirk
Signed-off-by: Maxim Levitsky <maximlevitsky@gmail.com>
Acked-by: Philip Langdale <philipl@overt.org>
Acked-by: Wolfram Sang <w.sang@pengutronix.de>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 00:43:20 +03:00
*
* This has to be done early , because as soon as we disable the MMC controller
2018-05-02 16:53:19 +03:00
* other PCI functions shift up one level , e . g . function # 2 becomes function
* # 1 , and this will confuse the PCI core .
ricoh_mmc: port from driver to pci quirk
This patch solves nasty problem original driver has.
Original goal of the ricoh_mmc was to disable this device because then,
mmc cards can be read using standard SDHCI controller, thus avoiding
writing of yet another driver.
However, the act of disablement, makes other pci functions that belong to
this controller (xD and memstick) shift up one level, thus pci core has
now wrong idea about these devices.
To fix this issue, this patch moves the driver into the pci quirk section,
thus it is executes before the pci is enumerated, and therefore solving
that issue, also same sequence of commands is performed on resume for same
reasons.
Also regardless of the above, this way is cleaner. You still need to set
CONFIG_MMC_RICOH_MMC to enable this quirk
Signed-off-by: Maxim Levitsky <maximlevitsky@gmail.com>
Acked-by: Philip Langdale <philipl@overt.org>
Acked-by: Wolfram Sang <w.sang@pengutronix.de>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 00:43:20 +03:00
*/
# ifdef CONFIG_MMC_RICOH_MMC
static void ricoh_mmc_fixup_rl5c476 ( struct pci_dev * dev )
{
u8 write_enable ;
u8 write_target ;
u8 disable ;
2018-05-02 16:53:19 +03:00
/*
* Disable via CardBus interface
*
* This must be done via function # 0
*/
ricoh_mmc: port from driver to pci quirk
This patch solves nasty problem original driver has.
Original goal of the ricoh_mmc was to disable this device because then,
mmc cards can be read using standard SDHCI controller, thus avoiding
writing of yet another driver.
However, the act of disablement, makes other pci functions that belong to
this controller (xD and memstick) shift up one level, thus pci core has
now wrong idea about these devices.
To fix this issue, this patch moves the driver into the pci quirk section,
thus it is executes before the pci is enumerated, and therefore solving
that issue, also same sequence of commands is performed on resume for same
reasons.
Also regardless of the above, this way is cleaner. You still need to set
CONFIG_MMC_RICOH_MMC to enable this quirk
Signed-off-by: Maxim Levitsky <maximlevitsky@gmail.com>
Acked-by: Philip Langdale <philipl@overt.org>
Acked-by: Wolfram Sang <w.sang@pengutronix.de>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 00:43:20 +03:00
if ( PCI_FUNC ( dev - > devfn ) )
return ;
pci_read_config_byte ( dev , 0xB7 , & disable ) ;
if ( disable & 0x02 )
return ;
pci_read_config_byte ( dev , 0x8E , & write_enable ) ;
pci_write_config_byte ( dev , 0x8E , 0xAA ) ;
pci_read_config_byte ( dev , 0x8D , & write_target ) ;
pci_write_config_byte ( dev , 0x8D , 0xB7 ) ;
pci_write_config_byte ( dev , 0xB7 , disable | 0x02 ) ;
pci_write_config_byte ( dev , 0x8E , write_enable ) ;
pci_write_config_byte ( dev , 0x8D , write_target ) ;
2018-05-02 16:53:19 +03:00
pci_notice ( dev , " proprietary Ricoh MMC controller disabled (via CardBus function) \n " ) ;
2018-01-18 21:55:24 +03:00
pci_notice ( dev , " MMC cards are now supported by standard SDHCI controller \n " ) ;
ricoh_mmc: port from driver to pci quirk
This patch solves nasty problem original driver has.
Original goal of the ricoh_mmc was to disable this device because then,
mmc cards can be read using standard SDHCI controller, thus avoiding
writing of yet another driver.
However, the act of disablement, makes other pci functions that belong to
this controller (xD and memstick) shift up one level, thus pci core has
now wrong idea about these devices.
To fix this issue, this patch moves the driver into the pci quirk section,
thus it is executes before the pci is enumerated, and therefore solving
that issue, also same sequence of commands is performed on resume for same
reasons.
Also regardless of the above, this way is cleaner. You still need to set
CONFIG_MMC_RICOH_MMC to enable this quirk
Signed-off-by: Maxim Levitsky <maximlevitsky@gmail.com>
Acked-by: Philip Langdale <philipl@overt.org>
Acked-by: Wolfram Sang <w.sang@pengutronix.de>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 00:43:20 +03:00
}
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_RICOH , PCI_DEVICE_ID_RICOH_RL5C476 , ricoh_mmc_fixup_rl5c476 ) ;
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_RICOH , PCI_DEVICE_ID_RICOH_RL5C476 , ricoh_mmc_fixup_rl5c476 ) ;
static void ricoh_mmc_fixup_r5c832 ( struct pci_dev * dev )
{
u8 write_enable ;
u8 disable ;
2018-05-02 16:53:19 +03:00
/*
* Disable via FireWire interface
*
* This must be done via function # 0
*/
ricoh_mmc: port from driver to pci quirk
This patch solves nasty problem original driver has.
Original goal of the ricoh_mmc was to disable this device because then,
mmc cards can be read using standard SDHCI controller, thus avoiding
writing of yet another driver.
However, the act of disablement, makes other pci functions that belong to
this controller (xD and memstick) shift up one level, thus pci core has
now wrong idea about these devices.
To fix this issue, this patch moves the driver into the pci quirk section,
thus it is executes before the pci is enumerated, and therefore solving
that issue, also same sequence of commands is performed on resume for same
reasons.
Also regardless of the above, this way is cleaner. You still need to set
CONFIG_MMC_RICOH_MMC to enable this quirk
Signed-off-by: Maxim Levitsky <maximlevitsky@gmail.com>
Acked-by: Philip Langdale <philipl@overt.org>
Acked-by: Wolfram Sang <w.sang@pengutronix.de>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 00:43:20 +03:00
if ( PCI_FUNC ( dev - > devfn ) )
return ;
2011-07-12 01:28:35 +04:00
/*
2012-12-02 00:37:20 +04:00
* RICOH 0xe822 and 0xe823 SD / MMC card readers fail to recognize
2018-05-02 16:53:19 +03:00
* certain types of SD / MMC cards . Lowering the SD base clock
* frequency from 200 Mhz to 50 Mhz fixes this issue .
2011-07-12 01:28:35 +04:00
*
* 0x150 - SD2 .0 mode enable for changing base clock
* frequency to 50 Mhz
* 0xe1 - Base clock frequency
* 0x32 - 50 Mhz new clock frequency
* 0xf9 - Key register for 0x150
* 0xfc - key register for 0xe1
*/
2012-12-02 00:37:20 +04:00
if ( dev - > device = = PCI_DEVICE_ID_RICOH_R5CE822 | |
dev - > device = = PCI_DEVICE_ID_RICOH_R5CE823 ) {
2011-07-12 01:28:35 +04:00
pci_write_config_byte ( dev , 0xf9 , 0xfc ) ;
pci_write_config_byte ( dev , 0x150 , 0x10 ) ;
pci_write_config_byte ( dev , 0xf9 , 0x00 ) ;
pci_write_config_byte ( dev , 0xfc , 0x01 ) ;
pci_write_config_byte ( dev , 0xe1 , 0x32 ) ;
pci_write_config_byte ( dev , 0xfc , 0x00 ) ;
2018-01-18 21:55:24 +03:00
pci_notice ( dev , " MMC controller base frequency changed to 50Mhz. \n " ) ;
2011-07-12 01:28:35 +04:00
}
2011-10-05 19:44:50 +04:00
pci_read_config_byte ( dev , 0xCB , & disable ) ;
if ( disable & 0x02 )
return ;
pci_read_config_byte ( dev , 0xCA , & write_enable ) ;
pci_write_config_byte ( dev , 0xCA , 0x57 ) ;
pci_write_config_byte ( dev , 0xCB , disable | 0x02 ) ;
pci_write_config_byte ( dev , 0xCA , write_enable ) ;
2018-05-02 16:53:19 +03:00
pci_notice ( dev , " proprietary Ricoh MMC controller disabled (via FireWire function) \n " ) ;
2018-01-18 21:55:24 +03:00
pci_notice ( dev , " MMC cards are now supported by standard SDHCI controller \n " ) ;
2011-10-05 19:44:50 +04:00
ricoh_mmc: port from driver to pci quirk
This patch solves nasty problem original driver has.
Original goal of the ricoh_mmc was to disable this device because then,
mmc cards can be read using standard SDHCI controller, thus avoiding
writing of yet another driver.
However, the act of disablement, makes other pci functions that belong to
this controller (xD and memstick) shift up one level, thus pci core has
now wrong idea about these devices.
To fix this issue, this patch moves the driver into the pci quirk section,
thus it is executes before the pci is enumerated, and therefore solving
that issue, also same sequence of commands is performed on resume for same
reasons.
Also regardless of the above, this way is cleaner. You still need to set
CONFIG_MMC_RICOH_MMC to enable this quirk
Signed-off-by: Maxim Levitsky <maximlevitsky@gmail.com>
Acked-by: Philip Langdale <philipl@overt.org>
Acked-by: Wolfram Sang <w.sang@pengutronix.de>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 00:43:20 +03:00
}
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_RICOH , PCI_DEVICE_ID_RICOH_R5C832 , ricoh_mmc_fixup_r5c832 ) ;
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_RICOH , PCI_DEVICE_ID_RICOH_R5C832 , ricoh_mmc_fixup_r5c832 ) ;
2012-12-02 00:37:20 +04:00
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_RICOH , PCI_DEVICE_ID_RICOH_R5CE822 , ricoh_mmc_fixup_r5c832 ) ;
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_RICOH , PCI_DEVICE_ID_RICOH_R5CE822 , ricoh_mmc_fixup_r5c832 ) ;
2011-05-26 20:19:05 +04:00
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_RICOH , PCI_DEVICE_ID_RICOH_R5CE823 , ricoh_mmc_fixup_r5c832 ) ;
DECLARE_PCI_FIXUP_RESUME_EARLY ( PCI_VENDOR_ID_RICOH , PCI_DEVICE_ID_RICOH_R5CE823 , ricoh_mmc_fixup_r5c832 ) ;
ricoh_mmc: port from driver to pci quirk
This patch solves nasty problem original driver has.
Original goal of the ricoh_mmc was to disable this device because then,
mmc cards can be read using standard SDHCI controller, thus avoiding
writing of yet another driver.
However, the act of disablement, makes other pci functions that belong to
this controller (xD and memstick) shift up one level, thus pci core has
now wrong idea about these devices.
To fix this issue, this patch moves the driver into the pci quirk section,
thus it is executes before the pci is enumerated, and therefore solving
that issue, also same sequence of commands is performed on resume for same
reasons.
Also regardless of the above, this way is cleaner. You still need to set
CONFIG_MMC_RICOH_MMC to enable this quirk
Signed-off-by: Maxim Levitsky <maximlevitsky@gmail.com>
Acked-by: Philip Langdale <philipl@overt.org>
Acked-by: Wolfram Sang <w.sang@pengutronix.de>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 00:43:20 +03:00
# endif /*CONFIG_MMC_RICOH_MMC*/
2011-08-24 04:05:25 +04:00
# ifdef CONFIG_DMAR_TABLE
2010-12-06 23:26:30 +03:00
# define VTUNCERRMSK_REG 0x1ac
# define VTD_MSK_SPEC_ERRORS (1 << 31)
/*
2018-05-02 16:53:19 +03:00
* This is a quirk for masking VT - d spec - defined errors to platform error
* handling logic . Without this , platforms using Intel 7500 , 5500 chipsets
2010-12-06 23:26:30 +03:00
* ( and the derivative chipsets like X58 etc ) seem to generate NMI / SMI ( based
2018-05-02 16:53:19 +03:00
* on the RAS config settings of the platform ) when a VT - d fault happens .
2010-12-06 23:26:30 +03:00
* The resulting SMI caused the system to hang .
*
2018-05-02 16:53:19 +03:00
* VT - d spec - related errors are already handled by the VT - d OS code , so no
2010-12-06 23:26:30 +03:00
* need to report the same error through other channels .
*/
static void vtd_mask_spec_errors ( struct pci_dev * dev )
{
u32 word ;
pci_read_config_dword ( dev , VTUNCERRMSK_REG , & word ) ;
pci_write_config_dword ( dev , VTUNCERRMSK_REG , word | VTD_MSK_SPEC_ERRORS ) ;
}
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_INTEL , 0x342e , vtd_mask_spec_errors ) ;
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_INTEL , 0x3c28 , vtd_mask_spec_errors ) ;
# endif
ricoh_mmc: port from driver to pci quirk
This patch solves nasty problem original driver has.
Original goal of the ricoh_mmc was to disable this device because then,
mmc cards can be read using standard SDHCI controller, thus avoiding
writing of yet another driver.
However, the act of disablement, makes other pci functions that belong to
this controller (xD and memstick) shift up one level, thus pci core has
now wrong idea about these devices.
To fix this issue, this patch moves the driver into the pci quirk section,
thus it is executes before the pci is enumerated, and therefore solving
that issue, also same sequence of commands is performed on resume for same
reasons.
Also regardless of the above, this way is cleaner. You still need to set
CONFIG_MMC_RICOH_MMC to enable this quirk
Signed-off-by: Maxim Levitsky <maximlevitsky@gmail.com>
Acked-by: Philip Langdale <philipl@overt.org>
Acked-by: Wolfram Sang <w.sang@pengutronix.de>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 00:43:20 +03:00
2012-11-22 00:35:00 +04:00
static void fixup_ti816x_class ( struct pci_dev * dev )
2011-04-05 11:02:50 +04:00
{
2015-06-19 23:58:24 +03:00
u32 class = dev - > class ;
2011-04-05 11:02:50 +04:00
/* TI 816x devices do not have class code set when in PCIe boot mode */
2015-06-19 23:58:24 +03:00
dev - > class = PCI_CLASS_MULTIMEDIA_VIDEO < < 8 ;
2018-01-18 21:55:24 +03:00
pci_info ( dev , " PCI class overridden (%#08x -> %#08x) \n " ,
2015-06-19 23:58:24 +03:00
class , dev - > class ) ;
2011-04-05 11:02:50 +04:00
}
2012-02-24 11:46:58 +04:00
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_TI , 0xb800 ,
2015-06-20 00:20:58 +03:00
PCI_CLASS_NOT_DEFINED , 8 , fixup_ti816x_class ) ;
2011-04-05 11:02:50 +04:00
2018-05-02 16:53:19 +03:00
/*
* Some PCIe devices do not work reliably with the claimed maximum
2011-10-06 01:35:03 +04:00
* payload size supported .
*/
2012-11-22 00:35:00 +04:00
static void fixup_mpss_256 ( struct pci_dev * dev )
2011-10-06 01:35:03 +04:00
{
dev - > pcie_mpss = 1 ; /* 256 bytes */
}
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_SOLARFLARE ,
PCI_DEVICE_ID_SOLARFLARE_SFC4000A_0 , fixup_mpss_256 ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_SOLARFLARE ,
PCI_DEVICE_ID_SOLARFLARE_SFC4000A_1 , fixup_mpss_256 ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_SOLARFLARE ,
PCI_DEVICE_ID_SOLARFLARE_SFC4000B , fixup_mpss_256 ) ;
2018-05-02 16:53:19 +03:00
/*
* Intel 5000 and 5100 Memory controllers have an erratum with read completion
2011-10-14 23:56:13 +04:00
* coalescing ( which is enabled by default on some BIOSes ) and MPS of 256 B .
2018-05-02 16:53:19 +03:00
* Since there is no way of knowing what the PCIe MPS on each fabric will be
2011-10-14 23:56:13 +04:00
* until all of the devices are discovered and buses walked , read completion
* coalescing must be disabled . Unfortunately , it cannot be re - enabled because
* it is possible to hotplug a device with MPS of 256 B .
*/
2012-11-22 00:35:00 +04:00
static void quirk_intel_mc_errata ( struct pci_dev * dev )
2011-10-14 23:56:13 +04:00
{
int err ;
u16 rcc ;
PCI: Set MPS to match upstream bridge
Firmware typically configures the PCIe fabric with a consistent Max Payload
Size setting based on the devices present at boot. A hot-added device
typically has the power-on default MPS setting (128 bytes), which may not
match the fabric.
The previous Linux default, in the absence of any "pci=pcie_bus_*" options,
was PCIE_BUS_TUNE_OFF, in which we never touch MPS, even for hot-added
devices.
Add a new default setting, PCIE_BUS_DEFAULT, in which we make sure every
device's MPS setting matches the upstream bridge. This makes it more
likely that a hot-added device will work in a system with optimized MPS
configuration.
Note that if we hot-add a device that only supports 128-byte MPS, it still
likely won't work because we don't reconfigure the rest of the fabric.
Booting with "pci=pcie_bus_peer2peer" is a workaround for this because it
sets MPS to 128 for everything.
[bhelgaas: changelog, new default, rework for pci_configure_device() path]
Tested-by: Keith Busch <keith.busch@intel.com>
Tested-by: Jordan Hargrave <jharg93@gmail.com>
Signed-off-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-by: Yinghai Lu <yinghai@kernel.org>
2015-08-24 16:48:16 +03:00
if ( pcie_bus_config = = PCIE_BUS_TUNE_OFF | |
pcie_bus_config = = PCIE_BUS_DEFAULT )
2011-10-14 23:56:13 +04:00
return ;
2018-05-02 16:53:19 +03:00
/*
* Intel erratum specifies bits to change but does not say what
* they are . Keeping them magical until such time as the registers
* and values can be explained .
2011-10-14 23:56:13 +04:00
*/
err = pci_read_config_word ( dev , 0x48 , & rcc ) ;
if ( err ) {
2018-01-18 21:55:24 +03:00
pci_err ( dev , " Error attempting to read the read completion coalescing register \n " ) ;
2011-10-14 23:56:13 +04:00
return ;
}
if ( ! ( rcc & ( 1 < < 10 ) ) )
return ;
rcc & = ~ ( 1 < < 10 ) ;
err = pci_write_config_word ( dev , 0x48 , rcc ) ;
if ( err ) {
2018-01-18 21:55:24 +03:00
pci_err ( dev , " Error attempting to write the read completion coalescing register \n " ) ;
2011-10-14 23:56:13 +04:00
return ;
}
2018-05-02 16:53:19 +03:00
pr_info_once ( " Read completion coalescing disabled due to hardware erratum relating to 256B MPS \n " ) ;
2011-10-14 23:56:13 +04:00
}
/* Intel 5000 series memory controllers and ports 2-7 */
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x25c0 , quirk_intel_mc_errata ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x25d0 , quirk_intel_mc_errata ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x25d4 , quirk_intel_mc_errata ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x25d8 , quirk_intel_mc_errata ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x25e2 , quirk_intel_mc_errata ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x25e3 , quirk_intel_mc_errata ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x25e4 , quirk_intel_mc_errata ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x25e5 , quirk_intel_mc_errata ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x25e6 , quirk_intel_mc_errata ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x25e7 , quirk_intel_mc_errata ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x25f7 , quirk_intel_mc_errata ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x25f8 , quirk_intel_mc_errata ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x25f9 , quirk_intel_mc_errata ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x25fa , quirk_intel_mc_errata ) ;
/* Intel 5100 series memory controllers and ports 2-7 */
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x65c0 , quirk_intel_mc_errata ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x65e2 , quirk_intel_mc_errata ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x65e3 , quirk_intel_mc_errata ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x65e4 , quirk_intel_mc_errata ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x65e5 , quirk_intel_mc_errata ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x65e6 , quirk_intel_mc_errata ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x65e7 , quirk_intel_mc_errata ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x65f7 , quirk_intel_mc_errata ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x65f8 , quirk_intel_mc_errata ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x65f9 , quirk_intel_mc_errata ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x65fa , quirk_intel_mc_errata ) ;
2013-05-06 12:03:33 +04:00
/*
2018-05-02 16:53:19 +03:00
* Ivytown NTB BAR sizes are misreported by the hardware due to an erratum .
* To work around this , query the size it should be configured to by the
* device and modify the resource end to correspond to this new size .
2013-05-06 12:03:33 +04:00
*/
static void quirk_intel_ntb ( struct pci_dev * dev )
{
int rc ;
u8 val ;
rc = pci_read_config_byte ( dev , 0x00D0 , & val ) ;
if ( rc )
return ;
dev - > resource [ 2 ] . end = dev - > resource [ 2 ] . start + ( ( u64 ) 1 < < val ) - 1 ;
rc = pci_read_config_byte ( dev , 0x00D1 , & val ) ;
if ( rc )
return ;
dev - > resource [ 4 ] . end = dev - > resource [ 4 ] . start + ( ( u64 ) 1 < < val ) - 1 ;
}
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x0e08 , quirk_intel_ntb ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x0e0d , quirk_intel_ntb ) ;
2011-12-08 01:08:11 +04:00
/*
2018-05-02 16:53:19 +03:00
* Some BIOS implementations leave the Intel GPU interrupts enabled , even
* though no one is handling them ( e . g . , if the i915 driver is never
* loaded ) . Additionally the interrupt destination is not set up properly
2011-12-08 01:08:11 +04:00
* and the interrupt ends up - somewhere - .
*
2018-05-02 16:53:19 +03:00
* These spurious interrupts are " sticky " and the kernel disables the
* ( shared ) interrupt line after 100 , 000 + generated interrupts .
2011-12-08 01:08:11 +04:00
*
2018-05-02 16:53:19 +03:00
* Fix it by disabling the still enabled interrupts . This resolves crashes
* often seen on monitor unplug .
2011-12-08 01:08:11 +04:00
*/
# define I915_DEIER_REG 0x4400c
2012-11-22 00:35:00 +04:00
static void disable_igfx_irq ( struct pci_dev * dev )
2011-12-08 01:08:11 +04:00
{
void __iomem * regs = pci_iomap ( dev , 0 , 0 ) ;
if ( regs = = NULL ) {
2018-01-18 21:55:24 +03:00
pci_warn ( dev , " igfx quirk: Can't iomap PCI device \n " ) ;
2011-12-08 01:08:11 +04:00
return ;
}
/* Check if any interrupt line is still enabled */
if ( readl ( regs + I915_DEIER_REG ) ! = 0 ) {
2018-01-18 21:55:24 +03:00
pci_warn ( dev , " BIOS left Intel GPU interrupts enabled; disabling \n " ) ;
2011-12-08 01:08:11 +04:00
writel ( 0 , regs + I915_DEIER_REG ) ;
}
pci_iounmap ( dev , regs ) ;
}
2018-09-26 18:14:01 +03:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x0042 , disable_igfx_irq ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x0046 , disable_igfx_irq ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x004a , disable_igfx_irq ) ;
2011-12-08 01:08:11 +04:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x0102 , disable_igfx_irq ) ;
2018-09-26 18:14:01 +03:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x0106 , disable_igfx_irq ) ;
2011-12-08 01:08:11 +04:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x010a , disable_igfx_irq ) ;
2014-04-07 17:10:32 +04:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x0152 , disable_igfx_irq ) ;
2011-12-08 01:08:11 +04:00
2013-09-11 03:10:43 +04:00
/*
* PCI devices which are on Intel chips can skip the 10 ms delay
* before entering D3 mode .
*/
2020-07-31 00:08:48 +03:00
static void quirk_remove_d3hot_delay ( struct pci_dev * dev )
{
dev - > d3hot_delay = 0 ;
}
/* C600 Series devices do not need 10ms d3hot_delay */
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x0412 , quirk_remove_d3hot_delay ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x0c00 , quirk_remove_d3hot_delay ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x0c0c , quirk_remove_d3hot_delay ) ;
/* Lynxpoint-H PCH devices do not need 10ms d3hot_delay */
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x8c02 , quirk_remove_d3hot_delay ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x8c18 , quirk_remove_d3hot_delay ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x8c1c , quirk_remove_d3hot_delay ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x8c20 , quirk_remove_d3hot_delay ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x8c22 , quirk_remove_d3hot_delay ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x8c26 , quirk_remove_d3hot_delay ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x8c2d , quirk_remove_d3hot_delay ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x8c31 , quirk_remove_d3hot_delay ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x8c3a , quirk_remove_d3hot_delay ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x8c3d , quirk_remove_d3hot_delay ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x8c4e , quirk_remove_d3hot_delay ) ;
/* Intel Cherrytrail devices do not need 10ms d3hot_delay */
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x2280 , quirk_remove_d3hot_delay ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x2298 , quirk_remove_d3hot_delay ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x229c , quirk_remove_d3hot_delay ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x22b0 , quirk_remove_d3hot_delay ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x22b5 , quirk_remove_d3hot_delay ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x22b7 , quirk_remove_d3hot_delay ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x22b8 , quirk_remove_d3hot_delay ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x22d8 , quirk_remove_d3hot_delay ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x22dc , quirk_remove_d3hot_delay ) ;
2016-11-15 10:59:59 +03:00
2012-06-17 00:40:22 +04:00
/*
2016-11-15 10:59:59 +03:00
* Some devices may pass our check in pci_intx_mask_supported ( ) if
2012-06-17 00:40:22 +04:00
* PCI_COMMAND_INTX_DISABLE works though they actually do not properly
* support this feature .
*/
2012-11-22 00:35:00 +04:00
static void quirk_broken_intx_masking ( struct pci_dev * dev )
2012-06-17 00:40:22 +04:00
{
dev - > broken_intx_masking = 1 ;
}
2016-11-15 10:59:58 +03:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_CHELSIO , 0x0030 ,
quirk_broken_intx_masking ) ;
DECLARE_PCI_FIXUP_FINAL ( 0x1814 , 0x0601 , /* Ralink RT2800 802.11n PCI */
quirk_broken_intx_masking ) ;
2017-12-15 23:51:44 +03:00
DECLARE_PCI_FIXUP_FINAL ( 0x1b7c , 0x0004 , /* Ceton InfiniTV4 */
quirk_broken_intx_masking ) ;
2016-11-15 10:59:59 +03:00
2014-05-02 00:36:31 +04:00
/*
* Realtek RTL8169 PCI Gigabit Ethernet Controller ( rev 10 )
* Subsystem : Realtek RTL8169 / 8110 Family PCI Gigabit Ethernet NIC
*
* RTL8110SC - Fails under PCI device assignment using DisINTx masking .
*/
2016-11-15 10:59:58 +03:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_REALTEK , 0x8169 ,
quirk_broken_intx_masking ) ;
2012-06-17 00:40:22 +04:00
2016-03-24 22:03:49 +03:00
/*
* Intel i40e ( XL710 / X710 ) 10 / 20 / 40 GbE NICs all have broken INTx masking ,
* DisINTx can be set but the interrupt status bit is non - functional .
*/
2018-05-02 16:53:19 +03:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x1572 , quirk_broken_intx_masking ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x1574 , quirk_broken_intx_masking ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x1580 , quirk_broken_intx_masking ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x1581 , quirk_broken_intx_masking ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x1583 , quirk_broken_intx_masking ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x1584 , quirk_broken_intx_masking ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x1585 , quirk_broken_intx_masking ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x1586 , quirk_broken_intx_masking ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x1587 , quirk_broken_intx_masking ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x1588 , quirk_broken_intx_masking ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x1589 , quirk_broken_intx_masking ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x158a , quirk_broken_intx_masking ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x158b , quirk_broken_intx_masking ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x37d0 , quirk_broken_intx_masking ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x37d1 , quirk_broken_intx_masking ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , 0x37d2 , quirk_broken_intx_masking ) ;
2016-03-24 22:03:49 +03:00
2016-11-15 10:59:59 +03:00
static u16 mellanox_broken_intx_devs [ ] = {
PCI_DEVICE_ID_MELLANOX_HERMON_SDR ,
PCI_DEVICE_ID_MELLANOX_HERMON_DDR ,
PCI_DEVICE_ID_MELLANOX_HERMON_QDR ,
PCI_DEVICE_ID_MELLANOX_HERMON_DDR_GEN2 ,
PCI_DEVICE_ID_MELLANOX_HERMON_QDR_GEN2 ,
PCI_DEVICE_ID_MELLANOX_HERMON_EN ,
PCI_DEVICE_ID_MELLANOX_HERMON_EN_GEN2 ,
PCI_DEVICE_ID_MELLANOX_CONNECTX_EN ,
PCI_DEVICE_ID_MELLANOX_CONNECTX_EN_T_GEN2 ,
PCI_DEVICE_ID_MELLANOX_CONNECTX_EN_GEN2 ,
PCI_DEVICE_ID_MELLANOX_CONNECTX_EN_5_GEN2 ,
PCI_DEVICE_ID_MELLANOX_CONNECTX2 ,
PCI_DEVICE_ID_MELLANOX_CONNECTX3 ,
PCI_DEVICE_ID_MELLANOX_CONNECTX3_PRO ,
} ;
2016-11-15 11:00:00 +03:00
# define CONNECTX_4_CURR_MAX_MINOR 99
# define CONNECTX_4_INTX_SUPPORT_MINOR 14
/*
* Check ConnectX - 4 / LX FW version to see if it supports legacy interrupts .
* If so , don ' t mark it as broken .
* FW minor > 99 means older FW version format and no INTx masking support .
* FW minor < 14 means new FW version format and no INTx masking support .
*/
2016-11-15 10:59:59 +03:00
static void mellanox_check_broken_intx_masking ( struct pci_dev * pdev )
{
2016-11-15 11:00:00 +03:00
__be32 __iomem * fw_ver ;
u16 fw_major ;
u16 fw_minor ;
u16 fw_subminor ;
u32 fw_maj_min ;
u32 fw_sub_min ;
2016-11-15 10:59:59 +03:00
int i ;
for ( i = 0 ; i < ARRAY_SIZE ( mellanox_broken_intx_devs ) ; i + + ) {
if ( pdev - > device = = mellanox_broken_intx_devs [ i ] ) {
pdev - > broken_intx_masking = 1 ;
return ;
}
}
2016-11-15 11:00:00 +03:00
2018-05-02 16:53:19 +03:00
/*
* Getting here means Connect - IB cards and up . Connect - IB has no INTx
2016-11-15 11:00:00 +03:00
* support so shouldn ' t be checked further
*/
if ( pdev - > device = = PCI_DEVICE_ID_MELLANOX_CONNECTIB )
return ;
if ( pdev - > device ! = PCI_DEVICE_ID_MELLANOX_CONNECTX4 & &
pdev - > device ! = PCI_DEVICE_ID_MELLANOX_CONNECTX4_LX )
return ;
/* For ConnectX-4 and ConnectX-4LX, need to check FW support */
if ( pci_enable_device_mem ( pdev ) ) {
2018-01-18 21:55:24 +03:00
pci_warn ( pdev , " Can't enable device memory \n " ) ;
2016-11-15 11:00:00 +03:00
return ;
}
fw_ver = ioremap ( pci_resource_start ( pdev , 0 ) , 4 ) ;
if ( ! fw_ver ) {
2018-01-18 21:55:24 +03:00
pci_warn ( pdev , " Can't map ConnectX-4 initialization segment \n " ) ;
2016-11-15 11:00:00 +03:00
goto out ;
}
/* Reading from resource space should be 32b aligned */
fw_maj_min = ioread32be ( fw_ver ) ;
fw_sub_min = ioread32be ( fw_ver + 1 ) ;
fw_major = fw_maj_min & 0xffff ;
fw_minor = fw_maj_min > > 16 ;
fw_subminor = fw_sub_min & 0xffff ;
if ( fw_minor > CONNECTX_4_CURR_MAX_MINOR | |
fw_minor < CONNECTX_4_INTX_SUPPORT_MINOR ) {
2018-01-18 21:55:24 +03:00
pci_warn ( pdev , " ConnectX-4: FW %u.%u.%u doesn't support INTx masking, disabling. Please upgrade FW to %d.14.1100 and up for INTx support \n " ,
2016-11-15 11:00:00 +03:00
fw_major , fw_minor , fw_subminor , pdev - > device = =
PCI_DEVICE_ID_MELLANOX_CONNECTX4 ? 12 : 14 ) ;
pdev - > broken_intx_masking = 1 ;
}
iounmap ( fw_ver ) ;
out :
pci_disable_device ( pdev ) ;
2016-11-15 10:59:59 +03:00
}
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_MELLANOX , PCI_ANY_ID ,
mellanox_check_broken_intx_masking ) ;
2016-03-24 22:03:49 +03:00
2015-01-16 03:17:12 +03:00
static void quirk_no_bus_reset ( struct pci_dev * dev )
{
dev - > dev_flags | = PCI_DEV_FLAGS_NO_BUS_RESET ;
}
/*
2016-05-30 15:26:37 +03:00
* Some Atheros AR9xxx and QCA988x chips do not behave after a bus reset .
* The device will throw a Link Down error on AER - capable systems and
* regardless of AER , config space of the device is never accessible again
* and typically causes the system to hang or reset when access is attempted .
2020-06-30 20:41:39 +03:00
* https : //lore.kernel.org/r/20140923210318.498dacbd@dualc.maya.org/
2015-01-16 03:17:12 +03:00
*/
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_ATHEROS , 0x0030 , quirk_no_bus_reset ) ;
2016-05-30 15:26:37 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_ATHEROS , 0x0032 , quirk_no_bus_reset ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_ATHEROS , 0x003c , quirk_no_bus_reset ) ;
2016-08-09 17:41:31 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_ATHEROS , 0x0033 , quirk_no_bus_reset ) ;
2019-01-08 00:32:48 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_ATHEROS , 0x0034 , quirk_no_bus_reset ) ;
2015-01-16 03:17:12 +03:00
2017-09-08 11:10:32 +03:00
/*
* Root port on some Cavium CN8xxx chips do not successfully complete a bus
* reset when used with certain child devices . After the reset , config
* accesses to the child may fail .
*/
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_CAVIUM , 0xa100 , quirk_no_bus_reset ) ;
2014-11-21 21:24:14 +03:00
static void quirk_no_pm_reset ( struct pci_dev * dev )
{
/*
* We can ' t do a bus reset on root bus devices , but an ineffective
* PM reset may be better than nothing .
*/
if ( ! pci_is_root_bus ( dev - > bus ) )
dev - > dev_flags | = PCI_DEV_FLAGS_NO_PM_RESET ;
}
/*
* Some AMD / ATI GPUS ( HD8570 - Oland ) report that a D3hot - > D0 transition
* causes a reset ( i . e . , they advertise NoSoftRst - ) . This transition seems
* to have no effect on the device : it retains the framebuffer contents and
* monitor sync . Advertising this support makes other layers , like VFIO ,
* assume pci_reset_function ( ) is viable for this device . Mark it as
* unavailable to skip it when testing reset methods .
*/
DECLARE_PCI_FIXUP_CLASS_HEADER ( PCI_VENDOR_ID_ATI , PCI_ANY_ID ,
PCI_CLASS_DISPLAY_VGA , 8 , quirk_no_pm_reset ) ;
thunderbolt: Support 1st gen Light Ridge controller
Add support for the 1st gen Light Ridge controller, which is built into
these systems:
iMac12,1 2011 21.5"
iMac12,2 2011 27"
Macmini5,1 2011 i5 2.3 GHz
Macmini5,2 2011 i5 2.5 GHz
Macmini5,3 2011 i7 2.0 GHz
MacBookPro8,1 2011 13"
MacBookPro8,2 2011 15"
MacBookPro8,3 2011 17"
MacBookPro9,1 2012 15"
MacBookPro9,2 2012 13"
Light Ridge (CV82524) was the very first copper Thunderbolt controller,
introduced 2010 alongside its fiber-optic cousin Light Peak (CVL2510).
Consequently the chip suffers from some teething troubles:
- MSI is broken for hotplug signaling on the downstream bridges: The chip
just never sends an interrupt. It requests 32 MSIs for each of its six
bridges and the pcieport driver only allocates one per bridge. However
I've verified that even if 32 MSIs are allocated there's no interrupt
on hotplug. The only option is thus to disable MSI, which is also what
OS X does. Apparently all Thunderbolt chips up to revision 1 of Cactus
Ridge 4C are plagued by this issue so quirk those as well.
- The chip supports a maximum hop_count of 32, unlike its successors
which support only 12. Fixup ring_interrupt_active() to cope with
values >= 32.
- Another peculiarity is that the chip supports a maximum of 13 ports
whereas its successors support 12. However the additional port (#5)
seems to be unusable as reading its TB_CFG_PORT config space results in
TB_CFG_ERROR_INVALID_CONFIG_SPACE. Add a quirk to mark the port
disabled on the root switch, assuming that's necessary on all Macs
using this chip.
Tested-by: Lukas Wunner <lukas@wunner.de> [MacBookPro9,1]
Tested-by: William Brown <william@blackhats.net.au> [MacBookPro8,2]
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-by: Andreas Noever <andreas.noever@gmail.com>
2016-03-20 15:57:20 +03:00
/*
* Thunderbolt controllers with broken MSI hotplug signaling :
* Entire 1 st generation ( Light Ridge , Eagle Ridge , Light Peak ) and part
* of the 2 nd generation ( Cactus Ridge 4 C up to revision 1 , Port Ridge ) .
*/
static void quirk_thunderbolt_hotplug_msi ( struct pci_dev * pdev )
{
if ( pdev - > is_hotplug_bridge & &
( pdev - > device ! = PCI_DEVICE_ID_INTEL_CACTUS_RIDGE_4C | |
pdev - > revision < = 1 ) )
pdev - > no_msi = 1 ;
}
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_LIGHT_RIDGE ,
quirk_thunderbolt_hotplug_msi ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_EAGLE_RIDGE ,
quirk_thunderbolt_hotplug_msi ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_LIGHT_PEAK ,
quirk_thunderbolt_hotplug_msi ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_CACTUS_RIDGE_4C ,
quirk_thunderbolt_hotplug_msi ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_PORT_RIDGE ,
quirk_thunderbolt_hotplug_msi ) ;
2014-06-04 00:04:10 +04:00
# ifdef CONFIG_ACPI
/*
* Apple : Shutdown Cactus Ridge Thunderbolt controller .
*
* On Apple hardware the Cactus Ridge Thunderbolt controller needs to be
* shutdown before suspend . Otherwise the native host interface ( NHI ) will not
* be present after resume if a device was plugged in before suspend .
*
2018-05-02 16:53:19 +03:00
* The Thunderbolt controller consists of a PCIe switch with downstream
* bridges leading to the NHI and to the tunnel PCI bridges .
2014-06-04 00:04:10 +04:00
*
* This quirk cuts power to the whole chip . Therefore we have to apply it
* during suspend_noirq of the upstream bridge .
*
* Power is automagically restored before resume . No action is needed .
*/
static void quirk_apple_poweroff_thunderbolt ( struct pci_dev * dev )
{
acpi_handle bridge , SXIO , SXFP , SXLV ;
2017-08-01 15:10:41 +03:00
if ( ! x86_apple_machine )
2014-06-04 00:04:10 +04:00
return ;
if ( pci_pcie_type ( dev ) ! = PCI_EXP_TYPE_UPSTREAM )
return ;
bridge = ACPI_HANDLE ( & dev - > dev ) ;
if ( ! bridge )
return ;
2018-05-02 16:53:19 +03:00
2014-06-04 00:04:10 +04:00
/*
* SXIO and SXLV are present only on machines requiring this quirk .
2018-05-02 16:53:19 +03:00
* Thunderbolt bridges in external devices might have the same
* device ID as those on the host , but they will not have the
* associated ACPI methods . This implicitly checks that we are at
* the right bridge .
2014-06-04 00:04:10 +04:00
*/
if ( ACPI_FAILURE ( acpi_get_handle ( bridge , " DSB0.NHI0.SXIO " , & SXIO ) )
| | ACPI_FAILURE ( acpi_get_handle ( bridge , " DSB0.NHI0.SXFP " , & SXFP ) )
| | ACPI_FAILURE ( acpi_get_handle ( bridge , " DSB0.NHI0.SXLV " , & SXLV ) ) )
return ;
2018-05-02 16:53:19 +03:00
pci_info ( dev , " quirk: cutting power to Thunderbolt controller... \n " ) ;
2014-06-04 00:04:10 +04:00
/* magic sequence */
acpi_execute_simple_method ( SXIO , NULL , 1 ) ;
acpi_execute_simple_method ( SXFP , NULL , 0 ) ;
msleep ( 300 ) ;
acpi_execute_simple_method ( SXLV , NULL , 0 ) ;
acpi_execute_simple_method ( SXIO , NULL , 0 ) ;
acpi_execute_simple_method ( SXLV , NULL , 0 ) ;
}
2016-03-20 15:57:20 +03:00
DECLARE_PCI_FIXUP_SUSPEND_LATE ( PCI_VENDOR_ID_INTEL ,
PCI_DEVICE_ID_INTEL_CACTUS_RIDGE_4C ,
2014-06-04 00:04:10 +04:00
quirk_apple_poweroff_thunderbolt ) ;
# endif
2009-12-07 08:03:21 +03:00
/*
2017-02-28 01:29:56 +03:00
* Following are device - specific reset methods which can be used to
2009-12-07 08:03:21 +03:00
* reset a single function if other methods ( e . g . FLR , PM D0 - > D3 ) are
* not available .
*/
2009-12-07 08:03:23 +03:00
static int reset_intel_82599_sfp_virtfn ( struct pci_dev * dev , int probe )
{
2012-08-22 19:41:27 +04:00
/*
* http : //www.intel.com/content/dam/doc/datasheet/82599-10-gbe-controller-datasheet.pdf
*
* The 82599 supports FLR on VFs , but FLR support is reported only
* in the PF DEVCAP ( sec 9.3 .10 .4 ) , not in the VF DEVCAP ( sec 9.5 ) .
2017-04-14 22:11:26 +03:00
* Thus we must call pcie_flr ( ) directly without first checking if it is
* supported .
2012-08-22 19:41:27 +04:00
*/
2017-04-14 22:11:26 +03:00
if ( ! probe )
pcie_flr ( dev ) ;
2009-12-07 08:03:23 +03:00
return 0 ;
}
2015-11-05 00:19:49 +03:00
# define SOUTH_CHICKEN2 0xc2004
# define PCH_PP_STATUS 0xc7200
# define PCH_PP_CONTROL 0xc7204
2012-04-27 19:16:46 +04:00
# define MSG_CTL 0x45010
# define NSDE_PWR_STATE 0xd0100
# define IGD_OPERATION_TIMEOUT 10000 /* set timeout 10 seconds */
static int reset_ivb_igd ( struct pci_dev * dev , int probe )
{
void __iomem * mmio_base ;
unsigned long timeout ;
u32 val ;
if ( probe )
return 0 ;
mmio_base = pci_iomap ( dev , 0 , 0 ) ;
if ( ! mmio_base )
return - ENOMEM ;
iowrite32 ( 0x00000002 , mmio_base + MSG_CTL ) ;
/*
* Clobbering SOUTH_CHICKEN2 register is fine only if the next
* driver loaded sets the right bits . However , this ' s a reset and
* the bits have been set by i915 previously , so we clobber
* SOUTH_CHICKEN2 register directly here .
*/
iowrite32 ( 0x00000005 , mmio_base + SOUTH_CHICKEN2 ) ;
val = ioread32 ( mmio_base + PCH_PP_CONTROL ) & 0xfffffffe ;
iowrite32 ( val , mmio_base + PCH_PP_CONTROL ) ;
timeout = jiffies + msecs_to_jiffies ( IGD_OPERATION_TIMEOUT ) ;
do {
val = ioread32 ( mmio_base + PCH_PP_STATUS ) ;
if ( ( val & 0xb0000000 ) = = 0 )
goto reset_complete ;
msleep ( 10 ) ;
} while ( time_before ( jiffies , timeout ) ) ;
2018-01-18 21:55:24 +03:00
pci_warn ( dev , " timeout during reset \n " ) ;
2012-04-27 19:16:46 +04:00
reset_complete :
iowrite32 ( 0x00000002 , mmio_base + NSDE_PWR_STATE ) ;
pci_iounmap ( dev , mmio_base ) ;
return 0 ;
}
2018-05-02 16:53:19 +03:00
/* Device-specific reset method for Chelsio T4-based adapters */
2013-08-06 14:18:37 +04:00
static int reset_chelsio_generic_dev ( struct pci_dev * dev , int probe )
{
u16 old_command ;
u16 msix_flags ;
/*
* If this isn ' t a Chelsio T4 - based device , return - ENOTTY indicating
* that we have no device - specific reset method .
*/
if ( ( dev - > device & 0xf000 ) ! = 0x4000 )
return - ENOTTY ;
/*
* If this is the " probe " phase , return 0 indicating that we can
* reset this device .
*/
if ( probe )
return 0 ;
/*
* T4 can wedge if there are DMAs in flight within the chip and Bus
* Master has been disabled . We need to have it on till the Function
* Level Reset completes . ( BUS_MASTER is disabled in
* pci_reset_function ( ) ) .
*/
pci_read_config_word ( dev , PCI_COMMAND , & old_command ) ;
pci_write_config_word ( dev , PCI_COMMAND ,
old_command | PCI_COMMAND_MASTER ) ;
/*
* Perform the actual device function reset , saving and restoring
* configuration information around the reset .
*/
pci_save_state ( dev ) ;
/*
* T4 also suffers a Head - Of - Line blocking problem if MSI - X interrupts
* are disabled when an MSI - X interrupt message needs to be delivered .
* So we briefly re - enable MSI - X interrupts for the duration of the
* FLR . The pci_restore_state ( ) below will restore the original
* MSI - X state .
*/
pci_read_config_word ( dev , dev - > msix_cap + PCI_MSIX_FLAGS , & msix_flags ) ;
if ( ( msix_flags & PCI_MSIX_FLAGS_ENABLE ) = = 0 )
pci_write_config_word ( dev , dev - > msix_cap + PCI_MSIX_FLAGS ,
msix_flags |
PCI_MSIX_FLAGS_ENABLE |
PCI_MSIX_FLAGS_MASKALL ) ;
2017-04-14 22:11:27 +03:00
pcie_flr ( dev ) ;
2013-08-06 14:18:37 +04:00
/*
* Restore the configuration information ( BAR values , etc . ) including
* the original PCI Configuration Space Command word , and return
* success .
*/
pci_restore_state ( dev ) ;
pci_write_config_word ( dev , PCI_COMMAND , old_command ) ;
return 0 ;
}
2009-12-07 08:03:23 +03:00
# define PCI_DEVICE_ID_INTEL_82599_SFP_VF 0x10ed
2012-04-27 19:16:46 +04:00
# define PCI_DEVICE_ID_INTEL_IVB_M_VGA 0x0156
# define PCI_DEVICE_ID_INTEL_IVB_M2_VGA 0x0166
2009-12-07 08:03:23 +03:00
2018-08-09 23:18:33 +03:00
/*
* The Samsung SM961 / PM961 controller can sometimes enter a fatal state after
* FLR where config space reads from the device return - 1. We seem to be
* able to avoid this condition if we disable the NVMe controller prior to
* FLR . This quirk is generic for any NVMe class device requiring similar
* assistance to quiesce the device prior to FLR .
*
* NVMe specification : https : //nvmexpress.org/resources/specifications/
* Revision 1.0 e :
* Chapter 2 : Required and optional PCI config registers
* Chapter 3 : NVMe control registers
* Chapter 7.3 : Reset behavior
*/
static int nvme_disable_and_flr ( struct pci_dev * dev , int probe )
{
void __iomem * bar ;
u16 cmd ;
u32 cfg ;
if ( dev - > class ! = PCI_CLASS_STORAGE_EXPRESS | |
! pcie_has_flr ( dev ) | | ! pci_resource_start ( dev , 0 ) )
return - ENOTTY ;
if ( probe )
return 0 ;
bar = pci_iomap ( dev , 0 , NVME_REG_CC + sizeof ( cfg ) ) ;
if ( ! bar )
return - ENOTTY ;
pci_read_config_word ( dev , PCI_COMMAND , & cmd ) ;
pci_write_config_word ( dev , PCI_COMMAND , cmd | PCI_COMMAND_MEMORY ) ;
cfg = readl ( bar + NVME_REG_CC ) ;
/* Disable controller if enabled */
if ( cfg & NVME_CC_ENABLE ) {
u32 cap = readl ( bar + NVME_REG_CAP ) ;
unsigned long timeout ;
/*
* Per nvme_disable_ctrl ( ) skip shutdown notification as it
* could complete commands to the admin queue . We only intend
* to quiesce the device before reset .
*/
cfg & = ~ ( NVME_CC_SHN_MASK | NVME_CC_ENABLE ) ;
writel ( cfg , bar + NVME_REG_CC ) ;
/*
* Some controllers require an additional delay here , see
* NVME_QUIRK_DELAY_BEFORE_CHK_RDY . None of those are yet
* supported by this quirk .
*/
/* Cap register provides max timeout in 500ms increments */
timeout = ( ( NVME_CAP_TIMEOUT ( cap ) + 1 ) * HZ / 2 ) + jiffies ;
for ( ; ; ) {
u32 status = readl ( bar + NVME_REG_CSTS ) ;
/* Ready status becomes zero on disable complete */
if ( ! ( status & NVME_CSTS_RDY ) )
break ;
msleep ( 100 ) ;
if ( time_after ( jiffies , timeout ) ) {
pci_warn ( dev , " Timeout waiting for NVMe ready status to clear after disable \n " ) ;
break ;
}
}
}
pci_iounmap ( dev , bar ) ;
pcie_flr ( dev ) ;
return 0 ;
}
2018-08-09 23:04:31 +03:00
/*
* Intel DC P3700 NVMe controller will timeout waiting for ready status
* to change after NVMe enable if the driver starts interacting with the
* device too soon after FLR . A 250 ms delay after FLR has heuristically
* proven to produce reliably working results for device assignment cases .
*/
static int delay_250ms_after_flr ( struct pci_dev * dev , int probe )
{
if ( ! pcie_has_flr ( dev ) )
return - ENOTTY ;
if ( probe )
return 0 ;
pcie_flr ( dev ) ;
msleep ( 250 ) ;
return 0 ;
}
2009-12-31 21:06:35 +03:00
static const struct pci_dev_reset_methods pci_dev_reset_methods [ ] = {
2009-12-07 08:03:23 +03:00
{ PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_82599_SFP_VF ,
reset_intel_82599_sfp_virtfn } ,
2012-04-27 19:16:46 +04:00
{ PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_IVB_M_VGA ,
reset_ivb_igd } ,
{ PCI_VENDOR_ID_INTEL , PCI_DEVICE_ID_INTEL_IVB_M2_VGA ,
reset_ivb_igd } ,
2018-08-09 23:18:33 +03:00
{ PCI_VENDOR_ID_SAMSUNG , 0xa804 , nvme_disable_and_flr } ,
2018-08-09 23:04:31 +03:00
{ PCI_VENDOR_ID_INTEL , 0x0953 , delay_250ms_after_flr } ,
2013-08-06 14:18:37 +04:00
{ PCI_VENDOR_ID_CHELSIO , PCI_ANY_ID ,
reset_chelsio_generic_dev } ,
2009-12-07 08:03:21 +03:00
{ 0 }
} ;
2009-12-31 21:06:35 +03:00
2012-04-27 19:16:46 +04:00
/*
* These device - specific reset methods are here rather than in a driver
* because when a host assigns a device to a guest VM , the host may need
* to reset the device but probably doesn ' t have a driver for it .
*/
2009-12-31 21:06:35 +03:00
int pci_dev_specific_reset ( struct pci_dev * dev , int probe )
{
2010-01-01 03:44:43 +03:00
const struct pci_dev_reset_methods * i ;
2009-12-31 21:06:35 +03:00
for ( i = pci_dev_reset_methods ; i - > reset ; i + + ) {
if ( ( i - > vendor = = dev - > vendor | |
i - > vendor = = ( u16 ) PCI_ANY_ID ) & &
( i - > device = = dev - > device | |
i - > device = = ( u16 ) PCI_ANY_ID ) )
return i - > reset ( dev , probe ) ;
}
return - ENOTTY ;
}
2012-06-11 09:26:55 +04:00
2014-05-23 03:07:49 +04:00
static void quirk_dma_func0_alias ( struct pci_dev * dev )
{
2016-02-24 22:43:45 +03:00
if ( PCI_FUNC ( dev - > devfn ) ! = 0 )
2019-12-11 01:07:30 +03:00
pci_add_dma_alias ( dev , PCI_DEVFN ( PCI_SLOT ( dev - > devfn ) , 0 ) , 1 ) ;
2014-05-23 03:07:49 +04:00
}
/*
* https : //bugzilla.redhat.com/show_bug.cgi?id=605888
*
* Some Ricoh devices use function 0 as the PCIe requester ID for DMA .
*/
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_RICOH , 0xe832 , quirk_dma_func0_alias ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_RICOH , 0xe476 , quirk_dma_func0_alias ) ;
2014-05-29 00:54:00 +04:00
static void quirk_dma_func1_alias ( struct pci_dev * dev )
{
2016-02-24 22:43:45 +03:00
if ( PCI_FUNC ( dev - > devfn ) ! = 1 )
2019-12-11 01:07:30 +03:00
pci_add_dma_alias ( dev , PCI_DEVFN ( PCI_SLOT ( dev - > devfn ) , 1 ) , 1 ) ;
2014-05-29 00:54:00 +04:00
}
/*
* Marvell 88 SE9123 uses function 1 as the requester ID for DMA . In some
* SKUs function 1 is present and is a legacy IDE controller , in other
* SKUs this function is not present , making this a ghost requester .
* https : //bugzilla.kernel.org/show_bug.cgi?id=42679
*/
2015-05-22 00:03:38 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_MARVELL_EXT , 0x9120 ,
quirk_dma_func1_alias ) ;
2014-05-29 00:54:00 +04:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_MARVELL_EXT , 0x9123 ,
quirk_dma_func1_alias ) ;
2018-01-16 20:05:26 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_MARVELL_EXT , 0x9128 ,
quirk_dma_func1_alias ) ;
2014-05-29 00:54:00 +04:00
/* https://bugzilla.kernel.org/show_bug.cgi?id=42679#c14 */
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_MARVELL_EXT , 0x9130 ,
quirk_dma_func1_alias ) ;
2019-04-05 18:20:47 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_MARVELL_EXT , 0x9170 ,
quirk_dma_func1_alias ) ;
2014-05-29 00:54:00 +04:00
/* https://bugzilla.kernel.org/show_bug.cgi?id=42679#c47 + c57 */
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_MARVELL_EXT , 0x9172 ,
quirk_dma_func1_alias ) ;
/* https://bugzilla.kernel.org/show_bug.cgi?id=42679#c59 */
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_MARVELL_EXT , 0x917a ,
quirk_dma_func1_alias ) ;
2016-05-18 17:04:19 +03:00
/* https://bugzilla.kernel.org/show_bug.cgi?id=42679#c78 */
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_MARVELL_EXT , 0x9182 ,
quirk_dma_func1_alias ) ;
2018-08-13 22:30:41 +03:00
/* https://bugzilla.kernel.org/show_bug.cgi?id=42679#c134 */
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_MARVELL_EXT , 0x9183 ,
quirk_dma_func1_alias ) ;
2014-05-29 00:54:00 +04:00
/* https://bugzilla.kernel.org/show_bug.cgi?id=42679#c46 */
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_MARVELL_EXT , 0x91a0 ,
quirk_dma_func1_alias ) ;
2018-02-27 12:20:36 +03:00
/* https://bugzilla.kernel.org/show_bug.cgi?id=42679#c127 */
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_MARVELL_EXT , 0x9220 ,
quirk_dma_func1_alias ) ;
2014-05-29 00:54:00 +04:00
/* https://bugzilla.kernel.org/show_bug.cgi?id=42679#c49 */
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_MARVELL_EXT , 0x9230 ,
quirk_dma_func1_alias ) ;
2014-06-03 23:41:56 +04:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_TTI , 0x0642 ,
quirk_dma_func1_alias ) ;
2018-03-02 13:36:33 +03:00
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_TTI , 0x0645 ,
quirk_dma_func1_alias ) ;
2014-05-29 00:54:00 +04:00
/* https://bugs.gentoo.org/show_bug.cgi?id=497630 */
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_JMICRON ,
PCI_DEVICE_ID_JMICRON_JMB388_ESD ,
quirk_dma_func1_alias ) ;
2016-01-19 23:32:29 +03:00
/* https://bugzilla.kernel.org/show_bug.cgi?id=42679#c117 */
DECLARE_PCI_FIXUP_HEADER ( 0x1c28 , /* Lite-On */
0x0122 , /* Plextor M6E (Marvell 88SS9183)*/
quirk_dma_func1_alias ) ;
2014-05-29 00:54:00 +04:00
2015-01-13 21:26:50 +03:00
/*
* Some devices DMA with the wrong devfn , not just the wrong function .
* quirk_fixed_dma_alias ( ) uses this table to create fixed aliases , where
* the alias is " fixed " and independent of the device devfn .
*
* For example , the Adaptec 3405 is a PCIe card with an Intel 80333 I / O
* processor . To software , this appears as a PCIe - to - PCI / X bridge with a
* single device on the secondary bus . In reality , the single exposed
* device at 0 e .0 is the Address Translation Unit ( ATU ) of the controller
* that provides a bridge to the internal bus of the I / O processor . The
* controller supports private devices , which can be hidden from PCI config
* space . In the case of the Adaptec 3405 , a private device at 01.0
* appears to be the DMA engine , which therefore needs to become a DMA
* alias for the device .
*/
static const struct pci_device_id fixed_dma_alias_tbl [ ] = {
{ PCI_DEVICE_SUB ( PCI_VENDOR_ID_ADAPTEC2 , 0x0285 ,
PCI_VENDOR_ID_ADAPTEC2 , 0x02bb ) , /* Adaptec 3405 */
. driver_data = PCI_DEVFN ( 1 , 0 ) } ,
2016-07-18 17:32:45 +03:00
{ PCI_DEVICE_SUB ( PCI_VENDOR_ID_ADAPTEC2 , 0x0285 ,
PCI_VENDOR_ID_ADAPTEC2 , 0x02bc ) , /* Adaptec 3805 */
. driver_data = PCI_DEVFN ( 1 , 0 ) } ,
2015-01-13 21:26:50 +03:00
{ 0 }
} ;
static void quirk_fixed_dma_alias ( struct pci_dev * dev )
{
const struct pci_device_id * id ;
id = pci_match_id ( fixed_dma_alias_tbl , dev ) ;
2016-02-24 22:43:54 +03:00
if ( id )
2019-12-11 01:07:30 +03:00
pci_add_dma_alias ( dev , id - > driver_data , 1 ) ;
2015-01-13 21:26:50 +03:00
}
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_ADAPTEC2 , 0x0285 , quirk_fixed_dma_alias ) ;
2014-05-23 03:08:07 +04:00
/*
* A few PCIe - to - PCI bridges fail to expose a PCIe capability , resulting in
* using the wrong DMA alias for the device . Some of these devices can be
* used as either forward or reverse bridges , so we need to test whether the
* device is operating in the correct mode . We could probably apply this
* quirk to PCI_ANY_ID , but for now we ' ll just use known offenders . The test
* is for a non - root , non - PCIe bridge where the upstream device is PCIe and
* is not a PCIe - to - PCI bridge , then @ pdev is actually a PCIe - to - PCI bridge .
*/
static void quirk_use_pcie_bridge_dma_alias ( struct pci_dev * pdev )
{
if ( ! pci_is_root_bus ( pdev - > bus ) & &
pdev - > hdr_type = = PCI_HEADER_TYPE_BRIDGE & &
! pci_is_pcie ( pdev ) & & pci_is_pcie ( pdev - > bus - > self ) & &
pci_pcie_type ( pdev - > bus - > self ) ! = PCI_EXP_TYPE_PCI_BRIDGE )
pdev - > dev_flags | = PCI_DEV_FLAG_PCIE_BRIDGE_ALIAS ;
}
/* ASM1083/1085, https://bugzilla.kernel.org/show_bug.cgi?id=44881#c46 */
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_ASMEDIA , 0x1080 ,
quirk_use_pcie_bridge_dma_alias ) ;
/* Tundra 8113, https://bugzilla.kernel.org/show_bug.cgi?id=44881#c43 */
DECLARE_PCI_FIXUP_HEADER ( 0x10e3 , 0x8113 , quirk_use_pcie_bridge_dma_alias ) ;
2014-06-09 22:43:25 +04:00
/* ITE 8892, https://bugzilla.kernel.org/show_bug.cgi?id=73551 */
DECLARE_PCI_FIXUP_HEADER ( 0x1283 , 0x8892 , quirk_use_pcie_bridge_dma_alias ) ;
2017-04-12 20:33:04 +03:00
/* ITE 8893 has the same problem as the 8892 */
DECLARE_PCI_FIXUP_HEADER ( 0x1283 , 0x8893 , quirk_use_pcie_bridge_dma_alias ) ;
2014-07-06 01:26:52 +04:00
/* Intel 82801, https://bugzilla.kernel.org/show_bug.cgi?id=44881#c49 */
DECLARE_PCI_FIXUP_HEADER ( 0x8086 , 0x244e , quirk_use_pcie_bridge_dma_alias ) ;
2014-05-23 03:08:07 +04:00
2016-03-03 17:53:20 +03:00
/*
* MIC x200 NTB forwards PCIe traffic using multiple alien RIDs . They have to
* be added as aliases to the DMA device in order to allow buffer access
* when IOMMU is enabled . Following devfns have to match RIT - LUT table
* programmed in the EEPROM .
*/
static void quirk_mic_x200_dma_alias ( struct pci_dev * pdev )
{
2019-12-11 01:07:30 +03:00
pci_add_dma_alias ( pdev , PCI_DEVFN ( 0x10 , 0x0 ) , 1 ) ;
pci_add_dma_alias ( pdev , PCI_DEVFN ( 0x11 , 0x0 ) , 1 ) ;
pci_add_dma_alias ( pdev , PCI_DEVFN ( 0x12 , 0x3 ) , 1 ) ;
2016-03-03 17:53:20 +03:00
}
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x2260 , quirk_mic_x200_dma_alias ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x2264 , quirk_mic_x200_dma_alias ) ;
2019-09-17 12:20:48 +03:00
/*
* Intel Visual Compute Accelerator ( VCA ) is a family of PCIe add - in devices
* exposing computational units via Non Transparent Bridges ( NTB , PEX 87 xx ) .
*
* Similarly to MIC x200 , we need to add DMA aliases to allow buffer access
* when IOMMU is enabled . These aliases allow computational unit access to
* host memory . These aliases mark the whole VCA device as one IOMMU
* group .
*
* All possible slot numbers ( 0x20 ) are used , since we are unable to tell
* what slot is used on other side . This quirk is intended for both host
* and computational unit sides . The VCA devices have up to five functions
* ( four for DMA channels and one additional ) .
*/
static void quirk_pex_vca_alias ( struct pci_dev * pdev )
{
const unsigned int num_pci_slots = 0x20 ;
unsigned int slot ;
2019-12-11 01:07:30 +03:00
for ( slot = 0 ; slot < num_pci_slots ; slot + + )
pci_add_dma_alias ( pdev , PCI_DEVFN ( slot , 0x0 ) , 5 ) ;
2019-09-17 12:20:48 +03:00
}
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x2954 , quirk_pex_vca_alias ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x2955 , quirk_pex_vca_alias ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x2956 , quirk_pex_vca_alias ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x2958 , quirk_pex_vca_alias ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x2959 , quirk_pex_vca_alias ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_INTEL , 0x295A , quirk_pex_vca_alias ) ;
PCI: Avoid generating invalid ThunderX2 DMA aliases
On Cavium ThunderX2 arm64 SoCs (formerly known as Broadcom Vulcan), the PCI
topology is slightly unusual. For a multi-node system, it looks like:
00:00.0 PCI bridge to [bus 01-1e]
01:0a.0 PCI-to-PCIe bridge to [bus 02-04]
02:00.0 PCIe Root Port bridge to [bus 03-04] (XLATE_ROOT)
03:00.0 PCIe Endpoint
pci_for_each_dma_alias() assumes IOMMU translation is done at the root of
the PCI hierarchy. It generates 03:00.0, 01:0a.0, and 00:00.0 as DMA
aliases for 03:00.0 because buses 01 and 00 are non-PCIe buses that don't
carry the Requester ID.
Because the ThunderX2 IOMMU is at 02:00.0, the Requester IDs 01:0a.0 and
00:00.0 are never valid for the endpoint. This quirk stops alias
generation at the XLATE_ROOT bridge so we won't generate 01:0a.0 or
00:00.0.
The current IOMMU code only maps the last alias (this is a separate bug in
itself). Prior to this quirk, we only created IOMMU mappings for the
invalid Requester ID 00:00:0, which never matched any DMA transactions.
With this quirk, we create IOMMU mappings for a valid Requester ID, which
fixes devices with no aliases but leaves devices with aliases still broken.
The last alias for the endpoint is also used by the ARM GICv3 MSI-X code.
Without this quirk, the GIC Interrupt Translation Tables are setup with the
invalid Requester ID, and the MSI-X generated by the device fails to be
translated and routed.
Link: https://bugzilla.kernel.org/show_bug.cgi?id=195447
Signed-off-by: Jayachandran C <jnair@caviumnetworks.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Robin Murphy <robin.murphy@arm.com>
Acked-by: David Daney <david.daney@cavium.com>
2017-04-13 23:30:45 +03:00
/*
* The IOMMU and interrupt controller on Broadcom Vulcan / Cavium ThunderX2 are
* associated not at the root bus , but at a bridge below . This quirk avoids
* generating invalid DMA aliases .
*/
static void quirk_bridge_cavm_thrx2_pcie_root ( struct pci_dev * pdev )
{
pdev - > dev_flags | = PCI_DEV_FLAGS_BRIDGE_XLATE_ROOT ;
}
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_BROADCOM , 0x9000 ,
quirk_bridge_cavm_thrx2_pcie_root ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_BROADCOM , 0x9084 ,
quirk_bridge_cavm_thrx2_pcie_root ) ;
2015-06-19 11:00:15 +03:00
/*
* Intersil / Techwell TW686 [ 4589 ] - based video capture cards have an empty ( zero )
* class code . Fix it .
*/
static void quirk_tw686x_class ( struct pci_dev * pdev )
{
u32 class = pdev - > class ;
/* Use "Multimedia controller" class */
pdev - > class = ( PCI_CLASS_MULTIMEDIA_OTHER < < 8 ) | 0x01 ;
2018-01-18 21:55:24 +03:00
pci_info ( pdev , " TW686x PCI class overridden (%#08x -> %#08x) \n " ,
2015-06-19 11:00:15 +03:00
class , pdev - > class ) ;
}
2015-06-20 00:20:58 +03:00
DECLARE_PCI_FIXUP_CLASS_EARLY ( 0x1797 , 0x6864 , PCI_CLASS_NOT_DEFINED , 8 ,
2015-06-19 11:00:15 +03:00
quirk_tw686x_class ) ;
2015-06-20 00:20:58 +03:00
DECLARE_PCI_FIXUP_CLASS_EARLY ( 0x1797 , 0x6865 , PCI_CLASS_NOT_DEFINED , 8 ,
2015-06-19 11:00:15 +03:00
quirk_tw686x_class ) ;
2015-06-20 00:20:58 +03:00
DECLARE_PCI_FIXUP_CLASS_EARLY ( 0x1797 , 0x6868 , PCI_CLASS_NOT_DEFINED , 8 ,
2015-06-19 11:00:15 +03:00
quirk_tw686x_class ) ;
2015-06-20 00:20:58 +03:00
DECLARE_PCI_FIXUP_CLASS_EARLY ( 0x1797 , 0x6869 , PCI_CLASS_NOT_DEFINED , 8 ,
2015-06-19 11:00:15 +03:00
quirk_tw686x_class ) ;
2017-08-15 06:23:23 +03:00
/*
* Some devices have problems with Transaction Layer Packets with the Relaxed
* Ordering Attribute set . Such devices should mark themselves and other
2018-05-02 16:53:19 +03:00
* device drivers should check before sending TLPs with RO set .
2017-08-15 06:23:23 +03:00
*/
static void quirk_relaxedordering_disable ( struct pci_dev * dev )
{
dev - > dev_flags | = PCI_DEV_FLAGS_NO_RELAXED_ORDERING ;
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Disable Relaxed Ordering Attributes to avoid PCIe Completion erratum \n " ) ;
2017-08-15 06:23:23 +03:00
}
2017-08-15 06:23:24 +03:00
/*
* Intel Xeon processors based on Broadwell / Haswell microarchitecture Root
2018-05-02 16:53:19 +03:00
* Complex have a Flow Control Credit issue which can cause performance
2017-08-15 06:23:24 +03:00
* problems with Upstream Transaction Layer Packets with Relaxed Ordering set .
*/
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_INTEL , 0x6f01 , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_INTEL , 0x6f02 , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_INTEL , 0x6f03 , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_INTEL , 0x6f04 , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_INTEL , 0x6f05 , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_INTEL , 0x6f06 , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_INTEL , 0x6f07 , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_INTEL , 0x6f08 , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_INTEL , 0x6f09 , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_INTEL , 0x6f0a , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_INTEL , 0x6f0b , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_INTEL , 0x6f0c , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_INTEL , 0x6f0d , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_INTEL , 0x6f0e , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_INTEL , 0x2f01 , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_INTEL , 0x2f02 , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_INTEL , 0x2f03 , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_INTEL , 0x2f04 , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_INTEL , 0x2f05 , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_INTEL , 0x2f06 , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_INTEL , 0x2f07 , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_INTEL , 0x2f08 , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_INTEL , 0x2f09 , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_INTEL , 0x2f0a , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_INTEL , 0x2f0b , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_INTEL , 0x2f0c , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_INTEL , 0x2f0d , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_INTEL , 0x2f0e , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
2017-08-15 06:23:25 +03:00
/*
2018-05-02 16:53:19 +03:00
* The AMD ARM A1100 ( aka " SEATTLE " ) SoC has a bug in its PCIe Root Complex
2017-08-15 06:23:25 +03:00
* where Upstream Transaction Layer Packets with the Relaxed Ordering
* Attribute clear are allowed to bypass earlier TLPs with Relaxed Ordering
* set . This is a violation of the PCIe 3.0 Transaction Ordering Rules
* outlined in Section 2.4 .1 ( PCI Express ( r ) Base Specification Revision 3.0
* November 10 , 2010 ) . As a result , on this platform we can ' t use Relaxed
* Ordering for Upstream TLPs .
*/
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_AMD , 0x1a00 , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_AMD , 0x1a01 , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
DECLARE_PCI_FIXUP_CLASS_EARLY ( PCI_VENDOR_ID_AMD , 0x1a02 , PCI_CLASS_NOT_DEFINED , 8 ,
quirk_relaxedordering_disable ) ;
PCI: Turn off Request Attributes to avoid Chelsio T5 Completion erratum
The Chelsio T5 has a PCIe compliance erratum that causes Malformed TLP or
Unexpected Completion errors in some systems, which may cause device access
timeouts.
Per PCIe r3.0, sec 2.2.9, "Completion headers must supply the same values
for the Attribute as were supplied in the header of the corresponding
Request, except as explicitly allowed when IDO is used."
Instead of copying the Attributes from the Request to the Completion, the
T5 always generates Completions with zero Attributes. The receiver of a
Completion whose Attributes don't match the Request may accept it (which
itself seems non-compliant based on sec 2.3.2), or it may handle it as a
Malformed TLP or an Unexpected Completion, which will probably lead to a
device access timeout.
Work around this by disabling "Relaxed Ordering" and "No Snoop" in the Root
Port so it always generate Requests with zero Attributes.
This does affect all other devices which are downstream of that Root Port,
but these are performance optimizations that should not make a functional
difference.
Note that Configuration Space accesses are never supposed to have TLP
Attributes, so we're safe waiting till after any Configuration Space
accesses to do the Root Port "fixup".
Based on original work by Casey Leedom <leedom@chelsio.com>
[bhelgaas: changelog, comments, rename to pci_find_pcie_root_port(), rework
to use pci_upstream_bridge() and check for Root Port device type, edit
diagnostics to clarify intent and devices affected]
Signed-off-by: Hariprasad Shenai <hariprasad@chelsio.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2015-10-18 17:25:04 +03:00
/*
* Per PCIe r3 .0 , sec 2.2 .9 , " Completion headers must supply the same
* values for the Attribute as were supplied in the header of the
* corresponding Request , except as explicitly allowed when IDO is used . "
*
* If a non - compliant device generates a completion with a different
* attribute than the request , the receiver may accept it ( which itself
* seems non - compliant based on sec 2.3 .2 ) , or it may handle it as a
* Malformed TLP or an Unexpected Completion , which will probably lead to a
* device access timeout .
*
* If the non - compliant device generates completions with zero attributes
* ( instead of copying the attributes from the request ) , we can work around
* this by disabling the " Relaxed Ordering " and " No Snoop " attributes in
* upstream devices so they always generate requests with zero attributes .
*
* This affects other devices under the same Root Port , but since these
* attributes are performance hints , there should be no functional problem .
*
* Note that Configuration Space accesses are never supposed to have TLP
* Attributes , so we ' re safe waiting till after any Configuration Space
* accesses to do the Root Port fixup .
*/
static void quirk_disable_root_port_attributes ( struct pci_dev * pdev )
{
2020-05-09 13:19:28 +03:00
struct pci_dev * root_port = pcie_find_root_port ( pdev ) ;
PCI: Turn off Request Attributes to avoid Chelsio T5 Completion erratum
The Chelsio T5 has a PCIe compliance erratum that causes Malformed TLP or
Unexpected Completion errors in some systems, which may cause device access
timeouts.
Per PCIe r3.0, sec 2.2.9, "Completion headers must supply the same values
for the Attribute as were supplied in the header of the corresponding
Request, except as explicitly allowed when IDO is used."
Instead of copying the Attributes from the Request to the Completion, the
T5 always generates Completions with zero Attributes. The receiver of a
Completion whose Attributes don't match the Request may accept it (which
itself seems non-compliant based on sec 2.3.2), or it may handle it as a
Malformed TLP or an Unexpected Completion, which will probably lead to a
device access timeout.
Work around this by disabling "Relaxed Ordering" and "No Snoop" in the Root
Port so it always generate Requests with zero Attributes.
This does affect all other devices which are downstream of that Root Port,
but these are performance optimizations that should not make a functional
difference.
Note that Configuration Space accesses are never supposed to have TLP
Attributes, so we're safe waiting till after any Configuration Space
accesses to do the Root Port "fixup".
Based on original work by Casey Leedom <leedom@chelsio.com>
[bhelgaas: changelog, comments, rename to pci_find_pcie_root_port(), rework
to use pci_upstream_bridge() and check for Root Port device type, edit
diagnostics to clarify intent and devices affected]
Signed-off-by: Hariprasad Shenai <hariprasad@chelsio.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2015-10-18 17:25:04 +03:00
if ( ! root_port ) {
2018-01-18 21:55:24 +03:00
pci_warn ( pdev , " PCIe Completion erratum may cause device errors \n " ) ;
PCI: Turn off Request Attributes to avoid Chelsio T5 Completion erratum
The Chelsio T5 has a PCIe compliance erratum that causes Malformed TLP or
Unexpected Completion errors in some systems, which may cause device access
timeouts.
Per PCIe r3.0, sec 2.2.9, "Completion headers must supply the same values
for the Attribute as were supplied in the header of the corresponding
Request, except as explicitly allowed when IDO is used."
Instead of copying the Attributes from the Request to the Completion, the
T5 always generates Completions with zero Attributes. The receiver of a
Completion whose Attributes don't match the Request may accept it (which
itself seems non-compliant based on sec 2.3.2), or it may handle it as a
Malformed TLP or an Unexpected Completion, which will probably lead to a
device access timeout.
Work around this by disabling "Relaxed Ordering" and "No Snoop" in the Root
Port so it always generate Requests with zero Attributes.
This does affect all other devices which are downstream of that Root Port,
but these are performance optimizations that should not make a functional
difference.
Note that Configuration Space accesses are never supposed to have TLP
Attributes, so we're safe waiting till after any Configuration Space
accesses to do the Root Port "fixup".
Based on original work by Casey Leedom <leedom@chelsio.com>
[bhelgaas: changelog, comments, rename to pci_find_pcie_root_port(), rework
to use pci_upstream_bridge() and check for Root Port device type, edit
diagnostics to clarify intent and devices affected]
Signed-off-by: Hariprasad Shenai <hariprasad@chelsio.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2015-10-18 17:25:04 +03:00
return ;
}
2018-01-18 21:55:24 +03:00
pci_info ( root_port , " Disabling No Snoop/Relaxed Ordering Attributes to avoid PCIe Completion erratum in %s \n " ,
PCI: Turn off Request Attributes to avoid Chelsio T5 Completion erratum
The Chelsio T5 has a PCIe compliance erratum that causes Malformed TLP or
Unexpected Completion errors in some systems, which may cause device access
timeouts.
Per PCIe r3.0, sec 2.2.9, "Completion headers must supply the same values
for the Attribute as were supplied in the header of the corresponding
Request, except as explicitly allowed when IDO is used."
Instead of copying the Attributes from the Request to the Completion, the
T5 always generates Completions with zero Attributes. The receiver of a
Completion whose Attributes don't match the Request may accept it (which
itself seems non-compliant based on sec 2.3.2), or it may handle it as a
Malformed TLP or an Unexpected Completion, which will probably lead to a
device access timeout.
Work around this by disabling "Relaxed Ordering" and "No Snoop" in the Root
Port so it always generate Requests with zero Attributes.
This does affect all other devices which are downstream of that Root Port,
but these are performance optimizations that should not make a functional
difference.
Note that Configuration Space accesses are never supposed to have TLP
Attributes, so we're safe waiting till after any Configuration Space
accesses to do the Root Port "fixup".
Based on original work by Casey Leedom <leedom@chelsio.com>
[bhelgaas: changelog, comments, rename to pci_find_pcie_root_port(), rework
to use pci_upstream_bridge() and check for Root Port device type, edit
diagnostics to clarify intent and devices affected]
Signed-off-by: Hariprasad Shenai <hariprasad@chelsio.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2015-10-18 17:25:04 +03:00
dev_name ( & pdev - > dev ) ) ;
pcie_capability_clear_and_set_word ( root_port , PCI_EXP_DEVCTL ,
PCI_EXP_DEVCTL_RELAX_EN |
PCI_EXP_DEVCTL_NOSNOOP_EN , 0 ) ;
}
/*
* The Chelsio T5 chip fails to copy TLP Attributes from a Request to the
* Completion it generates .
*/
static void quirk_chelsio_T5_disable_root_port_attributes ( struct pci_dev * pdev )
{
/*
* This mask / compare operation selects for Physical Function 4 on a
* T5 . We only need to fix up the Root Port once for any of the
* PFs . PF [ 0. .3 ] have PCI Device IDs of 0x50 xx , but PF4 is uniquely
2018-05-02 16:53:19 +03:00
* 0x54 xx so we use that one .
PCI: Turn off Request Attributes to avoid Chelsio T5 Completion erratum
The Chelsio T5 has a PCIe compliance erratum that causes Malformed TLP or
Unexpected Completion errors in some systems, which may cause device access
timeouts.
Per PCIe r3.0, sec 2.2.9, "Completion headers must supply the same values
for the Attribute as were supplied in the header of the corresponding
Request, except as explicitly allowed when IDO is used."
Instead of copying the Attributes from the Request to the Completion, the
T5 always generates Completions with zero Attributes. The receiver of a
Completion whose Attributes don't match the Request may accept it (which
itself seems non-compliant based on sec 2.3.2), or it may handle it as a
Malformed TLP or an Unexpected Completion, which will probably lead to a
device access timeout.
Work around this by disabling "Relaxed Ordering" and "No Snoop" in the Root
Port so it always generate Requests with zero Attributes.
This does affect all other devices which are downstream of that Root Port,
but these are performance optimizations that should not make a functional
difference.
Note that Configuration Space accesses are never supposed to have TLP
Attributes, so we're safe waiting till after any Configuration Space
accesses to do the Root Port "fixup".
Based on original work by Casey Leedom <leedom@chelsio.com>
[bhelgaas: changelog, comments, rename to pci_find_pcie_root_port(), rework
to use pci_upstream_bridge() and check for Root Port device type, edit
diagnostics to clarify intent and devices affected]
Signed-off-by: Hariprasad Shenai <hariprasad@chelsio.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2015-10-18 17:25:04 +03:00
*/
if ( ( pdev - > device & 0xff00 ) = = 0x5400 )
quirk_disable_root_port_attributes ( pdev ) ;
}
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_CHELSIO , PCI_ANY_ID ,
quirk_chelsio_T5_disable_root_port_attributes ) ;
2019-09-07 02:36:06 +03:00
/*
* pci_acs_ctrl_enabled - compare desired ACS controls with those provided
* by a device
* @ acs_ctrl_req : Bitmask of desired ACS controls
* @ acs_ctrl_ena : Bitmask of ACS controls enabled or provided implicitly by
* the hardware design
*
* Return 1 if all ACS controls in the @ acs_ctrl_req bitmask are included
* in @ acs_ctrl_ena , i . e . , the device provides all the access controls the
* caller desires . Return 0 otherwise .
*/
static int pci_acs_ctrl_enabled ( u16 acs_ctrl_req , u16 acs_ctrl_ena )
{
if ( ( acs_ctrl_req & acs_ctrl_ena ) = = acs_ctrl_req )
return 1 ;
return 0 ;
}
2013-06-28 02:40:00 +04:00
/*
* AMD has indicated that the devices below do not support peer - to - peer
* in any system where they are found in the southbridge with an AMD
* IOMMU in the system . Multifunction devices that do not support
* peer - to - peer between functions can claim to support a subset of ACS .
* Such devices effectively enable request redirect ( RR ) and completion
* redirect ( CR ) since all transactions are redirected to the upstream
* root complex .
*
2020-06-30 20:41:39 +03:00
* https : //lore.kernel.org/r/201207111426.q6BEQTbh002928@mail.maya.org/
* https : //lore.kernel.org/r/20120711165854.GM25282@amd.com/
* https : //lore.kernel.org/r/20121005130857.GX4009@amd.com/
2013-06-28 02:40:00 +04:00
*
* 1002 : 4385 SBx00 SMBus Controller
* 1002 : 439 c SB7x0 / SB8x0 / SB9x0 IDE Controller
* 1002 : 4383 SBx00 Azalia ( Intel HDA )
* 1002 : 439 d SB7x0 / SB8x0 / SB9x0 LPC host controller
* 1002 : 4384 SBx00 PCI to PCI Bridge
* 1002 : 4399 SB7x0 / SB8x0 / SB9x0 USB OHCI2 Controller
2014-10-02 18:50:31 +04:00
*
* https : //bugzilla.kernel.org/show_bug.cgi?id=81841#c15
*
* 1022 : 780f [ AMD ] FCH PCI Bridge
* 1022 : 7809 [ AMD ] FCH USB OHCI Controller
2013-06-28 02:40:00 +04:00
*/
static int pci_quirk_amd_sb_acs ( struct pci_dev * dev , u16 acs_flags )
{
# ifdef CONFIG_ACPI
struct acpi_table_header * header = NULL ;
acpi_status status ;
/* Targeting multifunction devices on the SB (appears on root bus) */
if ( ! dev - > multifunction | | ! pci_is_root_bus ( dev - > bus ) )
return - ENODEV ;
/* The IVRS table describes the AMD IOMMU */
status = acpi_get_table ( " IVRS " , 0 , & header ) ;
if ( ACPI_FAILURE ( status ) )
return - ENODEV ;
2020-07-22 12:44:28 +03:00
acpi_put_table ( header ) ;
2013-06-28 02:40:00 +04:00
/* Filter out flags not applicable to multifunction */
acs_flags & = ( PCI_ACS_RR | PCI_ACS_CR | PCI_ACS_EC | PCI_ACS_DT ) ;
2019-09-07 02:36:06 +03:00
return pci_acs_ctrl_enabled ( acs_flags , PCI_ACS_RR | PCI_ACS_CR ) ;
2013-06-28 02:40:00 +04:00
# else
return - ENODEV ;
# endif
}
2017-10-17 15:47:39 +03:00
static bool pci_quirk_cavium_acs_match ( struct pci_dev * dev )
{
2019-11-11 05:43:03 +03:00
if ( ! pci_is_pcie ( dev ) | | pci_pcie_type ( dev ) ! = PCI_EXP_TYPE_ROOT_PORT )
return false ;
switch ( dev - > device ) {
2017-10-17 15:47:39 +03:00
/*
2019-11-11 05:43:03 +03:00
* Effectively selects all downstream ports for whole ThunderX1
* ( which represents 8 SoCs ) .
2017-10-17 15:47:39 +03:00
*/
2019-11-11 05:43:03 +03:00
case 0xa000 . . . 0xa7ff : /* ThunderX1 */
case 0xaf84 : /* ThunderX2 */
case 0xb884 : /* ThunderX3 */
return true ;
default :
return false ;
}
2017-10-17 15:47:39 +03:00
}
2016-01-29 23:03:58 +03:00
static int pci_quirk_cavium_acs ( struct pci_dev * dev , u16 acs_flags )
{
2019-09-06 01:54:42 +03:00
if ( ! pci_quirk_cavium_acs_match ( dev ) )
return - ENOTTY ;
2016-01-29 23:03:58 +03:00
/*
2019-09-06 01:54:42 +03:00
* Cavium Root Ports don ' t advertise an ACS capability . However ,
2017-10-17 15:47:38 +03:00
* the RTL internally implements similar protection as if ACS had
2019-09-06 01:54:42 +03:00
* Source Validation , Request Redirection , Completion Redirection ,
2017-10-17 15:47:38 +03:00
* and Upstream Forwarding features enabled . Assert that the
* hardware implements and enables equivalent ACS functionality for
* these flags .
2016-01-29 23:03:58 +03:00
*/
2019-09-07 02:36:06 +03:00
return pci_acs_ctrl_enabled ( acs_flags ,
PCI_ACS_SV | PCI_ACS_RR | PCI_ACS_CR | PCI_ACS_UF ) ;
2016-01-29 23:03:58 +03:00
}
2017-08-11 00:06:33 +03:00
static int pci_quirk_xgene_acs ( struct pci_dev * dev , u16 acs_flags )
{
/*
2018-05-02 16:53:19 +03:00
* X - Gene Root Ports matching this quirk do not allow peer - to - peer
2017-08-11 00:06:33 +03:00
* transactions with others , allowing masking out these bits as if they
* were unimplemented in the ACS capability .
*/
2019-09-07 02:36:06 +03:00
return pci_acs_ctrl_enabled ( acs_flags ,
PCI_ACS_SV | PCI_ACS_RR | PCI_ACS_CR | PCI_ACS_UF ) ;
2017-08-11 00:06:33 +03:00
}
2020-03-27 12:11:48 +03:00
/*
* Many Zhaoxin Root Ports and Switch Downstream Ports have no ACS capability .
* But the implementation could block peer - to - peer transactions between them
* and provide ACS - like functionality .
*/
static int pci_quirk_zhaoxin_pcie_ports_acs ( struct pci_dev * dev , u16 acs_flags )
{
if ( ! pci_is_pcie ( dev ) | |
( ( pci_pcie_type ( dev ) ! = PCI_EXP_TYPE_ROOT_PORT ) & &
( pci_pcie_type ( dev ) ! = PCI_EXP_TYPE_DOWNSTREAM ) ) )
return - ENOTTY ;
switch ( dev - > device ) {
case 0x0710 . . . 0x071e :
case 0x0721 :
case 0x0723 . . . 0x0732 :
return pci_acs_ctrl_enabled ( acs_flags ,
PCI_ACS_SV | PCI_ACS_RR | PCI_ACS_CR | PCI_ACS_UF ) ;
}
return false ;
}
2014-02-04 01:27:46 +04:00
/*
2019-09-06 01:54:42 +03:00
* Many Intel PCH Root Ports do provide ACS - like features to disable peer
2014-02-04 01:27:46 +04:00
* transactions and validate bus numbers in requests , but do not provide an
* actual PCIe ACS capability . This is the list of device IDs known to fall
* into that category as provided by Intel in Red Hat bugzilla 1037684.
*/
static const u16 pci_quirk_intel_pch_acs_ids [ ] = {
/* Ibexpeak PCH */
0x3b42 , 0x3b43 , 0x3b44 , 0x3b45 , 0x3b46 , 0x3b47 , 0x3b48 , 0x3b49 ,
0x3b4a , 0x3b4b , 0x3b4c , 0x3b4d , 0x3b4e , 0x3b4f , 0x3b50 , 0x3b51 ,
/* Cougarpoint PCH */
0x1c10 , 0x1c11 , 0x1c12 , 0x1c13 , 0x1c14 , 0x1c15 , 0x1c16 , 0x1c17 ,
0x1c18 , 0x1c19 , 0x1c1a , 0x1c1b , 0x1c1c , 0x1c1d , 0x1c1e , 0x1c1f ,
/* Pantherpoint PCH */
0x1e10 , 0x1e11 , 0x1e12 , 0x1e13 , 0x1e14 , 0x1e15 , 0x1e16 , 0x1e17 ,
0x1e18 , 0x1e19 , 0x1e1a , 0x1e1b , 0x1e1c , 0x1e1d , 0x1e1e , 0x1e1f ,
/* Lynxpoint-H PCH */
0x8c10 , 0x8c11 , 0x8c12 , 0x8c13 , 0x8c14 , 0x8c15 , 0x8c16 , 0x8c17 ,
0x8c18 , 0x8c19 , 0x8c1a , 0x8c1b , 0x8c1c , 0x8c1d , 0x8c1e , 0x8c1f ,
/* Lynxpoint-LP PCH */
0x9c10 , 0x9c11 , 0x9c12 , 0x9c13 , 0x9c14 , 0x9c15 , 0x9c16 , 0x9c17 ,
0x9c18 , 0x9c19 , 0x9c1a , 0x9c1b ,
/* Wildcat PCH */
0x9c90 , 0x9c91 , 0x9c92 , 0x9c93 , 0x9c94 , 0x9c95 , 0x9c96 , 0x9c97 ,
0x9c98 , 0x9c99 , 0x9c9a , 0x9c9b ,
2014-03-31 22:21:38 +04:00
/* Patsburg (X79) PCH */
0x1d10 , 0x1d12 , 0x1d14 , 0x1d16 , 0x1d18 , 0x1d1a , 0x1d1c , 0x1d1e ,
2015-01-22 21:15:43 +03:00
/* Wellsburg (X99) PCH */
0x8d10 , 0x8d11 , 0x8d12 , 0x8d13 , 0x8d14 , 0x8d15 , 0x8d16 , 0x8d17 ,
0x8d18 , 0x8d19 , 0x8d1a , 0x8d1b , 0x8d1c , 0x8d1d , 0x8d1e ,
2015-05-01 22:20:13 +03:00
/* Lynx Point (9 series) PCH */
0x8c90 , 0x8c92 , 0x8c94 , 0x8c96 , 0x8c98 , 0x8c9a , 0x8c9c , 0x8c9e ,
2014-02-04 01:27:46 +04:00
} ;
static bool pci_quirk_intel_pch_acs_match ( struct pci_dev * dev )
{
int i ;
/* Filter out a few obvious non-matches first */
if ( ! pci_is_pcie ( dev ) | | pci_pcie_type ( dev ) ! = PCI_EXP_TYPE_ROOT_PORT )
return false ;
for ( i = 0 ; i < ARRAY_SIZE ( pci_quirk_intel_pch_acs_ids ) ; i + + )
if ( pci_quirk_intel_pch_acs_ids [ i ] = = dev - > device )
return true ;
return false ;
}
static int pci_quirk_intel_pch_acs ( struct pci_dev * dev , u16 acs_flags )
{
if ( ! pci_quirk_intel_pch_acs_match ( dev ) )
return - ENOTTY ;
2019-09-06 01:54:42 +03:00
if ( dev - > dev_flags & PCI_DEV_FLAGS_ACS_ENABLED_QUIRK )
2019-09-07 02:36:06 +03:00
return pci_acs_ctrl_enabled ( acs_flags ,
PCI_ACS_SV | PCI_ACS_RR | PCI_ACS_CR | PCI_ACS_UF ) ;
2019-09-06 01:54:42 +03:00
2019-09-07 02:36:06 +03:00
return pci_acs_ctrl_enabled ( acs_flags , 0 ) ;
2014-02-04 01:27:46 +04:00
}
2017-02-17 01:01:45 +03:00
/*
2019-09-06 01:54:42 +03:00
* These QCOM Root Ports do provide ACS - like features to disable peer
2017-02-17 01:01:45 +03:00
* transactions and validate bus numbers in requests , but do not provide an
* actual PCIe ACS capability . Hardware supports source validation but it
* will report the issue as Completer Abort instead of ACS Violation .
2019-09-06 01:54:42 +03:00
* Hardware doesn ' t support peer - to - peer and each Root Port is a Root
* Complex with unique segment numbers . It is not possible for one Root
* Port to pass traffic to another Root Port . All PCIe transactions are
* terminated inside the Root Port .
2017-02-17 01:01:45 +03:00
*/
static int pci_quirk_qcom_rp_acs ( struct pci_dev * dev , u16 acs_flags )
{
2019-09-07 02:36:06 +03:00
return pci_acs_ctrl_enabled ( acs_flags ,
PCI_ACS_SV | PCI_ACS_RR | PCI_ACS_CR | PCI_ACS_UF ) ;
2017-02-17 01:01:45 +03:00
}
2019-09-12 16:00:40 +03:00
static int pci_quirk_al_acs ( struct pci_dev * dev , u16 acs_flags )
{
if ( pci_pcie_type ( dev ) ! = PCI_EXP_TYPE_ROOT_PORT )
return - ENOTTY ;
/*
* Amazon ' s Annapurna Labs root ports don ' t include an ACS capability ,
* but do include ACS - like functionality . The hardware doesn ' t support
* peer - to - peer transactions via the root port and each has a unique
* segment number .
*
* Additionally , the root ports cannot send traffic to each other .
*/
acs_flags & = ~ ( PCI_ACS_SV | PCI_ACS_RR | PCI_ACS_CR | PCI_ACS_UF ) ;
return acs_flags ? 0 : 1 ;
}
2016-04-01 01:34:37 +03:00
/*
* Sunrise Point PCH root ports implement ACS , but unfortunately as shown in
* the datasheet ( Intel 100 Series Chipset Family PCH Datasheet , Vol . 2 ,
* 12.1 .46 , 12.1 .47 ) [ 1 ] this chipset uses dwords for the ACS capability and
* control registers whereas the PCIe spec packs them into words ( Rev 3.0 ,
* 7.16 ACS Extended Capability ) . The bit definitions are correct , but the
* control register is at offset 8 instead of 6 and we should probably use
* dword accesses to them . This applies to the following PCI Device IDs , as
* found in volume 1 of the datasheet [ 2 ] :
*
* 0xa110 - 0xa11f Sunrise Point - H PCI Express Root Port # { 0 - 16 }
* 0xa167 - 0xa16a Sunrise Point - H PCI Express Root Port # { 17 - 20 }
*
* N . B . This doesn ' t fix what lspci shows .
*
2017-01-19 18:51:30 +03:00
* The 100 series chipset specification update includes this as errata # 23 [ 3 ] .
*
* The 200 series chipset ( Union Point ) has the same bug according to the
* specification update ( Intel 200 Series Chipset Family Platform Controller
* Hub , Specification Update , January 2017 , Revision 001 , Document # 335194 - 001 ,
* Errata 22 ) [ 4 ] . Per the datasheet [ 5 ] , root port PCI Device IDs for this
* chipset include :
*
* 0xa290 - 0xa29f PCI Express Root port # { 0 - 16 }
* 0xa2e7 - 0xa2ee PCI Express Root port # { 17 - 24 }
*
2018-04-25 23:27:37 +03:00
* Mobile chipsets are also affected , 7 th & 8 th Generation
* Specification update confirms ACS errata 22 , status no fix : ( 7 th Generation
* Intel Processor Family I / O for U / Y Platforms and 8 th Generation Intel
* Processor Family I / O for U Quad Core Platforms Specification Update ,
* August 2017 , Revision 002 , Document # : 334660 - 002 ) [ 6 ]
* Device IDs from I / O datasheet : ( 7 th Generation Intel Processor Family I / O
* for U / Y Platforms and 8 th Generation Intel ® Processor Family I / O for U
* Quad Core Platforms , Vol 1 of 2 , August 2017 , Document # : 334658 - 003 ) [ 7 ]
*
* 0x9d10 - 0x9d1b PCI Express Root port # { 1 - 12 }
*
2020-06-27 13:30:50 +03:00
* [ 1 ] https : //www.intel.com/content/www/us/en/chipsets/100-series-chipset-datasheet-vol-2.html
* [ 2 ] https : //www.intel.com/content/www/us/en/chipsets/100-series-chipset-datasheet-vol-1.html
* [ 3 ] https : //www.intel.com/content/www/us/en/chipsets/100-series-chipset-spec-update.html
* [ 4 ] https : //www.intel.com/content/www/us/en/chipsets/200-series-chipset-pch-spec-update.html
* [ 5 ] https : //www.intel.com/content/www/us/en/chipsets/200-series-chipset-pch-datasheet-vol-1.html
2018-04-25 23:27:37 +03:00
* [ 6 ] https : //www.intel.com/content/www/us/en/processors/core/7th-gen-core-family-mobile-u-y-processor-lines-i-o-spec-update.html
* [ 7 ] https : //www.intel.com/content/www/us/en/processors/core/7th-gen-core-family-mobile-u-y-processor-lines-i-o-datasheet-vol-1.html
2016-04-01 01:34:37 +03:00
*/
static bool pci_quirk_intel_spt_pch_acs_match ( struct pci_dev * dev )
{
2017-01-19 18:51:30 +03:00
if ( ! pci_is_pcie ( dev ) | | pci_pcie_type ( dev ) ! = PCI_EXP_TYPE_ROOT_PORT )
return false ;
switch ( dev - > device ) {
case 0xa110 . . . 0xa11f : case 0xa167 . . . 0xa16a : /* Sunrise Point */
case 0xa290 . . . 0xa29f : case 0xa2e7 . . . 0xa2ee : /* Union Point */
2018-04-25 23:27:37 +03:00
case 0x9d10 . . . 0x9d1b : /* 7th & 8th Gen Mobile */
2017-01-19 18:51:30 +03:00
return true ;
}
return false ;
2016-04-01 01:34:37 +03:00
}
# define INTEL_SPT_ACS_CTRL (PCI_ACS_CAP + 4)
static int pci_quirk_intel_spt_pch_acs ( struct pci_dev * dev , u16 acs_flags )
{
int pos ;
u32 cap , ctrl ;
if ( ! pci_quirk_intel_spt_pch_acs_match ( dev ) )
return - ENOTTY ;
2020-07-08 01:46:02 +03:00
pos = dev - > acs_cap ;
2016-04-01 01:34:37 +03:00
if ( ! pos )
return - ENOTTY ;
/* see pci_acs_flags_enabled() */
pci_read_config_dword ( dev , pos + PCI_ACS_CAP , & cap ) ;
acs_flags & = ( cap | PCI_ACS_EC ) ;
pci_read_config_dword ( dev , pos + INTEL_SPT_ACS_CTRL , & ctrl ) ;
2019-09-07 02:36:06 +03:00
return pci_acs_ctrl_enabled ( acs_flags , ctrl ) ;
2016-04-01 01:34:37 +03:00
}
2014-09-27 03:07:59 +04:00
static int pci_quirk_mf_endpoint_acs ( struct pci_dev * dev , u16 acs_flags )
2014-09-17 18:59:36 +04:00
{
/*
* SV , TB , and UF are not relevant to multifunction endpoints .
*
2014-09-27 03:07:59 +04:00
* Multifunction devices are only required to implement RR , CR , and DT
* in their ACS capability if they support peer - to - peer transactions .
* Devices matching this quirk have been verified by the vendor to not
* perform peer - to - peer with other functions , allowing us to mask out
* these bits as if they were unimplemented in the ACS capability .
2014-09-17 18:59:36 +04:00
*/
2019-09-07 02:36:06 +03:00
return pci_acs_ctrl_enabled ( acs_flags ,
PCI_ACS_SV | PCI_ACS_TB | PCI_ACS_RR |
PCI_ACS_CR | PCI_ACS_UF | PCI_ACS_DT ) ;
2014-09-17 18:59:36 +04:00
}
PCI: Add ACS quirk for Intel Root Complex Integrated Endpoints
All Intel platforms guarantee that all root complex implementations must
send transactions up to IOMMU for address translations. Hence for Intel
RCiEP devices, we can assume some ACS-type isolation even without an ACS
capability.
From the Intel VT-d spec, r3.1, sec 3.16 ("Root-Complex Peer to Peer
Considerations"):
When DMA remapping is enabled, peer-to-peer requests through the
Root-Complex must be handled as follows:
- The input address in the request is translated (through first-level,
second-level or nested translation) to a host physical address (HPA).
The address decoding for peer addresses must be done only on the
translated HPA. Hardware implementations are free to further limit
peer-to-peer accesses to specific host physical address regions (or
to completely disallow peer-forwarding of translated requests).
- Since address translation changes the contents (address field) of
the PCI Express Transaction Layer Packet (TLP), for PCI Express
peer-to-peer requests with ECRC, the Root-Complex hardware must use
the new ECRC (re-computed with the translated address) if it
decides to forward the TLP as a peer request.
- Root-ports, and multi-function root-complex integrated endpoints, may
support additional peer-to-peer control features by supporting PCI
Express Access Control Services (ACS) capability. Refer to ACS
capability in PCI Express specifications for details.
Since Linux didn't give special treatment to allow this exception, certain
RCiEP MFD devices were grouped in a single IOMMU group. This doesn't permit
a single device to be assigned to a guest for instance.
In one vendor system: Device 14.x were grouped in a single IOMMU group.
/sys/kernel/iommu_groups/5/devices/0000:00:14.0
/sys/kernel/iommu_groups/5/devices/0000:00:14.2
/sys/kernel/iommu_groups/5/devices/0000:00:14.3
After this patch:
/sys/kernel/iommu_groups/5/devices/0000:00:14.0
/sys/kernel/iommu_groups/5/devices/0000:00:14.2
/sys/kernel/iommu_groups/6/devices/0000:00:14.3 <<< new group
14.0 and 14.2 are integrated devices, but legacy end points, whereas 14.3
was a PCIe-compliant RCiEP.
00:14.3 Network controller: Intel Corporation Device 9df0 (rev 30)
Capabilities: [40] Express (v2) Root Complex Integrated Endpoint, MSI 00
This permits assigning this device to a guest VM.
[bhelgaas: drop "Fixes" tag since this doesn't fix a bug in that commit]
Link: https://lore.kernel.org/r/1590699462-7131-1-git-send-email-ashok.raj@intel.com
Tested-by: Darrel Goeddel <dgoeddel@forcepoint.com>
Signed-off-by: Ashok Raj <ashok.raj@intel.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Alex Williamson <alex.williamson@redhat.com>
Cc: stable@vger.kernel.org
Cc: Lu Baolu <baolu.lu@linux.intel.com>
Cc: Mark Scott <mscott@forcepoint.com>,
Cc: Romil Sharma <rsharma@forcepoint.com>
2020-05-28 23:57:42 +03:00
static int pci_quirk_rciep_acs ( struct pci_dev * dev , u16 acs_flags )
{
/*
* Intel RCiEP ' s are required to allow p2p only on translated
* addresses . Refer to Intel VT - d specification , r3 .1 , sec 3.16 ,
* " Root-Complex Peer to Peer Considerations " .
*/
if ( pci_pcie_type ( dev ) ! = PCI_EXP_TYPE_RC_END )
return - ENOTTY ;
return pci_acs_ctrl_enabled ( acs_flags ,
PCI_ACS_SV | PCI_ACS_RR | PCI_ACS_CR | PCI_ACS_UF ) ;
}
2019-08-20 07:39:45 +03:00
static int pci_quirk_brcm_acs ( struct pci_dev * dev , u16 acs_flags )
{
/*
* iProc PAXB Root Ports don ' t advertise an ACS capability , but
* they do not allow peer - to - peer transactions between Root Ports .
* Allow each Root Port to be in a separate IOMMU group by masking
* SV / RR / CR / UF bits .
*/
2019-09-07 02:36:06 +03:00
return pci_acs_ctrl_enabled ( acs_flags ,
PCI_ACS_SV | PCI_ACS_RR | PCI_ACS_CR | PCI_ACS_UF ) ;
2019-08-20 07:39:45 +03:00
}
2012-06-11 09:27:07 +04:00
static const struct pci_dev_acs_enabled {
u16 vendor ;
u16 device ;
int ( * acs_enabled ) ( struct pci_dev * dev , u16 acs_flags ) ;
} pci_dev_acs_enabled [ ] = {
2013-06-28 02:40:00 +04:00
{ PCI_VENDOR_ID_ATI , 0x4385 , pci_quirk_amd_sb_acs } ,
{ PCI_VENDOR_ID_ATI , 0x439c , pci_quirk_amd_sb_acs } ,
{ PCI_VENDOR_ID_ATI , 0x4383 , pci_quirk_amd_sb_acs } ,
{ PCI_VENDOR_ID_ATI , 0x439d , pci_quirk_amd_sb_acs } ,
{ PCI_VENDOR_ID_ATI , 0x4384 , pci_quirk_amd_sb_acs } ,
{ PCI_VENDOR_ID_ATI , 0x4399 , pci_quirk_amd_sb_acs } ,
2014-10-02 18:50:31 +04:00
{ PCI_VENDOR_ID_AMD , 0x780f , pci_quirk_amd_sb_acs } ,
{ PCI_VENDOR_ID_AMD , 0x7809 , pci_quirk_amd_sb_acs } ,
2014-09-27 03:07:59 +04:00
{ PCI_VENDOR_ID_SOLARFLARE , 0x0903 , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_SOLARFLARE , 0x0923 , pci_quirk_mf_endpoint_acs } ,
2016-07-28 20:13:56 +03:00
{ PCI_VENDOR_ID_SOLARFLARE , 0x0A03 , pci_quirk_mf_endpoint_acs } ,
2014-09-27 03:07:59 +04:00
{ PCI_VENDOR_ID_INTEL , 0x10C6 , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x10DB , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x10DD , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x10E1 , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x10F1 , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x10F7 , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x10F8 , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x10F9 , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x10FA , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x10FB , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x10FC , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x1507 , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x1514 , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x151C , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x1529 , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x152A , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x154D , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x154F , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x1551 , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x1558 , pci_quirk_mf_endpoint_acs } ,
2015-03-20 21:27:57 +03:00
/* 82580 */
{ PCI_VENDOR_ID_INTEL , 0x1509 , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x150E , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x150F , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x1510 , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x1511 , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x1516 , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x1527 , pci_quirk_mf_endpoint_acs } ,
/* 82576 */
{ PCI_VENDOR_ID_INTEL , 0x10C9 , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x10E6 , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x10E7 , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x10E8 , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x150A , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x150D , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x1518 , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x1526 , pci_quirk_mf_endpoint_acs } ,
/* 82575 */
{ PCI_VENDOR_ID_INTEL , 0x10A7 , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x10A9 , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x10D6 , pci_quirk_mf_endpoint_acs } ,
/* I350 */
{ PCI_VENDOR_ID_INTEL , 0x1521 , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x1522 , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x1523 , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x1524 , pci_quirk_mf_endpoint_acs } ,
/* 82571 (Quads omitted due to non-ACS switch) */
{ PCI_VENDOR_ID_INTEL , 0x105E , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x105F , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x1060 , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x10D9 , pci_quirk_mf_endpoint_acs } ,
2015-08-10 21:32:04 +03:00
/* I219 */
{ PCI_VENDOR_ID_INTEL , 0x15b7 , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_INTEL , 0x15b8 , pci_quirk_mf_endpoint_acs } ,
PCI: Add ACS quirk for Intel Root Complex Integrated Endpoints
All Intel platforms guarantee that all root complex implementations must
send transactions up to IOMMU for address translations. Hence for Intel
RCiEP devices, we can assume some ACS-type isolation even without an ACS
capability.
From the Intel VT-d spec, r3.1, sec 3.16 ("Root-Complex Peer to Peer
Considerations"):
When DMA remapping is enabled, peer-to-peer requests through the
Root-Complex must be handled as follows:
- The input address in the request is translated (through first-level,
second-level or nested translation) to a host physical address (HPA).
The address decoding for peer addresses must be done only on the
translated HPA. Hardware implementations are free to further limit
peer-to-peer accesses to specific host physical address regions (or
to completely disallow peer-forwarding of translated requests).
- Since address translation changes the contents (address field) of
the PCI Express Transaction Layer Packet (TLP), for PCI Express
peer-to-peer requests with ECRC, the Root-Complex hardware must use
the new ECRC (re-computed with the translated address) if it
decides to forward the TLP as a peer request.
- Root-ports, and multi-function root-complex integrated endpoints, may
support additional peer-to-peer control features by supporting PCI
Express Access Control Services (ACS) capability. Refer to ACS
capability in PCI Express specifications for details.
Since Linux didn't give special treatment to allow this exception, certain
RCiEP MFD devices were grouped in a single IOMMU group. This doesn't permit
a single device to be assigned to a guest for instance.
In one vendor system: Device 14.x were grouped in a single IOMMU group.
/sys/kernel/iommu_groups/5/devices/0000:00:14.0
/sys/kernel/iommu_groups/5/devices/0000:00:14.2
/sys/kernel/iommu_groups/5/devices/0000:00:14.3
After this patch:
/sys/kernel/iommu_groups/5/devices/0000:00:14.0
/sys/kernel/iommu_groups/5/devices/0000:00:14.2
/sys/kernel/iommu_groups/6/devices/0000:00:14.3 <<< new group
14.0 and 14.2 are integrated devices, but legacy end points, whereas 14.3
was a PCIe-compliant RCiEP.
00:14.3 Network controller: Intel Corporation Device 9df0 (rev 30)
Capabilities: [40] Express (v2) Root Complex Integrated Endpoint, MSI 00
This permits assigning this device to a guest VM.
[bhelgaas: drop "Fixes" tag since this doesn't fix a bug in that commit]
Link: https://lore.kernel.org/r/1590699462-7131-1-git-send-email-ashok.raj@intel.com
Tested-by: Darrel Goeddel <dgoeddel@forcepoint.com>
Signed-off-by: Ashok Raj <ashok.raj@intel.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Alex Williamson <alex.williamson@redhat.com>
Cc: stable@vger.kernel.org
Cc: Lu Baolu <baolu.lu@linux.intel.com>
Cc: Mark Scott <mscott@forcepoint.com>,
Cc: Romil Sharma <rsharma@forcepoint.com>
2020-05-28 23:57:42 +03:00
{ PCI_VENDOR_ID_INTEL , PCI_ANY_ID , pci_quirk_rciep_acs } ,
2017-02-17 01:01:45 +03:00
/* QCOM QDF2xxx root ports */
2018-05-07 23:52:55 +03:00
{ PCI_VENDOR_ID_QCOM , 0x0400 , pci_quirk_qcom_rp_acs } ,
{ PCI_VENDOR_ID_QCOM , 0x0401 , pci_quirk_qcom_rp_acs } ,
2019-02-02 02:13:10 +03:00
/* HXT SD4800 root ports. The ACS design is same as QCOM QDF2xxx */
{ PCI_VENDOR_ID_HXT , 0x0401 , pci_quirk_qcom_rp_acs } ,
2015-03-20 21:27:57 +03:00
/* Intel PCH root ports */
2014-02-04 01:27:46 +04:00
{ PCI_VENDOR_ID_INTEL , PCI_ANY_ID , pci_quirk_intel_pch_acs } ,
2016-04-01 01:34:37 +03:00
{ PCI_VENDOR_ID_INTEL , PCI_ANY_ID , pci_quirk_intel_spt_pch_acs } ,
2015-01-13 09:22:23 +03:00
{ 0x19a2 , 0x710 , pci_quirk_mf_endpoint_acs } , /* Emulex BE3-R */
{ 0x10df , 0x720 , pci_quirk_mf_endpoint_acs } , /* Emulex Skyhawk-R */
2016-01-29 23:03:58 +03:00
/* Cavium ThunderX */
{ PCI_VENDOR_ID_CAVIUM , PCI_ANY_ID , pci_quirk_cavium_acs } ,
2017-08-11 00:06:33 +03:00
/* APM X-Gene */
{ PCI_VENDOR_ID_AMCC , 0xE004 , pci_quirk_xgene_acs } ,
2018-02-21 06:19:27 +03:00
/* Ampere Computing */
{ PCI_VENDOR_ID_AMPERE , 0xE005 , pci_quirk_xgene_acs } ,
{ PCI_VENDOR_ID_AMPERE , 0xE006 , pci_quirk_xgene_acs } ,
{ PCI_VENDOR_ID_AMPERE , 0xE007 , pci_quirk_xgene_acs } ,
{ PCI_VENDOR_ID_AMPERE , 0xE008 , pci_quirk_xgene_acs } ,
{ PCI_VENDOR_ID_AMPERE , 0xE009 , pci_quirk_xgene_acs } ,
{ PCI_VENDOR_ID_AMPERE , 0xE00A , pci_quirk_xgene_acs } ,
{ PCI_VENDOR_ID_AMPERE , 0xE00B , pci_quirk_xgene_acs } ,
{ PCI_VENDOR_ID_AMPERE , 0xE00C , pci_quirk_xgene_acs } ,
2019-08-20 07:39:45 +03:00
{ PCI_VENDOR_ID_BROADCOM , 0xD714 , pci_quirk_brcm_acs } ,
2019-09-12 16:00:40 +03:00
/* Amazon Annapurna Labs */
{ PCI_VENDOR_ID_AMAZON_ANNAPURNA_LABS , 0x0031 , pci_quirk_al_acs } ,
2020-03-27 12:11:47 +03:00
/* Zhaoxin multi-function devices */
{ PCI_VENDOR_ID_ZHAOXIN , 0x3038 , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_ZHAOXIN , 0x3104 , pci_quirk_mf_endpoint_acs } ,
{ PCI_VENDOR_ID_ZHAOXIN , 0x9083 , pci_quirk_mf_endpoint_acs } ,
2020-03-27 12:11:48 +03:00
/* Zhaoxin Root/Downstream Ports */
{ PCI_VENDOR_ID_ZHAOXIN , PCI_ANY_ID , pci_quirk_zhaoxin_pcie_ports_acs } ,
2012-06-11 09:27:07 +04:00
{ 0 }
} ;
2019-09-07 02:36:06 +03:00
/*
* pci_dev_specific_acs_enabled - check whether device provides ACS controls
* @ dev : PCI device
* @ acs_flags : Bitmask of desired ACS controls
*
* Returns :
* - ENOTTY : No quirk applies to this device ; we can ' t tell whether the
* device provides the desired controls
* 0 : Device does not provide all the desired controls
* > 0 : Device provides all the controls in @ acs_flags
*/
2012-06-11 09:27:07 +04:00
int pci_dev_specific_acs_enabled ( struct pci_dev * dev , u16 acs_flags )
{
const struct pci_dev_acs_enabled * i ;
int ret ;
/*
* Allow devices that do not expose standard PCIe ACS capabilities
* or control to indicate their support here . Multi - function express
* devices which do not allow internal peer - to - peer between functions ,
* but do not implement PCIe ACS may wish to return true here .
*/
for ( i = pci_dev_acs_enabled ; i - > acs_enabled ; i + + ) {
if ( ( i - > vendor = = dev - > vendor | |
i - > vendor = = ( u16 ) PCI_ANY_ID ) & &
( i - > device = = dev - > device | |
i - > device = = ( u16 ) PCI_ANY_ID ) ) {
ret = i - > acs_enabled ( dev , acs_flags ) ;
if ( ret > = 0 )
return ret ;
}
}
return - ENOTTY ;
}
2014-02-04 01:27:33 +04:00
2014-02-04 01:27:46 +04:00
/* Config space offset of Root Complex Base Address register */
# define INTEL_LPC_RCBA_REG 0xf0
/* 31:14 RCBA address */
# define INTEL_LPC_RCBA_MASK 0xffffc000
/* RCBA Enable */
# define INTEL_LPC_RCBA_ENABLE (1 << 0)
/* Backbone Scratch Pad Register */
# define INTEL_BSPR_REG 0x1104
/* Backbone Peer Non-Posted Disable */
# define INTEL_BSPR_REG_BPNPD (1 << 8)
/* Backbone Peer Posted Disable */
# define INTEL_BSPR_REG_BPPD (1 << 9)
/* Upstream Peer Decode Configuration Register */
2019-09-18 16:16:52 +03:00
# define INTEL_UPDCR_REG 0x1014
2014-02-04 01:27:46 +04:00
/* 5:0 Peer Decode Enable bits */
# define INTEL_UPDCR_REG_MASK 0x3f
static int pci_quirk_enable_intel_lpc_acs ( struct pci_dev * dev )
{
u32 rcba , bspr , updcr ;
void __iomem * rcba_mem ;
/*
* Read the RCBA register from the LPC ( D31 : F0 ) . PCH root ports
* are D28 : F * and therefore get probed before LPC , thus we can ' t
2018-05-02 16:53:19 +03:00
* use pci_get_slot ( ) / pci_read_config_dword ( ) here .
2014-02-04 01:27:46 +04:00
*/
pci_bus_read_config_dword ( dev - > bus , PCI_DEVFN ( 31 , 0 ) ,
INTEL_LPC_RCBA_REG , & rcba ) ;
if ( ! ( rcba & INTEL_LPC_RCBA_ENABLE ) )
return - EINVAL ;
2020-01-06 11:43:50 +03:00
rcba_mem = ioremap ( rcba & INTEL_LPC_RCBA_MASK ,
2014-02-04 01:27:46 +04:00
PAGE_ALIGN ( INTEL_UPDCR_REG ) ) ;
if ( ! rcba_mem )
return - ENOMEM ;
/*
* The BSPR can disallow peer cycles , but it ' s set by soft strap and
* therefore read - only . If both posted and non - posted peer cycles are
* disallowed , we ' re ok . If either are allowed , then we need to use
* the UPDCR to disable peer decodes for each port . This provides the
* PCIe ACS equivalent of PCI_ACS_RR | PCI_ACS_CR | PCI_ACS_UF
*/
bspr = readl ( rcba_mem + INTEL_BSPR_REG ) ;
bspr & = INTEL_BSPR_REG_BPNPD | INTEL_BSPR_REG_BPPD ;
if ( bspr ! = ( INTEL_BSPR_REG_BPNPD | INTEL_BSPR_REG_BPPD ) ) {
updcr = readl ( rcba_mem + INTEL_UPDCR_REG ) ;
if ( updcr & INTEL_UPDCR_REG_MASK ) {
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Disabling UPDCR peer decodes \n " ) ;
2014-02-04 01:27:46 +04:00
updcr & = ~ INTEL_UPDCR_REG_MASK ;
writel ( updcr , rcba_mem + INTEL_UPDCR_REG ) ;
}
}
iounmap ( rcba_mem ) ;
return 0 ;
}
/* Miscellaneous Port Configuration register */
# define INTEL_MPC_REG 0xd8
/* MPC: Invalid Receive Bus Number Check Enable */
# define INTEL_MPC_REG_IRBNCE (1 << 26)
static void pci_quirk_enable_intel_rp_mpc_acs ( struct pci_dev * dev )
{
u32 mpc ;
/*
* When enabled , the IRBNCE bit of the MPC register enables the
* equivalent of PCI ACS Source Validation ( PCI_ACS_SV ) , which
* ensures that requester IDs fall within the bus number range
* of the bridge . Enable if not already .
*/
pci_read_config_dword ( dev , INTEL_MPC_REG , & mpc ) ;
if ( ! ( mpc & INTEL_MPC_REG_IRBNCE ) ) {
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Enabling MPC IRBNCE \n " ) ;
2014-02-04 01:27:46 +04:00
mpc | = INTEL_MPC_REG_IRBNCE ;
pci_write_config_word ( dev , INTEL_MPC_REG , mpc ) ;
}
}
2020-07-08 01:46:04 +03:00
/*
* Currently this quirk does the equivalent of
* PCI_ACS_SV | PCI_ACS_RR | PCI_ACS_CR | PCI_ACS_UF
*
* TODO : This quirk also needs to do equivalent of PCI_ACS_TB ,
* if dev - > external_facing | | dev - > untrusted
*/
2014-02-04 01:27:46 +04:00
static int pci_quirk_enable_intel_pch_acs ( struct pci_dev * dev )
{
if ( ! pci_quirk_intel_pch_acs_match ( dev ) )
return - ENOTTY ;
if ( pci_quirk_enable_intel_lpc_acs ( dev ) ) {
2018-01-18 21:55:24 +03:00
pci_warn ( dev , " Failed to enable Intel PCH ACS quirk \n " ) ;
2014-02-04 01:27:46 +04:00
return 0 ;
}
pci_quirk_enable_intel_rp_mpc_acs ( dev ) ;
dev - > dev_flags | = PCI_DEV_FLAGS_ACS_ENABLED_QUIRK ;
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Intel PCH root port ACS workaround enabled \n " ) ;
2014-02-04 01:27:46 +04:00
return 0 ;
}
2016-04-01 01:34:37 +03:00
static int pci_quirk_enable_intel_spt_pch_acs ( struct pci_dev * dev )
{
int pos ;
u32 cap , ctrl ;
if ( ! pci_quirk_intel_spt_pch_acs_match ( dev ) )
return - ENOTTY ;
2020-07-08 01:46:02 +03:00
pos = dev - > acs_cap ;
2016-04-01 01:34:37 +03:00
if ( ! pos )
return - ENOTTY ;
pci_read_config_dword ( dev , pos + PCI_ACS_CAP , & cap ) ;
pci_read_config_dword ( dev , pos + INTEL_SPT_ACS_CTRL , & ctrl ) ;
ctrl | = ( cap & PCI_ACS_SV ) ;
ctrl | = ( cap & PCI_ACS_RR ) ;
ctrl | = ( cap & PCI_ACS_CR ) ;
ctrl | = ( cap & PCI_ACS_UF ) ;
2020-07-08 01:46:04 +03:00
if ( dev - > external_facing | | dev - > untrusted )
ctrl | = ( cap & PCI_ACS_TB ) ;
2016-04-01 01:34:37 +03:00
pci_write_config_dword ( dev , pos + INTEL_SPT_ACS_CTRL , ctrl ) ;
2018-01-18 21:55:24 +03:00
pci_info ( dev , " Intel SPT PCH root port ACS workaround enabled \n " ) ;
2016-04-01 01:34:37 +03:00
return 0 ;
}
2018-08-10 01:09:17 +03:00
static int pci_quirk_disable_intel_spt_pch_acs_redir ( struct pci_dev * dev )
{
int pos ;
u32 cap , ctrl ;
if ( ! pci_quirk_intel_spt_pch_acs_match ( dev ) )
return - ENOTTY ;
2020-07-08 01:46:02 +03:00
pos = dev - > acs_cap ;
2018-08-10 01:09:17 +03:00
if ( ! pos )
return - ENOTTY ;
pci_read_config_dword ( dev , pos + PCI_ACS_CAP , & cap ) ;
pci_read_config_dword ( dev , pos + INTEL_SPT_ACS_CTRL , & ctrl ) ;
ctrl & = ~ ( PCI_ACS_RR | PCI_ACS_CR | PCI_ACS_EC ) ;
pci_write_config_dword ( dev , pos + INTEL_SPT_ACS_CTRL , ctrl ) ;
pci_info ( dev , " Intel SPT PCH root port workaround: disabled ACS redirect \n " ) ;
return 0 ;
}
2018-08-10 00:51:43 +03:00
static const struct pci_dev_acs_ops {
2014-02-04 01:27:33 +04:00
u16 vendor ;
u16 device ;
int ( * enable_acs ) ( struct pci_dev * dev ) ;
2018-08-10 00:51:43 +03:00
int ( * disable_acs_redir ) ( struct pci_dev * dev ) ;
} pci_dev_acs_ops [ ] = {
{ PCI_VENDOR_ID_INTEL , PCI_ANY_ID ,
. enable_acs = pci_quirk_enable_intel_pch_acs ,
} ,
{ PCI_VENDOR_ID_INTEL , PCI_ANY_ID ,
. enable_acs = pci_quirk_enable_intel_spt_pch_acs ,
2018-08-10 01:09:17 +03:00
. disable_acs_redir = pci_quirk_disable_intel_spt_pch_acs_redir ,
2018-08-10 00:51:43 +03:00
} ,
2014-02-04 01:27:33 +04:00
} ;
2016-04-01 01:34:32 +03:00
int pci_dev_specific_enable_acs ( struct pci_dev * dev )
2014-02-04 01:27:33 +04:00
{
2018-08-10 00:51:43 +03:00
const struct pci_dev_acs_ops * p ;
2018-08-10 00:45:47 +03:00
int i , ret ;
2014-02-04 01:27:33 +04:00
2018-08-10 00:51:43 +03:00
for ( i = 0 ; i < ARRAY_SIZE ( pci_dev_acs_ops ) ; i + + ) {
p = & pci_dev_acs_ops [ i ] ;
2018-08-10 00:45:47 +03:00
if ( ( p - > vendor = = dev - > vendor | |
p - > vendor = = ( u16 ) PCI_ANY_ID ) & &
( p - > device = = dev - > device | |
2018-08-10 00:51:43 +03:00
p - > device = = ( u16 ) PCI_ANY_ID ) & &
p - > enable_acs ) {
2018-08-10 00:45:47 +03:00
ret = p - > enable_acs ( dev ) ;
2014-02-04 01:27:33 +04:00
if ( ret > = 0 )
2018-08-10 00:51:43 +03:00
return ret ;
}
}
2014-02-04 01:27:33 +04:00
2018-08-10 00:51:43 +03:00
return - ENOTTY ;
}
int pci_dev_specific_disable_acs_redir ( struct pci_dev * dev )
{
const struct pci_dev_acs_ops * p ;
int i , ret ;
for ( i = 0 ; i < ARRAY_SIZE ( pci_dev_acs_ops ) ; i + + ) {
p = & pci_dev_acs_ops [ i ] ;
if ( ( p - > vendor = = dev - > vendor | |
p - > vendor = = ( u16 ) PCI_ANY_ID ) & &
( p - > device = = dev - > device | |
p - > device = = ( u16 ) PCI_ANY_ID ) & &
p - > disable_acs_redir ) {
ret = p - > disable_acs_redir ( dev ) ;
2014-02-04 01:27:33 +04:00
if ( ret > = 0 )
2016-04-01 01:34:32 +03:00
return ret ;
2014-02-04 01:27:33 +04:00
}
}
2016-04-01 01:34:32 +03:00
return - ENOTTY ;
2014-02-04 01:27:33 +04:00
}
2015-08-07 21:34:42 +03:00
/*
2018-05-02 16:53:19 +03:00
* The PCI capabilities list for Intel DH895xCC VFs ( device ID 0x0443 ) with
2015-08-07 21:34:42 +03:00
* QuickAssist Technology ( QAT ) is prematurely terminated in hardware . The
* Next Capability pointer in the MSI Capability Structure should point to
* the PCIe Capability Structure but is incorrectly hardwired as 0 terminating
* the list .
*/
static void quirk_intel_qat_vf_cap ( struct pci_dev * pdev )
{
int pos , i = 0 ;
u8 next_cap ;
u16 reg16 , * cap ;
struct pci_cap_saved_state * state ;
/* Bail if the hardware bug is fixed */
if ( pdev - > pcie_cap | | pci_find_capability ( pdev , PCI_CAP_ID_EXP ) )
return ;
/* Bail if MSI Capability Structure is not found for some reason */
pos = pci_find_capability ( pdev , PCI_CAP_ID_MSI ) ;
if ( ! pos )
return ;
/*
* Bail if Next Capability pointer in the MSI Capability Structure
* is not the expected incorrect 0x00 .
*/
pci_read_config_byte ( pdev , pos + 1 , & next_cap ) ;
if ( next_cap )
return ;
/*
* PCIe Capability Structure is expected to be at 0x50 and should
* terminate the list ( Next Capability pointer is 0x00 ) . Verify
* Capability Id and Next Capability pointer is as expected .
* Open - code some of set_pcie_port_type ( ) and pci_cfg_space_size_ext ( )
* to correctly set kernel data structures which have already been
* set incorrectly due to the hardware bug .
*/
pos = 0x50 ;
pci_read_config_word ( pdev , pos , & reg16 ) ;
if ( reg16 = = ( 0x0000 | PCI_CAP_ID_EXP ) ) {
u32 status ;
# ifndef PCI_EXP_SAVE_REGS
# define PCI_EXP_SAVE_REGS 7
# endif
int size = PCI_EXP_SAVE_REGS * sizeof ( u16 ) ;
pdev - > pcie_cap = pos ;
pci_read_config_word ( pdev , pos + PCI_EXP_FLAGS , & reg16 ) ;
pdev - > pcie_flags_reg = reg16 ;
pci_read_config_word ( pdev , pos + PCI_EXP_DEVCAP , & reg16 ) ;
pdev - > pcie_mpss = reg16 & PCI_EXP_DEVCAP_PAYLOAD ;
pdev - > cfg_size = PCI_CFG_SPACE_EXP_SIZE ;
if ( pci_read_config_dword ( pdev , PCI_CFG_SPACE_SIZE , & status ) ! =
PCIBIOS_SUCCESSFUL | | ( status = = 0xffffffff ) )
pdev - > cfg_size = PCI_CFG_SPACE_SIZE ;
if ( pci_find_saved_cap ( pdev , PCI_CAP_ID_EXP ) )
return ;
2018-05-02 16:53:19 +03:00
/* Save PCIe cap */
2015-08-07 21:34:42 +03:00
state = kzalloc ( sizeof ( * state ) + size , GFP_KERNEL ) ;
if ( ! state )
return ;
state - > cap . cap_nr = PCI_CAP_ID_EXP ;
state - > cap . cap_extended = 0 ;
state - > cap . size = size ;
cap = ( u16 * ) & state - > cap . data [ 0 ] ;
pcie_capability_read_word ( pdev , PCI_EXP_DEVCTL , & cap [ i + + ] ) ;
pcie_capability_read_word ( pdev , PCI_EXP_LNKCTL , & cap [ i + + ] ) ;
pcie_capability_read_word ( pdev , PCI_EXP_SLTCTL , & cap [ i + + ] ) ;
pcie_capability_read_word ( pdev , PCI_EXP_RTCTL , & cap [ i + + ] ) ;
pcie_capability_read_word ( pdev , PCI_EXP_DEVCTL2 , & cap [ i + + ] ) ;
pcie_capability_read_word ( pdev , PCI_EXP_LNKCTL2 , & cap [ i + + ] ) ;
pcie_capability_read_word ( pdev , PCI_EXP_SLTCTL2 , & cap [ i + + ] ) ;
hlist_add_head ( & state - > next , & pdev - > saved_cap_space ) ;
}
}
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_INTEL , 0x443 , quirk_intel_qat_vf_cap ) ;
2016-09-06 22:15:24 +03:00
2020-05-21 02:23:30 +03:00
/*
* FLR may cause the following to devices to hang :
*
* AMD Starship / Matisse HD Audio Controller 0x1487
2020-05-24 10:35:29 +03:00
* AMD Starship USB 3.0 Host Controller 0x148c
2020-05-21 02:23:30 +03:00
* AMD Matisse USB 3.0 Host Controller 0x149c
* Intel 82579L M Gigabit Ethernet Controller 0x1502
* Intel 82579 V Gigabit Ethernet Controller 0x1503
*
*/
static void quirk_no_flr ( struct pci_dev * dev )
2017-04-04 00:02:50 +03:00
{
dev - > dev_flags | = PCI_DEV_FLAGS_NO_FLR_RESET ;
}
2020-05-21 02:23:30 +03:00
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_AMD , 0x1487 , quirk_no_flr ) ;
2020-05-24 10:35:29 +03:00
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_AMD , 0x148c , quirk_no_flr ) ;
2020-05-21 02:23:30 +03:00
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_AMD , 0x149c , quirk_no_flr ) ;
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_INTEL , 0x1502 , quirk_no_flr ) ;
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_INTEL , 0x1503 , quirk_no_flr ) ;
2017-07-12 07:04:14 +03:00
static void quirk_no_ext_tags ( struct pci_dev * pdev )
{
struct pci_host_bridge * bridge = pci_find_host_bridge ( pdev - > bus ) ;
if ( ! bridge )
return ;
bridge - > no_ext_tags = 1 ;
2018-01-18 21:55:24 +03:00
pci_info ( pdev , " disabling Extended Tags (this device can't handle them) \n " ) ;
2017-07-12 07:04:14 +03:00
pci_walk_bus ( bridge - > bus , pci_configure_extended_tags , NULL ) ;
}
2018-04-10 22:44:21 +03:00
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_SERVERWORKS , 0x0132 , quirk_no_ext_tags ) ;
2017-07-12 07:04:14 +03:00
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_SERVERWORKS , 0x0140 , quirk_no_ext_tags ) ;
2018-04-10 22:44:21 +03:00
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_SERVERWORKS , 0x0141 , quirk_no_ext_tags ) ;
2017-07-12 07:04:14 +03:00
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_SERVERWORKS , 0x0142 , quirk_no_ext_tags ) ;
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_SERVERWORKS , 0x0144 , quirk_no_ext_tags ) ;
2018-04-10 22:44:21 +03:00
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_SERVERWORKS , 0x0420 , quirk_no_ext_tags ) ;
DECLARE_PCI_FIXUP_EARLY ( PCI_VENDOR_ID_SERVERWORKS , 0x0422 , quirk_no_ext_tags ) ;
2017-09-07 21:24:41 +03:00
2017-07-11 23:48:00 +03:00
# ifdef CONFIG_PCI_ATS
/*
2020-01-15 02:09:28 +03:00
* Some devices require additional driver setup to enable ATS . Don ' t use
* ATS for those devices as ATS will be enabled before the driver has had a
* chance to load and configure the device .
2017-07-11 23:48:00 +03:00
*/
2020-01-15 02:09:28 +03:00
static void quirk_amd_harvest_no_ats ( struct pci_dev * pdev )
2017-07-11 23:48:00 +03:00
{
2020-07-28 13:45:53 +03:00
if ( ( pdev - > device = = 0x7312 & & pdev - > revision ! = 0x00 ) | |
( pdev - > device = = 0x7340 & & pdev - > revision ! = 0xc5 ) )
2020-01-15 02:09:28 +03:00
return ;
pci_info ( pdev , " disabling ATS \n " ) ;
2017-07-11 23:48:00 +03:00
pdev - > ats_cap = 0 ;
}
/* AMD Stoney platform GPU */
2020-01-15 02:09:28 +03:00
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_ATI , 0x98e4 , quirk_amd_harvest_no_ats ) ;
/* AMD Iceland dGPU */
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_ATI , 0x6900 , quirk_amd_harvest_no_ats ) ;
2020-07-28 13:45:53 +03:00
/* AMD Navi10 dGPU */
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_ATI , 0x7312 , quirk_amd_harvest_no_ats ) ;
2020-01-15 02:09:28 +03:00
/* AMD Navi14 dGPU */
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_ATI , 0x7340 , quirk_amd_harvest_no_ats ) ;
2017-07-11 23:48:00 +03:00
# endif /* CONFIG_PCI_ATS */
2017-10-12 12:44:47 +03:00
/* Freescale PCIe doesn't support MSI in RC mode */
static void quirk_fsl_no_msi ( struct pci_dev * pdev )
{
if ( pci_pcie_type ( pdev ) = = PCI_EXP_TYPE_ROOT_PORT )
pdev - > no_msi = 1 ;
}
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_FREESCALE , PCI_ANY_ID , quirk_fsl_no_msi ) ;
vga_switcheroo: Use device link for HDA controller
Back in 2013, runtime PM for GPUs with integrated HDA controller was
introduced with commits 0d69704ae348 ("gpu/vga_switcheroo: add driver
control power feature. (v3)") and 246efa4a072f ("snd/hda: add runtime
suspend/resume on optimus support (v4)").
Briefly, the idea was that the HDA controller is forced on and off in
unison with the GPU.
The original code is mostly still in place even though it was never a
100% perfect solution: E.g. on access to the HDA controller, the GPU
is powered up via vga_switcheroo_runtime_resume_hdmi_audio() but there
are no provisions to keep it resumed until access to the HDA controller
has ceased: The GPU autosuspends after 5 seconds, rendering the HDA
controller inaccessible.
Additionally, a kludge is required when hda_intel.c probes: It has to
check whether the GPU is powered down (check_hdmi_disabled()) and defer
probing if so.
However in the meantime (in v4.10) the driver core has gained a feature
called device links which promises to solve such issues in a clean way:
It allows us to declare a dependency from the HDA controller (consumer)
to the GPU (supplier). The PM core then automagically ensures that the
GPU is runtime resumed as long as the HDA controller's ->probe hook is
executed and whenever the HDA controller is accessed.
By default, the HDA controller has a dependency on its parent, a PCIe
Root Port. Adding a device link creates another dependency on its
sibling:
PCIe Root Port
^ ^
| |
| |
HDA ===> GPU
The device link is not only used for runtime PM, it also guarantees that
on system sleep, the HDA controller suspends before the GPU and resumes
after the GPU, and on system shutdown the HDA controller's ->shutdown
hook is executed before the one of the GPU. It is a complete solution.
Using this functionality is as simple as calling device_link_add(),
which results in a dmesg entry like this:
pci 0000:01:00.1: Linked as a consumer to 0000:01:00.0
The code for the GPU-governed audio power management can thus be removed
(except where it's still needed for legacy manual power control).
The device link is added in a PCI quirk rather than in hda_intel.c.
It is therefore legal for the GPU to runtime suspend to D3cold even if
the HDA controller is not bound to a driver or if CONFIG_SND_HDA_INTEL
is not enabled, for accesses to the HDA controller will cause the GPU to
wake up regardless if they're occurring outside of hda_intel.c (think
config space readout via sysfs).
Contrary to the previous implementation, the HDA controller's power
state is now self-governed, rather than GPU-governed, whereas the GPU's
power state is no longer fully self-governed. (The HDA controller needs
to runtime suspend before the GPU can.)
It is thus crucial that runtime PM is always activated on the HDA
controller even if CONFIG_SND_HDA_POWER_SAVE_DEFAULT is set to 0 (which
is the default), lest the GPU stays awake. This is achieved by setting
the auto_runtime_pm flag on every codec and the AZX_DCAPS_PM_RUNTIME
flag on the HDA controller.
A side effect is that power consumption might be reduced if the GPU is
in use but the HDA controller is not, because the HDA controller is now
allowed to go to D3hot. Before, it was forced to stay in D0 as long as
the GPU was in use. (There is no reduction in power consumption on my
Nvidia GK107, but there might be on other chips.)
The code paths for legacy manual power control are adjusted such that
runtime PM is disabled during power off, thereby preventing the PM core
from resuming the HDA controller.
Note that the device link is not only added on vga_switcheroo capable
systems, but for *any* GPU with integrated HDA controller. The idea is
that the HDA controller streams audio via connectors located on the GPU,
so the GPU needs to be on for the HDA controller to do anything useful.
This commit implicitly fixes an unbalanced runtime PM ref upon unbind of
hda_intel.c: On ->probe, a runtime PM ref was previously released under
the condition "azx_has_pm_runtime(chip) || hda->use_vga_switcheroo", but
on ->remove a runtime PM ref was only acquired under the first of those
conditions. Thus, binding and unbinding the driver twice on a
vga_switcheroo capable system caused the runtime PM refcount to drop
below zero. The issue is resolved because the AZX_DCAPS_PM_RUNTIME flag
is now always set if use_vga_switcheroo is true.
For more information on device links please refer to:
https://www.kernel.org/doc/html/latest/driver-api/device_link.html
Documentation/driver-api/device_link.rst
Cc: Dave Airlie <airlied@redhat.com>
Cc: Ben Skeggs <bskeggs@redhat.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Takashi Iwai <tiwai@suse.de>
Reviewed-by: Peter Wu <peter@lekensteyn.nl>
Tested-by: Kai Heng Feng <kai.heng.feng@canonical.com> # AMD PowerXpress
Tested-by: Mike Lothian <mike@fireburn.co.uk> # AMD PowerXpress
Tested-by: Denis Lisov <dennis.lissov@gmail.com> # Nvidia Optimus
Tested-by: Peter Wu <peter@lekensteyn.nl> # Nvidia Optimus
Tested-by: Lukas Wunner <lukas@wunner.de> # MacBook Pro
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Link: https://patchwork.freedesktop.org/patch/msgid/51bd38360ff502a8c42b1ebf4405ee1d3f27118d.1520068884.git.lukas@wunner.de
2018-03-03 12:53:24 +03:00
/*
2019-06-06 12:22:24 +03:00
* Although not allowed by the spec , some multi - function devices have
* dependencies of one function ( consumer ) on another ( supplier ) . For the
* consumer to work in D0 , the supplier must also be in D0 . Create a
* device link from the consumer to the supplier to enforce this
* dependency . Runtime PM is allowed by default on the consumer to prevent
* it from permanently keeping the supplier awake .
vga_switcheroo: Use device link for HDA controller
Back in 2013, runtime PM for GPUs with integrated HDA controller was
introduced with commits 0d69704ae348 ("gpu/vga_switcheroo: add driver
control power feature. (v3)") and 246efa4a072f ("snd/hda: add runtime
suspend/resume on optimus support (v4)").
Briefly, the idea was that the HDA controller is forced on and off in
unison with the GPU.
The original code is mostly still in place even though it was never a
100% perfect solution: E.g. on access to the HDA controller, the GPU
is powered up via vga_switcheroo_runtime_resume_hdmi_audio() but there
are no provisions to keep it resumed until access to the HDA controller
has ceased: The GPU autosuspends after 5 seconds, rendering the HDA
controller inaccessible.
Additionally, a kludge is required when hda_intel.c probes: It has to
check whether the GPU is powered down (check_hdmi_disabled()) and defer
probing if so.
However in the meantime (in v4.10) the driver core has gained a feature
called device links which promises to solve such issues in a clean way:
It allows us to declare a dependency from the HDA controller (consumer)
to the GPU (supplier). The PM core then automagically ensures that the
GPU is runtime resumed as long as the HDA controller's ->probe hook is
executed and whenever the HDA controller is accessed.
By default, the HDA controller has a dependency on its parent, a PCIe
Root Port. Adding a device link creates another dependency on its
sibling:
PCIe Root Port
^ ^
| |
| |
HDA ===> GPU
The device link is not only used for runtime PM, it also guarantees that
on system sleep, the HDA controller suspends before the GPU and resumes
after the GPU, and on system shutdown the HDA controller's ->shutdown
hook is executed before the one of the GPU. It is a complete solution.
Using this functionality is as simple as calling device_link_add(),
which results in a dmesg entry like this:
pci 0000:01:00.1: Linked as a consumer to 0000:01:00.0
The code for the GPU-governed audio power management can thus be removed
(except where it's still needed for legacy manual power control).
The device link is added in a PCI quirk rather than in hda_intel.c.
It is therefore legal for the GPU to runtime suspend to D3cold even if
the HDA controller is not bound to a driver or if CONFIG_SND_HDA_INTEL
is not enabled, for accesses to the HDA controller will cause the GPU to
wake up regardless if they're occurring outside of hda_intel.c (think
config space readout via sysfs).
Contrary to the previous implementation, the HDA controller's power
state is now self-governed, rather than GPU-governed, whereas the GPU's
power state is no longer fully self-governed. (The HDA controller needs
to runtime suspend before the GPU can.)
It is thus crucial that runtime PM is always activated on the HDA
controller even if CONFIG_SND_HDA_POWER_SAVE_DEFAULT is set to 0 (which
is the default), lest the GPU stays awake. This is achieved by setting
the auto_runtime_pm flag on every codec and the AZX_DCAPS_PM_RUNTIME
flag on the HDA controller.
A side effect is that power consumption might be reduced if the GPU is
in use but the HDA controller is not, because the HDA controller is now
allowed to go to D3hot. Before, it was forced to stay in D0 as long as
the GPU was in use. (There is no reduction in power consumption on my
Nvidia GK107, but there might be on other chips.)
The code paths for legacy manual power control are adjusted such that
runtime PM is disabled during power off, thereby preventing the PM core
from resuming the HDA controller.
Note that the device link is not only added on vga_switcheroo capable
systems, but for *any* GPU with integrated HDA controller. The idea is
that the HDA controller streams audio via connectors located on the GPU,
so the GPU needs to be on for the HDA controller to do anything useful.
This commit implicitly fixes an unbalanced runtime PM ref upon unbind of
hda_intel.c: On ->probe, a runtime PM ref was previously released under
the condition "azx_has_pm_runtime(chip) || hda->use_vga_switcheroo", but
on ->remove a runtime PM ref was only acquired under the first of those
conditions. Thus, binding and unbinding the driver twice on a
vga_switcheroo capable system caused the runtime PM refcount to drop
below zero. The issue is resolved because the AZX_DCAPS_PM_RUNTIME flag
is now always set if use_vga_switcheroo is true.
For more information on device links please refer to:
https://www.kernel.org/doc/html/latest/driver-api/device_link.html
Documentation/driver-api/device_link.rst
Cc: Dave Airlie <airlied@redhat.com>
Cc: Ben Skeggs <bskeggs@redhat.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Takashi Iwai <tiwai@suse.de>
Reviewed-by: Peter Wu <peter@lekensteyn.nl>
Tested-by: Kai Heng Feng <kai.heng.feng@canonical.com> # AMD PowerXpress
Tested-by: Mike Lothian <mike@fireburn.co.uk> # AMD PowerXpress
Tested-by: Denis Lisov <dennis.lissov@gmail.com> # Nvidia Optimus
Tested-by: Peter Wu <peter@lekensteyn.nl> # Nvidia Optimus
Tested-by: Lukas Wunner <lukas@wunner.de> # MacBook Pro
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Link: https://patchwork.freedesktop.org/patch/msgid/51bd38360ff502a8c42b1ebf4405ee1d3f27118d.1520068884.git.lukas@wunner.de
2018-03-03 12:53:24 +03:00
*/
2019-06-06 12:22:24 +03:00
static void pci_create_device_link ( struct pci_dev * pdev , unsigned int consumer ,
unsigned int supplier , unsigned int class ,
unsigned int class_shift )
vga_switcheroo: Use device link for HDA controller
Back in 2013, runtime PM for GPUs with integrated HDA controller was
introduced with commits 0d69704ae348 ("gpu/vga_switcheroo: add driver
control power feature. (v3)") and 246efa4a072f ("snd/hda: add runtime
suspend/resume on optimus support (v4)").
Briefly, the idea was that the HDA controller is forced on and off in
unison with the GPU.
The original code is mostly still in place even though it was never a
100% perfect solution: E.g. on access to the HDA controller, the GPU
is powered up via vga_switcheroo_runtime_resume_hdmi_audio() but there
are no provisions to keep it resumed until access to the HDA controller
has ceased: The GPU autosuspends after 5 seconds, rendering the HDA
controller inaccessible.
Additionally, a kludge is required when hda_intel.c probes: It has to
check whether the GPU is powered down (check_hdmi_disabled()) and defer
probing if so.
However in the meantime (in v4.10) the driver core has gained a feature
called device links which promises to solve such issues in a clean way:
It allows us to declare a dependency from the HDA controller (consumer)
to the GPU (supplier). The PM core then automagically ensures that the
GPU is runtime resumed as long as the HDA controller's ->probe hook is
executed and whenever the HDA controller is accessed.
By default, the HDA controller has a dependency on its parent, a PCIe
Root Port. Adding a device link creates another dependency on its
sibling:
PCIe Root Port
^ ^
| |
| |
HDA ===> GPU
The device link is not only used for runtime PM, it also guarantees that
on system sleep, the HDA controller suspends before the GPU and resumes
after the GPU, and on system shutdown the HDA controller's ->shutdown
hook is executed before the one of the GPU. It is a complete solution.
Using this functionality is as simple as calling device_link_add(),
which results in a dmesg entry like this:
pci 0000:01:00.1: Linked as a consumer to 0000:01:00.0
The code for the GPU-governed audio power management can thus be removed
(except where it's still needed for legacy manual power control).
The device link is added in a PCI quirk rather than in hda_intel.c.
It is therefore legal for the GPU to runtime suspend to D3cold even if
the HDA controller is not bound to a driver or if CONFIG_SND_HDA_INTEL
is not enabled, for accesses to the HDA controller will cause the GPU to
wake up regardless if they're occurring outside of hda_intel.c (think
config space readout via sysfs).
Contrary to the previous implementation, the HDA controller's power
state is now self-governed, rather than GPU-governed, whereas the GPU's
power state is no longer fully self-governed. (The HDA controller needs
to runtime suspend before the GPU can.)
It is thus crucial that runtime PM is always activated on the HDA
controller even if CONFIG_SND_HDA_POWER_SAVE_DEFAULT is set to 0 (which
is the default), lest the GPU stays awake. This is achieved by setting
the auto_runtime_pm flag on every codec and the AZX_DCAPS_PM_RUNTIME
flag on the HDA controller.
A side effect is that power consumption might be reduced if the GPU is
in use but the HDA controller is not, because the HDA controller is now
allowed to go to D3hot. Before, it was forced to stay in D0 as long as
the GPU was in use. (There is no reduction in power consumption on my
Nvidia GK107, but there might be on other chips.)
The code paths for legacy manual power control are adjusted such that
runtime PM is disabled during power off, thereby preventing the PM core
from resuming the HDA controller.
Note that the device link is not only added on vga_switcheroo capable
systems, but for *any* GPU with integrated HDA controller. The idea is
that the HDA controller streams audio via connectors located on the GPU,
so the GPU needs to be on for the HDA controller to do anything useful.
This commit implicitly fixes an unbalanced runtime PM ref upon unbind of
hda_intel.c: On ->probe, a runtime PM ref was previously released under
the condition "azx_has_pm_runtime(chip) || hda->use_vga_switcheroo", but
on ->remove a runtime PM ref was only acquired under the first of those
conditions. Thus, binding and unbinding the driver twice on a
vga_switcheroo capable system caused the runtime PM refcount to drop
below zero. The issue is resolved because the AZX_DCAPS_PM_RUNTIME flag
is now always set if use_vga_switcheroo is true.
For more information on device links please refer to:
https://www.kernel.org/doc/html/latest/driver-api/device_link.html
Documentation/driver-api/device_link.rst
Cc: Dave Airlie <airlied@redhat.com>
Cc: Ben Skeggs <bskeggs@redhat.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Takashi Iwai <tiwai@suse.de>
Reviewed-by: Peter Wu <peter@lekensteyn.nl>
Tested-by: Kai Heng Feng <kai.heng.feng@canonical.com> # AMD PowerXpress
Tested-by: Mike Lothian <mike@fireburn.co.uk> # AMD PowerXpress
Tested-by: Denis Lisov <dennis.lissov@gmail.com> # Nvidia Optimus
Tested-by: Peter Wu <peter@lekensteyn.nl> # Nvidia Optimus
Tested-by: Lukas Wunner <lukas@wunner.de> # MacBook Pro
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Link: https://patchwork.freedesktop.org/patch/msgid/51bd38360ff502a8c42b1ebf4405ee1d3f27118d.1520068884.git.lukas@wunner.de
2018-03-03 12:53:24 +03:00
{
2019-06-06 12:22:24 +03:00
struct pci_dev * supplier_pdev ;
vga_switcheroo: Use device link for HDA controller
Back in 2013, runtime PM for GPUs with integrated HDA controller was
introduced with commits 0d69704ae348 ("gpu/vga_switcheroo: add driver
control power feature. (v3)") and 246efa4a072f ("snd/hda: add runtime
suspend/resume on optimus support (v4)").
Briefly, the idea was that the HDA controller is forced on and off in
unison with the GPU.
The original code is mostly still in place even though it was never a
100% perfect solution: E.g. on access to the HDA controller, the GPU
is powered up via vga_switcheroo_runtime_resume_hdmi_audio() but there
are no provisions to keep it resumed until access to the HDA controller
has ceased: The GPU autosuspends after 5 seconds, rendering the HDA
controller inaccessible.
Additionally, a kludge is required when hda_intel.c probes: It has to
check whether the GPU is powered down (check_hdmi_disabled()) and defer
probing if so.
However in the meantime (in v4.10) the driver core has gained a feature
called device links which promises to solve such issues in a clean way:
It allows us to declare a dependency from the HDA controller (consumer)
to the GPU (supplier). The PM core then automagically ensures that the
GPU is runtime resumed as long as the HDA controller's ->probe hook is
executed and whenever the HDA controller is accessed.
By default, the HDA controller has a dependency on its parent, a PCIe
Root Port. Adding a device link creates another dependency on its
sibling:
PCIe Root Port
^ ^
| |
| |
HDA ===> GPU
The device link is not only used for runtime PM, it also guarantees that
on system sleep, the HDA controller suspends before the GPU and resumes
after the GPU, and on system shutdown the HDA controller's ->shutdown
hook is executed before the one of the GPU. It is a complete solution.
Using this functionality is as simple as calling device_link_add(),
which results in a dmesg entry like this:
pci 0000:01:00.1: Linked as a consumer to 0000:01:00.0
The code for the GPU-governed audio power management can thus be removed
(except where it's still needed for legacy manual power control).
The device link is added in a PCI quirk rather than in hda_intel.c.
It is therefore legal for the GPU to runtime suspend to D3cold even if
the HDA controller is not bound to a driver or if CONFIG_SND_HDA_INTEL
is not enabled, for accesses to the HDA controller will cause the GPU to
wake up regardless if they're occurring outside of hda_intel.c (think
config space readout via sysfs).
Contrary to the previous implementation, the HDA controller's power
state is now self-governed, rather than GPU-governed, whereas the GPU's
power state is no longer fully self-governed. (The HDA controller needs
to runtime suspend before the GPU can.)
It is thus crucial that runtime PM is always activated on the HDA
controller even if CONFIG_SND_HDA_POWER_SAVE_DEFAULT is set to 0 (which
is the default), lest the GPU stays awake. This is achieved by setting
the auto_runtime_pm flag on every codec and the AZX_DCAPS_PM_RUNTIME
flag on the HDA controller.
A side effect is that power consumption might be reduced if the GPU is
in use but the HDA controller is not, because the HDA controller is now
allowed to go to D3hot. Before, it was forced to stay in D0 as long as
the GPU was in use. (There is no reduction in power consumption on my
Nvidia GK107, but there might be on other chips.)
The code paths for legacy manual power control are adjusted such that
runtime PM is disabled during power off, thereby preventing the PM core
from resuming the HDA controller.
Note that the device link is not only added on vga_switcheroo capable
systems, but for *any* GPU with integrated HDA controller. The idea is
that the HDA controller streams audio via connectors located on the GPU,
so the GPU needs to be on for the HDA controller to do anything useful.
This commit implicitly fixes an unbalanced runtime PM ref upon unbind of
hda_intel.c: On ->probe, a runtime PM ref was previously released under
the condition "azx_has_pm_runtime(chip) || hda->use_vga_switcheroo", but
on ->remove a runtime PM ref was only acquired under the first of those
conditions. Thus, binding and unbinding the driver twice on a
vga_switcheroo capable system caused the runtime PM refcount to drop
below zero. The issue is resolved because the AZX_DCAPS_PM_RUNTIME flag
is now always set if use_vga_switcheroo is true.
For more information on device links please refer to:
https://www.kernel.org/doc/html/latest/driver-api/device_link.html
Documentation/driver-api/device_link.rst
Cc: Dave Airlie <airlied@redhat.com>
Cc: Ben Skeggs <bskeggs@redhat.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Takashi Iwai <tiwai@suse.de>
Reviewed-by: Peter Wu <peter@lekensteyn.nl>
Tested-by: Kai Heng Feng <kai.heng.feng@canonical.com> # AMD PowerXpress
Tested-by: Mike Lothian <mike@fireburn.co.uk> # AMD PowerXpress
Tested-by: Denis Lisov <dennis.lissov@gmail.com> # Nvidia Optimus
Tested-by: Peter Wu <peter@lekensteyn.nl> # Nvidia Optimus
Tested-by: Lukas Wunner <lukas@wunner.de> # MacBook Pro
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Link: https://patchwork.freedesktop.org/patch/msgid/51bd38360ff502a8c42b1ebf4405ee1d3f27118d.1520068884.git.lukas@wunner.de
2018-03-03 12:53:24 +03:00
2019-06-06 12:22:24 +03:00
if ( PCI_FUNC ( pdev - > devfn ) ! = consumer )
vga_switcheroo: Use device link for HDA controller
Back in 2013, runtime PM for GPUs with integrated HDA controller was
introduced with commits 0d69704ae348 ("gpu/vga_switcheroo: add driver
control power feature. (v3)") and 246efa4a072f ("snd/hda: add runtime
suspend/resume on optimus support (v4)").
Briefly, the idea was that the HDA controller is forced on and off in
unison with the GPU.
The original code is mostly still in place even though it was never a
100% perfect solution: E.g. on access to the HDA controller, the GPU
is powered up via vga_switcheroo_runtime_resume_hdmi_audio() but there
are no provisions to keep it resumed until access to the HDA controller
has ceased: The GPU autosuspends after 5 seconds, rendering the HDA
controller inaccessible.
Additionally, a kludge is required when hda_intel.c probes: It has to
check whether the GPU is powered down (check_hdmi_disabled()) and defer
probing if so.
However in the meantime (in v4.10) the driver core has gained a feature
called device links which promises to solve such issues in a clean way:
It allows us to declare a dependency from the HDA controller (consumer)
to the GPU (supplier). The PM core then automagically ensures that the
GPU is runtime resumed as long as the HDA controller's ->probe hook is
executed and whenever the HDA controller is accessed.
By default, the HDA controller has a dependency on its parent, a PCIe
Root Port. Adding a device link creates another dependency on its
sibling:
PCIe Root Port
^ ^
| |
| |
HDA ===> GPU
The device link is not only used for runtime PM, it also guarantees that
on system sleep, the HDA controller suspends before the GPU and resumes
after the GPU, and on system shutdown the HDA controller's ->shutdown
hook is executed before the one of the GPU. It is a complete solution.
Using this functionality is as simple as calling device_link_add(),
which results in a dmesg entry like this:
pci 0000:01:00.1: Linked as a consumer to 0000:01:00.0
The code for the GPU-governed audio power management can thus be removed
(except where it's still needed for legacy manual power control).
The device link is added in a PCI quirk rather than in hda_intel.c.
It is therefore legal for the GPU to runtime suspend to D3cold even if
the HDA controller is not bound to a driver or if CONFIG_SND_HDA_INTEL
is not enabled, for accesses to the HDA controller will cause the GPU to
wake up regardless if they're occurring outside of hda_intel.c (think
config space readout via sysfs).
Contrary to the previous implementation, the HDA controller's power
state is now self-governed, rather than GPU-governed, whereas the GPU's
power state is no longer fully self-governed. (The HDA controller needs
to runtime suspend before the GPU can.)
It is thus crucial that runtime PM is always activated on the HDA
controller even if CONFIG_SND_HDA_POWER_SAVE_DEFAULT is set to 0 (which
is the default), lest the GPU stays awake. This is achieved by setting
the auto_runtime_pm flag on every codec and the AZX_DCAPS_PM_RUNTIME
flag on the HDA controller.
A side effect is that power consumption might be reduced if the GPU is
in use but the HDA controller is not, because the HDA controller is now
allowed to go to D3hot. Before, it was forced to stay in D0 as long as
the GPU was in use. (There is no reduction in power consumption on my
Nvidia GK107, but there might be on other chips.)
The code paths for legacy manual power control are adjusted such that
runtime PM is disabled during power off, thereby preventing the PM core
from resuming the HDA controller.
Note that the device link is not only added on vga_switcheroo capable
systems, but for *any* GPU with integrated HDA controller. The idea is
that the HDA controller streams audio via connectors located on the GPU,
so the GPU needs to be on for the HDA controller to do anything useful.
This commit implicitly fixes an unbalanced runtime PM ref upon unbind of
hda_intel.c: On ->probe, a runtime PM ref was previously released under
the condition "azx_has_pm_runtime(chip) || hda->use_vga_switcheroo", but
on ->remove a runtime PM ref was only acquired under the first of those
conditions. Thus, binding and unbinding the driver twice on a
vga_switcheroo capable system caused the runtime PM refcount to drop
below zero. The issue is resolved because the AZX_DCAPS_PM_RUNTIME flag
is now always set if use_vga_switcheroo is true.
For more information on device links please refer to:
https://www.kernel.org/doc/html/latest/driver-api/device_link.html
Documentation/driver-api/device_link.rst
Cc: Dave Airlie <airlied@redhat.com>
Cc: Ben Skeggs <bskeggs@redhat.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Takashi Iwai <tiwai@suse.de>
Reviewed-by: Peter Wu <peter@lekensteyn.nl>
Tested-by: Kai Heng Feng <kai.heng.feng@canonical.com> # AMD PowerXpress
Tested-by: Mike Lothian <mike@fireburn.co.uk> # AMD PowerXpress
Tested-by: Denis Lisov <dennis.lissov@gmail.com> # Nvidia Optimus
Tested-by: Peter Wu <peter@lekensteyn.nl> # Nvidia Optimus
Tested-by: Lukas Wunner <lukas@wunner.de> # MacBook Pro
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Link: https://patchwork.freedesktop.org/patch/msgid/51bd38360ff502a8c42b1ebf4405ee1d3f27118d.1520068884.git.lukas@wunner.de
2018-03-03 12:53:24 +03:00
return ;
2019-06-06 12:22:24 +03:00
supplier_pdev = pci_get_domain_bus_and_slot ( pci_domain_nr ( pdev - > bus ) ,
pdev - > bus - > number ,
PCI_DEVFN ( PCI_SLOT ( pdev - > devfn ) , supplier ) ) ;
if ( ! supplier_pdev | | ( supplier_pdev - > class > > class_shift ) ! = class ) {
pci_dev_put ( supplier_pdev ) ;
vga_switcheroo: Use device link for HDA controller
Back in 2013, runtime PM for GPUs with integrated HDA controller was
introduced with commits 0d69704ae348 ("gpu/vga_switcheroo: add driver
control power feature. (v3)") and 246efa4a072f ("snd/hda: add runtime
suspend/resume on optimus support (v4)").
Briefly, the idea was that the HDA controller is forced on and off in
unison with the GPU.
The original code is mostly still in place even though it was never a
100% perfect solution: E.g. on access to the HDA controller, the GPU
is powered up via vga_switcheroo_runtime_resume_hdmi_audio() but there
are no provisions to keep it resumed until access to the HDA controller
has ceased: The GPU autosuspends after 5 seconds, rendering the HDA
controller inaccessible.
Additionally, a kludge is required when hda_intel.c probes: It has to
check whether the GPU is powered down (check_hdmi_disabled()) and defer
probing if so.
However in the meantime (in v4.10) the driver core has gained a feature
called device links which promises to solve such issues in a clean way:
It allows us to declare a dependency from the HDA controller (consumer)
to the GPU (supplier). The PM core then automagically ensures that the
GPU is runtime resumed as long as the HDA controller's ->probe hook is
executed and whenever the HDA controller is accessed.
By default, the HDA controller has a dependency on its parent, a PCIe
Root Port. Adding a device link creates another dependency on its
sibling:
PCIe Root Port
^ ^
| |
| |
HDA ===> GPU
The device link is not only used for runtime PM, it also guarantees that
on system sleep, the HDA controller suspends before the GPU and resumes
after the GPU, and on system shutdown the HDA controller's ->shutdown
hook is executed before the one of the GPU. It is a complete solution.
Using this functionality is as simple as calling device_link_add(),
which results in a dmesg entry like this:
pci 0000:01:00.1: Linked as a consumer to 0000:01:00.0
The code for the GPU-governed audio power management can thus be removed
(except where it's still needed for legacy manual power control).
The device link is added in a PCI quirk rather than in hda_intel.c.
It is therefore legal for the GPU to runtime suspend to D3cold even if
the HDA controller is not bound to a driver or if CONFIG_SND_HDA_INTEL
is not enabled, for accesses to the HDA controller will cause the GPU to
wake up regardless if they're occurring outside of hda_intel.c (think
config space readout via sysfs).
Contrary to the previous implementation, the HDA controller's power
state is now self-governed, rather than GPU-governed, whereas the GPU's
power state is no longer fully self-governed. (The HDA controller needs
to runtime suspend before the GPU can.)
It is thus crucial that runtime PM is always activated on the HDA
controller even if CONFIG_SND_HDA_POWER_SAVE_DEFAULT is set to 0 (which
is the default), lest the GPU stays awake. This is achieved by setting
the auto_runtime_pm flag on every codec and the AZX_DCAPS_PM_RUNTIME
flag on the HDA controller.
A side effect is that power consumption might be reduced if the GPU is
in use but the HDA controller is not, because the HDA controller is now
allowed to go to D3hot. Before, it was forced to stay in D0 as long as
the GPU was in use. (There is no reduction in power consumption on my
Nvidia GK107, but there might be on other chips.)
The code paths for legacy manual power control are adjusted such that
runtime PM is disabled during power off, thereby preventing the PM core
from resuming the HDA controller.
Note that the device link is not only added on vga_switcheroo capable
systems, but for *any* GPU with integrated HDA controller. The idea is
that the HDA controller streams audio via connectors located on the GPU,
so the GPU needs to be on for the HDA controller to do anything useful.
This commit implicitly fixes an unbalanced runtime PM ref upon unbind of
hda_intel.c: On ->probe, a runtime PM ref was previously released under
the condition "azx_has_pm_runtime(chip) || hda->use_vga_switcheroo", but
on ->remove a runtime PM ref was only acquired under the first of those
conditions. Thus, binding and unbinding the driver twice on a
vga_switcheroo capable system caused the runtime PM refcount to drop
below zero. The issue is resolved because the AZX_DCAPS_PM_RUNTIME flag
is now always set if use_vga_switcheroo is true.
For more information on device links please refer to:
https://www.kernel.org/doc/html/latest/driver-api/device_link.html
Documentation/driver-api/device_link.rst
Cc: Dave Airlie <airlied@redhat.com>
Cc: Ben Skeggs <bskeggs@redhat.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Takashi Iwai <tiwai@suse.de>
Reviewed-by: Peter Wu <peter@lekensteyn.nl>
Tested-by: Kai Heng Feng <kai.heng.feng@canonical.com> # AMD PowerXpress
Tested-by: Mike Lothian <mike@fireburn.co.uk> # AMD PowerXpress
Tested-by: Denis Lisov <dennis.lissov@gmail.com> # Nvidia Optimus
Tested-by: Peter Wu <peter@lekensteyn.nl> # Nvidia Optimus
Tested-by: Lukas Wunner <lukas@wunner.de> # MacBook Pro
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Link: https://patchwork.freedesktop.org/patch/msgid/51bd38360ff502a8c42b1ebf4405ee1d3f27118d.1520068884.git.lukas@wunner.de
2018-03-03 12:53:24 +03:00
return ;
}
2019-06-06 12:22:24 +03:00
if ( device_link_add ( & pdev - > dev , & supplier_pdev - > dev ,
DL_FLAG_STATELESS | DL_FLAG_PM_RUNTIME ) )
pci_info ( pdev , " D0 power state depends on %s \n " ,
pci_name ( supplier_pdev ) ) ;
else
pci_err ( pdev , " Cannot enforce power dependency on %s \n " ,
pci_name ( supplier_pdev ) ) ;
pm_runtime_allow ( & pdev - > dev ) ;
pci_dev_put ( supplier_pdev ) ;
}
vga_switcheroo: Use device link for HDA controller
Back in 2013, runtime PM for GPUs with integrated HDA controller was
introduced with commits 0d69704ae348 ("gpu/vga_switcheroo: add driver
control power feature. (v3)") and 246efa4a072f ("snd/hda: add runtime
suspend/resume on optimus support (v4)").
Briefly, the idea was that the HDA controller is forced on and off in
unison with the GPU.
The original code is mostly still in place even though it was never a
100% perfect solution: E.g. on access to the HDA controller, the GPU
is powered up via vga_switcheroo_runtime_resume_hdmi_audio() but there
are no provisions to keep it resumed until access to the HDA controller
has ceased: The GPU autosuspends after 5 seconds, rendering the HDA
controller inaccessible.
Additionally, a kludge is required when hda_intel.c probes: It has to
check whether the GPU is powered down (check_hdmi_disabled()) and defer
probing if so.
However in the meantime (in v4.10) the driver core has gained a feature
called device links which promises to solve such issues in a clean way:
It allows us to declare a dependency from the HDA controller (consumer)
to the GPU (supplier). The PM core then automagically ensures that the
GPU is runtime resumed as long as the HDA controller's ->probe hook is
executed and whenever the HDA controller is accessed.
By default, the HDA controller has a dependency on its parent, a PCIe
Root Port. Adding a device link creates another dependency on its
sibling:
PCIe Root Port
^ ^
| |
| |
HDA ===> GPU
The device link is not only used for runtime PM, it also guarantees that
on system sleep, the HDA controller suspends before the GPU and resumes
after the GPU, and on system shutdown the HDA controller's ->shutdown
hook is executed before the one of the GPU. It is a complete solution.
Using this functionality is as simple as calling device_link_add(),
which results in a dmesg entry like this:
pci 0000:01:00.1: Linked as a consumer to 0000:01:00.0
The code for the GPU-governed audio power management can thus be removed
(except where it's still needed for legacy manual power control).
The device link is added in a PCI quirk rather than in hda_intel.c.
It is therefore legal for the GPU to runtime suspend to D3cold even if
the HDA controller is not bound to a driver or if CONFIG_SND_HDA_INTEL
is not enabled, for accesses to the HDA controller will cause the GPU to
wake up regardless if they're occurring outside of hda_intel.c (think
config space readout via sysfs).
Contrary to the previous implementation, the HDA controller's power
state is now self-governed, rather than GPU-governed, whereas the GPU's
power state is no longer fully self-governed. (The HDA controller needs
to runtime suspend before the GPU can.)
It is thus crucial that runtime PM is always activated on the HDA
controller even if CONFIG_SND_HDA_POWER_SAVE_DEFAULT is set to 0 (which
is the default), lest the GPU stays awake. This is achieved by setting
the auto_runtime_pm flag on every codec and the AZX_DCAPS_PM_RUNTIME
flag on the HDA controller.
A side effect is that power consumption might be reduced if the GPU is
in use but the HDA controller is not, because the HDA controller is now
allowed to go to D3hot. Before, it was forced to stay in D0 as long as
the GPU was in use. (There is no reduction in power consumption on my
Nvidia GK107, but there might be on other chips.)
The code paths for legacy manual power control are adjusted such that
runtime PM is disabled during power off, thereby preventing the PM core
from resuming the HDA controller.
Note that the device link is not only added on vga_switcheroo capable
systems, but for *any* GPU with integrated HDA controller. The idea is
that the HDA controller streams audio via connectors located on the GPU,
so the GPU needs to be on for the HDA controller to do anything useful.
This commit implicitly fixes an unbalanced runtime PM ref upon unbind of
hda_intel.c: On ->probe, a runtime PM ref was previously released under
the condition "azx_has_pm_runtime(chip) || hda->use_vga_switcheroo", but
on ->remove a runtime PM ref was only acquired under the first of those
conditions. Thus, binding and unbinding the driver twice on a
vga_switcheroo capable system caused the runtime PM refcount to drop
below zero. The issue is resolved because the AZX_DCAPS_PM_RUNTIME flag
is now always set if use_vga_switcheroo is true.
For more information on device links please refer to:
https://www.kernel.org/doc/html/latest/driver-api/device_link.html
Documentation/driver-api/device_link.rst
Cc: Dave Airlie <airlied@redhat.com>
Cc: Ben Skeggs <bskeggs@redhat.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Takashi Iwai <tiwai@suse.de>
Reviewed-by: Peter Wu <peter@lekensteyn.nl>
Tested-by: Kai Heng Feng <kai.heng.feng@canonical.com> # AMD PowerXpress
Tested-by: Mike Lothian <mike@fireburn.co.uk> # AMD PowerXpress
Tested-by: Denis Lisov <dennis.lissov@gmail.com> # Nvidia Optimus
Tested-by: Peter Wu <peter@lekensteyn.nl> # Nvidia Optimus
Tested-by: Lukas Wunner <lukas@wunner.de> # MacBook Pro
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Link: https://patchwork.freedesktop.org/patch/msgid/51bd38360ff502a8c42b1ebf4405ee1d3f27118d.1520068884.git.lukas@wunner.de
2018-03-03 12:53:24 +03:00
2019-06-06 12:22:24 +03:00
/*
* Create device link for GPUs with integrated HDA controller for streaming
* audio to attached displays .
*/
static void quirk_gpu_hda ( struct pci_dev * hda )
{
pci_create_device_link ( hda , 1 , 0 , PCI_BASE_CLASS_DISPLAY , 16 ) ;
vga_switcheroo: Use device link for HDA controller
Back in 2013, runtime PM for GPUs with integrated HDA controller was
introduced with commits 0d69704ae348 ("gpu/vga_switcheroo: add driver
control power feature. (v3)") and 246efa4a072f ("snd/hda: add runtime
suspend/resume on optimus support (v4)").
Briefly, the idea was that the HDA controller is forced on and off in
unison with the GPU.
The original code is mostly still in place even though it was never a
100% perfect solution: E.g. on access to the HDA controller, the GPU
is powered up via vga_switcheroo_runtime_resume_hdmi_audio() but there
are no provisions to keep it resumed until access to the HDA controller
has ceased: The GPU autosuspends after 5 seconds, rendering the HDA
controller inaccessible.
Additionally, a kludge is required when hda_intel.c probes: It has to
check whether the GPU is powered down (check_hdmi_disabled()) and defer
probing if so.
However in the meantime (in v4.10) the driver core has gained a feature
called device links which promises to solve such issues in a clean way:
It allows us to declare a dependency from the HDA controller (consumer)
to the GPU (supplier). The PM core then automagically ensures that the
GPU is runtime resumed as long as the HDA controller's ->probe hook is
executed and whenever the HDA controller is accessed.
By default, the HDA controller has a dependency on its parent, a PCIe
Root Port. Adding a device link creates another dependency on its
sibling:
PCIe Root Port
^ ^
| |
| |
HDA ===> GPU
The device link is not only used for runtime PM, it also guarantees that
on system sleep, the HDA controller suspends before the GPU and resumes
after the GPU, and on system shutdown the HDA controller's ->shutdown
hook is executed before the one of the GPU. It is a complete solution.
Using this functionality is as simple as calling device_link_add(),
which results in a dmesg entry like this:
pci 0000:01:00.1: Linked as a consumer to 0000:01:00.0
The code for the GPU-governed audio power management can thus be removed
(except where it's still needed for legacy manual power control).
The device link is added in a PCI quirk rather than in hda_intel.c.
It is therefore legal for the GPU to runtime suspend to D3cold even if
the HDA controller is not bound to a driver or if CONFIG_SND_HDA_INTEL
is not enabled, for accesses to the HDA controller will cause the GPU to
wake up regardless if they're occurring outside of hda_intel.c (think
config space readout via sysfs).
Contrary to the previous implementation, the HDA controller's power
state is now self-governed, rather than GPU-governed, whereas the GPU's
power state is no longer fully self-governed. (The HDA controller needs
to runtime suspend before the GPU can.)
It is thus crucial that runtime PM is always activated on the HDA
controller even if CONFIG_SND_HDA_POWER_SAVE_DEFAULT is set to 0 (which
is the default), lest the GPU stays awake. This is achieved by setting
the auto_runtime_pm flag on every codec and the AZX_DCAPS_PM_RUNTIME
flag on the HDA controller.
A side effect is that power consumption might be reduced if the GPU is
in use but the HDA controller is not, because the HDA controller is now
allowed to go to D3hot. Before, it was forced to stay in D0 as long as
the GPU was in use. (There is no reduction in power consumption on my
Nvidia GK107, but there might be on other chips.)
The code paths for legacy manual power control are adjusted such that
runtime PM is disabled during power off, thereby preventing the PM core
from resuming the HDA controller.
Note that the device link is not only added on vga_switcheroo capable
systems, but for *any* GPU with integrated HDA controller. The idea is
that the HDA controller streams audio via connectors located on the GPU,
so the GPU needs to be on for the HDA controller to do anything useful.
This commit implicitly fixes an unbalanced runtime PM ref upon unbind of
hda_intel.c: On ->probe, a runtime PM ref was previously released under
the condition "azx_has_pm_runtime(chip) || hda->use_vga_switcheroo", but
on ->remove a runtime PM ref was only acquired under the first of those
conditions. Thus, binding and unbinding the driver twice on a
vga_switcheroo capable system caused the runtime PM refcount to drop
below zero. The issue is resolved because the AZX_DCAPS_PM_RUNTIME flag
is now always set if use_vga_switcheroo is true.
For more information on device links please refer to:
https://www.kernel.org/doc/html/latest/driver-api/device_link.html
Documentation/driver-api/device_link.rst
Cc: Dave Airlie <airlied@redhat.com>
Cc: Ben Skeggs <bskeggs@redhat.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Takashi Iwai <tiwai@suse.de>
Reviewed-by: Peter Wu <peter@lekensteyn.nl>
Tested-by: Kai Heng Feng <kai.heng.feng@canonical.com> # AMD PowerXpress
Tested-by: Mike Lothian <mike@fireburn.co.uk> # AMD PowerXpress
Tested-by: Denis Lisov <dennis.lissov@gmail.com> # Nvidia Optimus
Tested-by: Peter Wu <peter@lekensteyn.nl> # Nvidia Optimus
Tested-by: Lukas Wunner <lukas@wunner.de> # MacBook Pro
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Link: https://patchwork.freedesktop.org/patch/msgid/51bd38360ff502a8c42b1ebf4405ee1d3f27118d.1520068884.git.lukas@wunner.de
2018-03-03 12:53:24 +03:00
}
DECLARE_PCI_FIXUP_CLASS_FINAL ( PCI_VENDOR_ID_ATI , PCI_ANY_ID ,
PCI_CLASS_MULTIMEDIA_HD_AUDIO , 8 , quirk_gpu_hda ) ;
DECLARE_PCI_FIXUP_CLASS_FINAL ( PCI_VENDOR_ID_AMD , PCI_ANY_ID ,
PCI_CLASS_MULTIMEDIA_HD_AUDIO , 8 , quirk_gpu_hda ) ;
DECLARE_PCI_FIXUP_CLASS_FINAL ( PCI_VENDOR_ID_NVIDIA , PCI_ANY_ID ,
PCI_CLASS_MULTIMEDIA_HD_AUDIO , 8 , quirk_gpu_hda ) ;
PCI: Workaround IDT switch ACS Source Validation erratum
Some IDT switches incorrectly flag an ACS Source Validation error on
completions for config read requests even though PCIe r4.0, sec 6.12.1.1,
says that completions are never affected by ACS Source Validation. Here's
the text of IDT 89H32H8G3-YC, erratum #36:
Item #36 - Downstream port applies ACS Source Validation to Completions
Section 6.12.1.1 of the PCI Express Base Specification 3.1 states that
completions are never affected by ACS Source Validation. However,
completions received by a downstream port of the PCIe switch from a
device that has not yet captured a PCIe bus number are incorrectly
dropped by ACS Source Validation by the switch downstream port.
Workaround: Issue a CfgWr1 to the downstream device before issuing the
first CfgRd1 to the device. This allows the downstream device to capture
its bus number; ACS Source Validation no longer stops completions from
being forwarded by the downstream port. It has been observed that
Microsoft Windows implements this workaround already; however, some
versions of Linux and other operating systems may not.
When doing the first config read to probe for a device, if the device is
behind an IDT switch with this erratum:
1. Disable ACS Source Validation if enabled
2. Wait for device to become ready to accept config accesses (by using
the Config Request Retry Status mechanism)
3. Do a config write to the endpoint
4. Enable ACS Source Validation (if it was enabled to begin with)
The workaround suggested by IDT is basically only step 3, but we don't know
when the device is ready to accept config requests. That means we need to
do config reads until we receive a non-Config Request Retry Status, which
means we need to disable ACS SV temporarily.
Signed-off-by: James Puthukattukaran <james.puthukattukaran@oracle.com>
[bhelgaas: changelog, clean up whitespace, fold in unused variable fix
from Anders Roxell <anders.roxell@linaro.org>]
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Alex Williamson <alex.williamson@redhat.com>
2018-07-09 18:31:25 +03:00
PCI: Add NVIDIA GPU multi-function power dependencies
The NVIDIA Turing GPU is a multi-function PCI device with the following
functions:
- Function 0: VGA display controller
- Function 1: Audio controller
- Function 2: USB xHCI Host controller
- Function 3: USB Type-C UCSI controller
Function 0 is tightly coupled with other functions in the hardware. When
function 0 is in D3, it gates power for hardware blocks used by other
functions, which means those functions only work when function 0 is in D0.
If any of these functions (1/2/3) are in D0, then function 0 should also be
in D0.
Commit 07f4f97d7b4b ("vga_switcheroo: Use device link for HDA controller")
already creates a device link to show the dependency of function 1 on
function 0 of this GPU. Create additional device links to express the
dependencies of functions 2 and 3 on function 0. This means function 0
will be in D0 if any other function is in D0.
[bhelgaas: I think the PCI spec expectation is that functions can be
power-managed independently, so I don't think this device is technically
compliant. For example, the PCIe r5.0 spec, sec 1.4, says "the PCI/PCIe
hardware/software model includes architectural constructs necessary to
discover, configure, and use a Function, without needing Function-specific
knowledge" and sec 5.1 says "D states are associated with a particular
Function" and "PM provides ... a mechanism to identify power management
capabilities of a given Function [and] the ability to transition a Function
into a certain power management state."]
Link: https://lore.kernel.org/lkml/20190606092225.17960-3-abhsahu@nvidia.com
Signed-off-by: Abhishek Sahu <abhsahu@nvidia.com>
[bhelgaas: commit log]
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2019-06-06 12:22:25 +03:00
/*
* Create device link for NVIDIA GPU with integrated USB xHCI Host
* controller to VGA .
*/
static void quirk_gpu_usb ( struct pci_dev * usb )
{
pci_create_device_link ( usb , 2 , 0 , PCI_BASE_CLASS_DISPLAY , 16 ) ;
}
DECLARE_PCI_FIXUP_CLASS_FINAL ( PCI_VENDOR_ID_NVIDIA , PCI_ANY_ID ,
PCI_CLASS_SERIAL_USB , 8 , quirk_gpu_usb ) ;
/*
* Create device link for NVIDIA GPU with integrated Type - C UCSI controller
* to VGA . Currently there is no class code defined for UCSI device over PCI
* so using UNKNOWN class for now and it will be updated when UCSI
* over PCI gets a class code .
*/
# define PCI_CLASS_SERIAL_UNKNOWN 0x0c80
static void quirk_gpu_usb_typec_ucsi ( struct pci_dev * ucsi )
{
pci_create_device_link ( ucsi , 3 , 0 , PCI_BASE_CLASS_DISPLAY , 16 ) ;
}
DECLARE_PCI_FIXUP_CLASS_FINAL ( PCI_VENDOR_ID_NVIDIA , PCI_ANY_ID ,
PCI_CLASS_SERIAL_UNKNOWN , 8 ,
quirk_gpu_usb_typec_ucsi ) ;
2019-07-08 08:17:44 +03:00
/*
* Enable the NVIDIA GPU integrated HDA controller if the BIOS left it
* disabled . https : //devtalk.nvidia.com/default/topic/1024022
*/
static void quirk_nvidia_hda ( struct pci_dev * gpu )
{
u8 hdr_type ;
u32 val ;
/* There was no integrated HDA controller before MCP89 */
if ( gpu - > device < PCI_DEVICE_ID_NVIDIA_GEFORCE_320M )
return ;
/* Bit 25 at offset 0x488 enables the HDA controller */
pci_read_config_dword ( gpu , 0x488 , & val ) ;
if ( val & BIT ( 25 ) )
return ;
pci_info ( gpu , " Enabling HDA controller \n " ) ;
pci_write_config_dword ( gpu , 0x488 , val | BIT ( 25 ) ) ;
/* The GPU becomes a multi-function device when the HDA is enabled */
pci_read_config_byte ( gpu , PCI_HEADER_TYPE , & hdr_type ) ;
gpu - > multifunction = ! ! ( hdr_type & 0x80 ) ;
}
DECLARE_PCI_FIXUP_CLASS_HEADER ( PCI_VENDOR_ID_NVIDIA , PCI_ANY_ID ,
PCI_BASE_CLASS_DISPLAY , 16 , quirk_nvidia_hda ) ;
DECLARE_PCI_FIXUP_CLASS_RESUME_EARLY ( PCI_VENDOR_ID_NVIDIA , PCI_ANY_ID ,
PCI_BASE_CLASS_DISPLAY , 16 , quirk_nvidia_hda ) ;
PCI: Workaround IDT switch ACS Source Validation erratum
Some IDT switches incorrectly flag an ACS Source Validation error on
completions for config read requests even though PCIe r4.0, sec 6.12.1.1,
says that completions are never affected by ACS Source Validation. Here's
the text of IDT 89H32H8G3-YC, erratum #36:
Item #36 - Downstream port applies ACS Source Validation to Completions
Section 6.12.1.1 of the PCI Express Base Specification 3.1 states that
completions are never affected by ACS Source Validation. However,
completions received by a downstream port of the PCIe switch from a
device that has not yet captured a PCIe bus number are incorrectly
dropped by ACS Source Validation by the switch downstream port.
Workaround: Issue a CfgWr1 to the downstream device before issuing the
first CfgRd1 to the device. This allows the downstream device to capture
its bus number; ACS Source Validation no longer stops completions from
being forwarded by the downstream port. It has been observed that
Microsoft Windows implements this workaround already; however, some
versions of Linux and other operating systems may not.
When doing the first config read to probe for a device, if the device is
behind an IDT switch with this erratum:
1. Disable ACS Source Validation if enabled
2. Wait for device to become ready to accept config accesses (by using
the Config Request Retry Status mechanism)
3. Do a config write to the endpoint
4. Enable ACS Source Validation (if it was enabled to begin with)
The workaround suggested by IDT is basically only step 3, but we don't know
when the device is ready to accept config requests. That means we need to
do config reads until we receive a non-Config Request Retry Status, which
means we need to disable ACS SV temporarily.
Signed-off-by: James Puthukattukaran <james.puthukattukaran@oracle.com>
[bhelgaas: changelog, clean up whitespace, fold in unused variable fix
from Anders Roxell <anders.roxell@linaro.org>]
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Alex Williamson <alex.williamson@redhat.com>
2018-07-09 18:31:25 +03:00
/*
* Some IDT switches incorrectly flag an ACS Source Validation error on
* completions for config read requests even though PCIe r4 .0 , sec
* 6.12 .1 .1 , says that completions are never affected by ACS Source
* Validation . Here ' s the text of IDT 89 H32H8G3 - YC , erratum # 36 :
*
* Item # 36 - Downstream port applies ACS Source Validation to Completions
* Section 6.12 .1 .1 of the PCI Express Base Specification 3.1 states that
* completions are never affected by ACS Source Validation . However ,
* completions received by a downstream port of the PCIe switch from a
* device that has not yet captured a PCIe bus number are incorrectly
* dropped by ACS Source Validation by the switch downstream port .
*
* The workaround suggested by IDT is to issue a config write to the
* downstream device before issuing the first config read . This allows the
* downstream device to capture its bus and device numbers ( see PCIe r4 .0 ,
* sec 2.2 .9 ) , thus avoiding the ACS error on the completion .
*
* However , we don ' t know when the device is ready to accept the config
* write , so we do config reads until we receive a non - Config Request Retry
* Status , then do the config write .
*
* To avoid hitting the erratum when doing the config reads , we disable ACS
* SV around this process .
*/
int pci_idt_bus_quirk ( struct pci_bus * bus , int devfn , u32 * l , int timeout )
{
int pos ;
u16 ctrl = 0 ;
bool found ;
struct pci_dev * bridge = bus - > self ;
2020-07-08 01:46:02 +03:00
pos = bridge - > acs_cap ;
PCI: Workaround IDT switch ACS Source Validation erratum
Some IDT switches incorrectly flag an ACS Source Validation error on
completions for config read requests even though PCIe r4.0, sec 6.12.1.1,
says that completions are never affected by ACS Source Validation. Here's
the text of IDT 89H32H8G3-YC, erratum #36:
Item #36 - Downstream port applies ACS Source Validation to Completions
Section 6.12.1.1 of the PCI Express Base Specification 3.1 states that
completions are never affected by ACS Source Validation. However,
completions received by a downstream port of the PCIe switch from a
device that has not yet captured a PCIe bus number are incorrectly
dropped by ACS Source Validation by the switch downstream port.
Workaround: Issue a CfgWr1 to the downstream device before issuing the
first CfgRd1 to the device. This allows the downstream device to capture
its bus number; ACS Source Validation no longer stops completions from
being forwarded by the downstream port. It has been observed that
Microsoft Windows implements this workaround already; however, some
versions of Linux and other operating systems may not.
When doing the first config read to probe for a device, if the device is
behind an IDT switch with this erratum:
1. Disable ACS Source Validation if enabled
2. Wait for device to become ready to accept config accesses (by using
the Config Request Retry Status mechanism)
3. Do a config write to the endpoint
4. Enable ACS Source Validation (if it was enabled to begin with)
The workaround suggested by IDT is basically only step 3, but we don't know
when the device is ready to accept config requests. That means we need to
do config reads until we receive a non-Config Request Retry Status, which
means we need to disable ACS SV temporarily.
Signed-off-by: James Puthukattukaran <james.puthukattukaran@oracle.com>
[bhelgaas: changelog, clean up whitespace, fold in unused variable fix
from Anders Roxell <anders.roxell@linaro.org>]
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Alex Williamson <alex.williamson@redhat.com>
2018-07-09 18:31:25 +03:00
/* Disable ACS SV before initial config reads */
if ( pos ) {
pci_read_config_word ( bridge , pos + PCI_ACS_CTRL , & ctrl ) ;
if ( ctrl & PCI_ACS_SV )
pci_write_config_word ( bridge , pos + PCI_ACS_CTRL ,
ctrl & ~ PCI_ACS_SV ) ;
}
found = pci_bus_generic_read_dev_vendor_id ( bus , devfn , l , timeout ) ;
/* Write Vendor ID (read-only) so the endpoint latches its bus/dev */
if ( found )
pci_bus_write_config_word ( bus , devfn , PCI_VENDOR_ID , 0 ) ;
/* Re-enable ACS_SV if it was previously enabled */
if ( ctrl & PCI_ACS_SV )
pci_write_config_word ( bridge , pos + PCI_ACS_CTRL , ctrl ) ;
return found ;
}
2018-08-15 22:59:03 +03:00
PCI: Add DMA alias quirk for Microsemi Switchtec NTB
Add a quirk for the Microsemi Switchtec parts to allow DMA access via
non-transparent bridging to work when the IOMMU is turned on.
This exclusively addresses the ability of a remote NT endpoint to perform
DMA accesses through the locally enumerated NT endpoint. Other aspects of
the Switchtec NTB functionality, such as interrupts for doorbells and
messages are independent of this quirk, and will work whether the IOMMU is
on or off.
When a requestor on one NT endpoint accesses memory on another NT endpoint,
it does this via a devfn proxy ID. Proxy IDs are statically assigned to
each NT endpoint by the NTB hardware as part of the release-from-reset
sequence prior to PCI enumeration. These proxy IDs cannot be modified
dynamically, and are not visible to the host during enumeration.
When the Switchtec NTB driver loads it will map local requestor IDs, such
as the root complex and transparent bridge DMA engines, to proxy IDs by
populating those requestor IDs in hardware mapping table table entries.
This establishes a fixed relationship between a requestor ID and a proxy
ID.
When a peer on a remote NT endpoint performs an access within a particular
translation window in it's NT endpoint BAR address space, that access is
translated to a DMA request on the local endpoint's bus. As part of the
translation process, the original requestor ID has its devfn replaced with
the proxy ID, and the bus portion of the BDF is replaced with the bus of
the local NT endpoint. Thus, the DMA access from a remote NT endpoint will
appear on the local bus to have come from the unknown devfn which the IOMMU
will reject.
Interrogate NTB hardware registers for each remote NT endpoint to obtain
the proxy IDs that have been assigned to it and alias them to the local
(enumerated) NT endpoint's device. The IOMMU then accepts the remote proxy
IDs as if they were requests coming directly from the enumerated endpoint,
giving remote requestors access to memory resources which the local host
has made available.
Note that the aliasing of the proxy IDs cannot be performed at the driver
level given the current IOMMU architecture. Superficially this is because
pci_add_dma_alias() symbol is not exported. Functionally, the current
IOMMU design requires the aliasing to be performed prior to the creation of
IOMMU groups. If a driver were to attempt to use pci_add_dma_alias() in
its probe routine it would fail since the IOMMU groups have been set up by
that time. If the Switchtec hardware supported dynamic proxy ID
(re-)assignment this would be an issue, but it does not.
To further clarify static proxy ID assignment: While the requester ID to
proxy ID mapping can be dynamically changed, the number and value of proxy
IDs given to an NT EP cannot, even for dynamic reconfiguration such as
hot-add. Therefore, the chip configuration must account a priori for the
proxy IDs needs, considering both static and dynamic system configurations.
For example, a port on the chip may not having anything plugged into it at
start of day; but it must have a sufficient number of proxy IDs assigned to
accommodate the supported devices which may be hot-added.
Switchtec NTB functionality with the IOMMU off is unchanged by this quirk.
Signed-off-by: Doug Meyer <dmeyer@gigaio.com>
[bhelgaas: use hard-coded Device IDs instead of adding #defines for each]
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Logan Gunthorpe <logang@deltatee.com>
2018-05-23 23:18:06 +03:00
/*
* Microsemi Switchtec NTB uses devfn proxy IDs to move TLPs between
* NT endpoints via the internal switch fabric . These IDs replace the
* originating requestor ID TLPs which access host memory on peer NTB
* ports . Therefore , all proxy IDs must be aliased to the NTB device
* to permit access when the IOMMU is turned on .
*/
static void quirk_switchtec_ntb_dma_alias ( struct pci_dev * pdev )
{
void __iomem * mmio ;
struct ntb_info_regs __iomem * mmio_ntb ;
struct ntb_ctrl_regs __iomem * mmio_ctrl ;
u64 partition_map ;
u8 partition ;
int pp ;
if ( pci_enable_device ( pdev ) ) {
pci_err ( pdev , " Cannot enable Switchtec device \n " ) ;
return ;
}
mmio = pci_iomap ( pdev , 0 , 0 ) ;
if ( mmio = = NULL ) {
pci_disable_device ( pdev ) ;
pci_err ( pdev , " Cannot iomap Switchtec device \n " ) ;
return ;
}
pci_info ( pdev , " Setting Switchtec proxy ID aliases \n " ) ;
mmio_ntb = mmio + SWITCHTEC_GAS_NTB_OFFSET ;
mmio_ctrl = ( void __iomem * ) mmio_ntb + SWITCHTEC_NTB_REG_CTRL_OFFSET ;
partition = ioread8 ( & mmio_ntb - > partition_id ) ;
partition_map = ioread32 ( & mmio_ntb - > ep_map ) ;
partition_map | = ( ( u64 ) ioread32 ( & mmio_ntb - > ep_map + 4 ) ) < < 32 ;
partition_map & = ~ ( 1ULL < < partition ) ;
for ( pp = 0 ; pp < ( sizeof ( partition_map ) * 8 ) ; pp + + ) {
struct ntb_ctrl_regs __iomem * mmio_peer_ctrl ;
u32 table_sz = 0 ;
int te ;
if ( ! ( partition_map & ( 1ULL < < pp ) ) )
continue ;
pci_dbg ( pdev , " Processing partition %d \n " , pp ) ;
mmio_peer_ctrl = & mmio_ctrl [ pp ] ;
table_sz = ioread16 ( & mmio_peer_ctrl - > req_id_table_size ) ;
if ( ! table_sz ) {
pci_warn ( pdev , " Partition %d table_sz 0 \n " , pp ) ;
continue ;
}
if ( table_sz > 512 ) {
pci_warn ( pdev ,
" Invalid Switchtec partition %d table_sz %d \n " ,
pp , table_sz ) ;
continue ;
}
for ( te = 0 ; te < table_sz ; te + + ) {
u32 rid_entry ;
u8 devfn ;
rid_entry = ioread32 ( & mmio_peer_ctrl - > req_id_table [ te ] ) ;
devfn = ( rid_entry > > 1 ) & 0xFF ;
pci_dbg ( pdev ,
" Aliasing Partition %d Proxy ID %02x.%d \n " ,
pp , PCI_SLOT ( devfn ) , PCI_FUNC ( devfn ) ) ;
2019-12-11 01:07:30 +03:00
pci_add_dma_alias ( pdev , devfn , 1 ) ;
PCI: Add DMA alias quirk for Microsemi Switchtec NTB
Add a quirk for the Microsemi Switchtec parts to allow DMA access via
non-transparent bridging to work when the IOMMU is turned on.
This exclusively addresses the ability of a remote NT endpoint to perform
DMA accesses through the locally enumerated NT endpoint. Other aspects of
the Switchtec NTB functionality, such as interrupts for doorbells and
messages are independent of this quirk, and will work whether the IOMMU is
on or off.
When a requestor on one NT endpoint accesses memory on another NT endpoint,
it does this via a devfn proxy ID. Proxy IDs are statically assigned to
each NT endpoint by the NTB hardware as part of the release-from-reset
sequence prior to PCI enumeration. These proxy IDs cannot be modified
dynamically, and are not visible to the host during enumeration.
When the Switchtec NTB driver loads it will map local requestor IDs, such
as the root complex and transparent bridge DMA engines, to proxy IDs by
populating those requestor IDs in hardware mapping table table entries.
This establishes a fixed relationship between a requestor ID and a proxy
ID.
When a peer on a remote NT endpoint performs an access within a particular
translation window in it's NT endpoint BAR address space, that access is
translated to a DMA request on the local endpoint's bus. As part of the
translation process, the original requestor ID has its devfn replaced with
the proxy ID, and the bus portion of the BDF is replaced with the bus of
the local NT endpoint. Thus, the DMA access from a remote NT endpoint will
appear on the local bus to have come from the unknown devfn which the IOMMU
will reject.
Interrogate NTB hardware registers for each remote NT endpoint to obtain
the proxy IDs that have been assigned to it and alias them to the local
(enumerated) NT endpoint's device. The IOMMU then accepts the remote proxy
IDs as if they were requests coming directly from the enumerated endpoint,
giving remote requestors access to memory resources which the local host
has made available.
Note that the aliasing of the proxy IDs cannot be performed at the driver
level given the current IOMMU architecture. Superficially this is because
pci_add_dma_alias() symbol is not exported. Functionally, the current
IOMMU design requires the aliasing to be performed prior to the creation of
IOMMU groups. If a driver were to attempt to use pci_add_dma_alias() in
its probe routine it would fail since the IOMMU groups have been set up by
that time. If the Switchtec hardware supported dynamic proxy ID
(re-)assignment this would be an issue, but it does not.
To further clarify static proxy ID assignment: While the requester ID to
proxy ID mapping can be dynamically changed, the number and value of proxy
IDs given to an NT EP cannot, even for dynamic reconfiguration such as
hot-add. Therefore, the chip configuration must account a priori for the
proxy IDs needs, considering both static and dynamic system configurations.
For example, a port on the chip may not having anything plugged into it at
start of day; but it must have a sufficient number of proxy IDs assigned to
accommodate the supported devices which may be hot-added.
Switchtec NTB functionality with the IOMMU off is unchanged by this quirk.
Signed-off-by: Doug Meyer <dmeyer@gigaio.com>
[bhelgaas: use hard-coded Device IDs instead of adding #defines for each]
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Logan Gunthorpe <logang@deltatee.com>
2018-05-23 23:18:06 +03:00
}
}
pci_iounmap ( pdev , mmio ) ;
pci_disable_device ( pdev ) ;
}
2018-10-10 23:55:05 +03:00
# define SWITCHTEC_QUIRK(vid) \
2018-10-05 18:49:40 +03:00
DECLARE_PCI_FIXUP_CLASS_FINAL ( PCI_VENDOR_ID_MICROSEMI , vid , \
PCI_CLASS_BRIDGE_OTHER , 8 , quirk_switchtec_ntb_dma_alias )
2018-10-10 23:55:05 +03:00
SWITCHTEC_QUIRK ( 0x8531 ) ; /* PFX 24xG3 */
SWITCHTEC_QUIRK ( 0x8532 ) ; /* PFX 32xG3 */
SWITCHTEC_QUIRK ( 0x8533 ) ; /* PFX 48xG3 */
SWITCHTEC_QUIRK ( 0x8534 ) ; /* PFX 64xG3 */
SWITCHTEC_QUIRK ( 0x8535 ) ; /* PFX 80xG3 */
SWITCHTEC_QUIRK ( 0x8536 ) ; /* PFX 96xG3 */
SWITCHTEC_QUIRK ( 0x8541 ) ; /* PSX 24xG3 */
SWITCHTEC_QUIRK ( 0x8542 ) ; /* PSX 32xG3 */
SWITCHTEC_QUIRK ( 0x8543 ) ; /* PSX 48xG3 */
SWITCHTEC_QUIRK ( 0x8544 ) ; /* PSX 64xG3 */
SWITCHTEC_QUIRK ( 0x8545 ) ; /* PSX 80xG3 */
SWITCHTEC_QUIRK ( 0x8546 ) ; /* PSX 96xG3 */
SWITCHTEC_QUIRK ( 0x8551 ) ; /* PAX 24XG3 */
SWITCHTEC_QUIRK ( 0x8552 ) ; /* PAX 32XG3 */
SWITCHTEC_QUIRK ( 0x8553 ) ; /* PAX 48XG3 */
SWITCHTEC_QUIRK ( 0x8554 ) ; /* PAX 64XG3 */
SWITCHTEC_QUIRK ( 0x8555 ) ; /* PAX 80XG3 */
SWITCHTEC_QUIRK ( 0x8556 ) ; /* PAX 96XG3 */
SWITCHTEC_QUIRK ( 0x8561 ) ; /* PFXL 24XG3 */
SWITCHTEC_QUIRK ( 0x8562 ) ; /* PFXL 32XG3 */
SWITCHTEC_QUIRK ( 0x8563 ) ; /* PFXL 48XG3 */
SWITCHTEC_QUIRK ( 0x8564 ) ; /* PFXL 64XG3 */
SWITCHTEC_QUIRK ( 0x8565 ) ; /* PFXL 80XG3 */
SWITCHTEC_QUIRK ( 0x8566 ) ; /* PFXL 96XG3 */
SWITCHTEC_QUIRK ( 0x8571 ) ; /* PFXI 24XG3 */
SWITCHTEC_QUIRK ( 0x8572 ) ; /* PFXI 32XG3 */
SWITCHTEC_QUIRK ( 0x8573 ) ; /* PFXI 48XG3 */
SWITCHTEC_QUIRK ( 0x8574 ) ; /* PFXI 64XG3 */
SWITCHTEC_QUIRK ( 0x8575 ) ; /* PFXI 80XG3 */
SWITCHTEC_QUIRK ( 0x8576 ) ; /* PFXI 96XG3 */
2020-01-15 06:56:48 +03:00
SWITCHTEC_QUIRK ( 0x4000 ) ; /* PFX 100XG4 */
SWITCHTEC_QUIRK ( 0x4084 ) ; /* PFX 84XG4 */
SWITCHTEC_QUIRK ( 0x4068 ) ; /* PFX 68XG4 */
SWITCHTEC_QUIRK ( 0x4052 ) ; /* PFX 52XG4 */
SWITCHTEC_QUIRK ( 0x4036 ) ; /* PFX 36XG4 */
SWITCHTEC_QUIRK ( 0x4028 ) ; /* PFX 28XG4 */
SWITCHTEC_QUIRK ( 0x4100 ) ; /* PSX 100XG4 */
SWITCHTEC_QUIRK ( 0x4184 ) ; /* PSX 84XG4 */
SWITCHTEC_QUIRK ( 0x4168 ) ; /* PSX 68XG4 */
SWITCHTEC_QUIRK ( 0x4152 ) ; /* PSX 52XG4 */
SWITCHTEC_QUIRK ( 0x4136 ) ; /* PSX 36XG4 */
SWITCHTEC_QUIRK ( 0x4128 ) ; /* PSX 28XG4 */
SWITCHTEC_QUIRK ( 0x4200 ) ; /* PAX 100XG4 */
SWITCHTEC_QUIRK ( 0x4284 ) ; /* PAX 84XG4 */
SWITCHTEC_QUIRK ( 0x4268 ) ; /* PAX 68XG4 */
SWITCHTEC_QUIRK ( 0x4252 ) ; /* PAX 52XG4 */
SWITCHTEC_QUIRK ( 0x4236 ) ; /* PAX 36XG4 */
SWITCHTEC_QUIRK ( 0x4228 ) ; /* PAX 28XG4 */
PCI: Reset Lenovo ThinkPad P50 nvgpu at boot if necessary
On ThinkPad P50 SKUs with an Nvidia Quadro M1000M instead of the M2000M
variant, the BIOS does not always reset the secondary Nvidia GPU during
reboot if the laptop is configured in Hybrid Graphics mode. The reason is
unknown, but the following steps and possibly a good bit of patience will
reproduce the issue:
1. Boot up the laptop normally in Hybrid Graphics mode
2. Make sure nouveau is loaded and that the GPU is awake
3. Allow the Nvidia GPU to runtime suspend itself after being idle
4. Reboot the machine, the more sudden the better (e.g. sysrq-b may help)
5. If nouveau loads up properly, reboot the machine again and go back to
step 2 until you reproduce the issue
This results in some very strange behavior: the GPU will be left in exactly
the same state it was in when the previously booted kernel started the
reboot. This has all sorts of bad side effects: for starters, this
completely breaks nouveau starting with a mysterious EVO channel failure
that happens well before we've actually used the EVO channel for anything:
nouveau 0000:01:00.0: disp: chid 0 mthd 0000 data 00000400 00001000 00000002
This causes a timeout trying to bring up the GR ctx:
nouveau 0000:01:00.0: timeout
WARNING: CPU: 0 PID: 12 at drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf100.c:1547 gf100_grctx_generate+0x7b2/0x850 [nouveau]
Hardware name: LENOVO 20EQS64N0B/20EQS64N0B, BIOS N1EET82W (1.55 ) 12/18/2018
Workqueue: events_long drm_dp_mst_link_probe_work [drm_kms_helper]
...
nouveau 0000:01:00.0: gr: wait for idle timeout (en: 1, ctxsw: 0, busy: 1)
nouveau 0000:01:00.0: gr: wait for idle timeout (en: 1, ctxsw: 0, busy: 1)
nouveau 0000:01:00.0: fifo: fault 01 [WRITE] at 0000000000008000 engine 00 [GR] client 15 [HUB/SCC_NB] reason c4 [] on channel -1 [0000000000 unknown]
The GPU never manages to recover. Booting without loading nouveau causes
issues as well, since the GPU starts sending spurious interrupts that cause
other device's IRQs to get disabled by the kernel:
irq 16: nobody cared (try booting with the "irqpoll" option)
...
handlers:
[<000000007faa9e99>] i801_isr [i2c_i801]
Disabling IRQ #16
...
serio: RMI4 PS/2 pass-through port at rmi4-00.fn03
i801_smbus 0000:00:1f.4: Timeout waiting for interrupt!
i801_smbus 0000:00:1f.4: Transaction timeout
rmi4_f03 rmi4-00.fn03: rmi_f03_pt_write: Failed to write to F03 TX register (-110).
i801_smbus 0000:00:1f.4: Timeout waiting for interrupt!
i801_smbus 0000:00:1f.4: Transaction timeout
rmi4_physical rmi4-00: rmi_driver_set_irq_bits: Failed to change enabled interrupts!
This causes the touchpad and sometimes other things to get disabled.
Since this happens without nouveau, we can't fix this problem from nouveau
itself.
Add a PCI quirk for the specific P50 variant of this GPU. Make sure the
GPU is advertising NoReset- so we don't reset the GPU when the machine is
in Dedicated graphics mode (where the GPU being initialized by the BIOS is
normal and expected). Map the GPU MMIO space and read the magic 0x2240c
register, which will have bit 1 set if the device was POSTed during a
previous boot. Once we've confirmed all of this, reset the GPU and
re-disable it - bringing it back to a healthy state.
Link: https://bugzilla.kernel.org/show_bug.cgi?id=203003
Link: https://lore.kernel.org/lkml/20190212220230.1568-1-lyude@redhat.com
Signed-off-by: Lyude Paul <lyude@redhat.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Cc: nouveau@lists.freedesktop.org
Cc: dri-devel@lists.freedesktop.org
Cc: Karol Herbst <kherbst@redhat.com>
Cc: Ben Skeggs <skeggsb@gmail.com>
Cc: stable@vger.kernel.org
2019-02-13 01:02:30 +03:00
2019-12-11 01:25:40 +03:00
/*
* The PLX NTB uses devfn proxy IDs to move TLPs between NT endpoints .
* These IDs are used to forward responses to the originator on the other
* side of the NTB . Alias all possible IDs to the NTB to permit access when
* the IOMMU is turned on .
*/
static void quirk_plx_ntb_dma_alias ( struct pci_dev * pdev )
{
pci_info ( pdev , " Setting PLX NTB proxy ID aliases \n " ) ;
/* PLX NTB may use all 256 devfns */
pci_add_dma_alias ( pdev , 0 , 256 ) ;
}
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_PLX , 0x87b0 , quirk_plx_ntb_dma_alias ) ;
DECLARE_PCI_FIXUP_HEADER ( PCI_VENDOR_ID_PLX , 0x87b1 , quirk_plx_ntb_dma_alias ) ;
PCI: Reset Lenovo ThinkPad P50 nvgpu at boot if necessary
On ThinkPad P50 SKUs with an Nvidia Quadro M1000M instead of the M2000M
variant, the BIOS does not always reset the secondary Nvidia GPU during
reboot if the laptop is configured in Hybrid Graphics mode. The reason is
unknown, but the following steps and possibly a good bit of patience will
reproduce the issue:
1. Boot up the laptop normally in Hybrid Graphics mode
2. Make sure nouveau is loaded and that the GPU is awake
3. Allow the Nvidia GPU to runtime suspend itself after being idle
4. Reboot the machine, the more sudden the better (e.g. sysrq-b may help)
5. If nouveau loads up properly, reboot the machine again and go back to
step 2 until you reproduce the issue
This results in some very strange behavior: the GPU will be left in exactly
the same state it was in when the previously booted kernel started the
reboot. This has all sorts of bad side effects: for starters, this
completely breaks nouveau starting with a mysterious EVO channel failure
that happens well before we've actually used the EVO channel for anything:
nouveau 0000:01:00.0: disp: chid 0 mthd 0000 data 00000400 00001000 00000002
This causes a timeout trying to bring up the GR ctx:
nouveau 0000:01:00.0: timeout
WARNING: CPU: 0 PID: 12 at drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf100.c:1547 gf100_grctx_generate+0x7b2/0x850 [nouveau]
Hardware name: LENOVO 20EQS64N0B/20EQS64N0B, BIOS N1EET82W (1.55 ) 12/18/2018
Workqueue: events_long drm_dp_mst_link_probe_work [drm_kms_helper]
...
nouveau 0000:01:00.0: gr: wait for idle timeout (en: 1, ctxsw: 0, busy: 1)
nouveau 0000:01:00.0: gr: wait for idle timeout (en: 1, ctxsw: 0, busy: 1)
nouveau 0000:01:00.0: fifo: fault 01 [WRITE] at 0000000000008000 engine 00 [GR] client 15 [HUB/SCC_NB] reason c4 [] on channel -1 [0000000000 unknown]
The GPU never manages to recover. Booting without loading nouveau causes
issues as well, since the GPU starts sending spurious interrupts that cause
other device's IRQs to get disabled by the kernel:
irq 16: nobody cared (try booting with the "irqpoll" option)
...
handlers:
[<000000007faa9e99>] i801_isr [i2c_i801]
Disabling IRQ #16
...
serio: RMI4 PS/2 pass-through port at rmi4-00.fn03
i801_smbus 0000:00:1f.4: Timeout waiting for interrupt!
i801_smbus 0000:00:1f.4: Transaction timeout
rmi4_f03 rmi4-00.fn03: rmi_f03_pt_write: Failed to write to F03 TX register (-110).
i801_smbus 0000:00:1f.4: Timeout waiting for interrupt!
i801_smbus 0000:00:1f.4: Transaction timeout
rmi4_physical rmi4-00: rmi_driver_set_irq_bits: Failed to change enabled interrupts!
This causes the touchpad and sometimes other things to get disabled.
Since this happens without nouveau, we can't fix this problem from nouveau
itself.
Add a PCI quirk for the specific P50 variant of this GPU. Make sure the
GPU is advertising NoReset- so we don't reset the GPU when the machine is
in Dedicated graphics mode (where the GPU being initialized by the BIOS is
normal and expected). Map the GPU MMIO space and read the magic 0x2240c
register, which will have bit 1 set if the device was POSTed during a
previous boot. Once we've confirmed all of this, reset the GPU and
re-disable it - bringing it back to a healthy state.
Link: https://bugzilla.kernel.org/show_bug.cgi?id=203003
Link: https://lore.kernel.org/lkml/20190212220230.1568-1-lyude@redhat.com
Signed-off-by: Lyude Paul <lyude@redhat.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Cc: nouveau@lists.freedesktop.org
Cc: dri-devel@lists.freedesktop.org
Cc: Karol Herbst <kherbst@redhat.com>
Cc: Ben Skeggs <skeggsb@gmail.com>
Cc: stable@vger.kernel.org
2019-02-13 01:02:30 +03:00
/*
* On Lenovo Thinkpad P50 SKUs with a Nvidia Quadro M1000M , the BIOS does
* not always reset the secondary Nvidia GPU between reboots if the system
* is configured to use Hybrid Graphics mode . This results in the GPU
* being left in whatever state it was in during the * previous * boot , which
* causes spurious interrupts from the GPU , which in turn causes us to
* disable the wrong IRQ and end up breaking the touchpad . Unsurprisingly ,
* this also completely breaks nouveau .
*
* Luckily , it seems a simple reset of the Nvidia GPU brings it back to a
* clean state and fixes all these issues .
*
* When the machine is configured in Dedicated display mode , the issue
* doesn ' t occur . Fortunately the GPU advertises NoReset + when in this
* mode , so we can detect that and avoid resetting it .
*/
static void quirk_reset_lenovo_thinkpad_p50_nvgpu ( struct pci_dev * pdev )
{
void __iomem * map ;
int ret ;
if ( pdev - > subsystem_vendor ! = PCI_VENDOR_ID_LENOVO | |
pdev - > subsystem_device ! = 0x222e | |
! pdev - > reset_fn )
return ;
if ( pci_enable_device_mem ( pdev ) )
return ;
/*
* Based on nvkm_device_ctor ( ) in
* drivers / gpu / drm / nouveau / nvkm / engine / device / base . c
*/
map = pci_iomap ( pdev , 0 , 0x23000 ) ;
if ( ! map ) {
pci_err ( pdev , " Can't map MMIO space \n " ) ;
goto out_disable ;
}
/*
* Make sure the GPU looks like it ' s been POSTed before resetting
* it .
*/
if ( ioread32 ( map + 0x2240c ) & 0x2 ) {
pci_info ( pdev , FW_BUG " GPU left initialized by EFI, resetting \n " ) ;
2019-08-02 01:01:17 +03:00
ret = pci_reset_bus ( pdev ) ;
PCI: Reset Lenovo ThinkPad P50 nvgpu at boot if necessary
On ThinkPad P50 SKUs with an Nvidia Quadro M1000M instead of the M2000M
variant, the BIOS does not always reset the secondary Nvidia GPU during
reboot if the laptop is configured in Hybrid Graphics mode. The reason is
unknown, but the following steps and possibly a good bit of patience will
reproduce the issue:
1. Boot up the laptop normally in Hybrid Graphics mode
2. Make sure nouveau is loaded and that the GPU is awake
3. Allow the Nvidia GPU to runtime suspend itself after being idle
4. Reboot the machine, the more sudden the better (e.g. sysrq-b may help)
5. If nouveau loads up properly, reboot the machine again and go back to
step 2 until you reproduce the issue
This results in some very strange behavior: the GPU will be left in exactly
the same state it was in when the previously booted kernel started the
reboot. This has all sorts of bad side effects: for starters, this
completely breaks nouveau starting with a mysterious EVO channel failure
that happens well before we've actually used the EVO channel for anything:
nouveau 0000:01:00.0: disp: chid 0 mthd 0000 data 00000400 00001000 00000002
This causes a timeout trying to bring up the GR ctx:
nouveau 0000:01:00.0: timeout
WARNING: CPU: 0 PID: 12 at drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf100.c:1547 gf100_grctx_generate+0x7b2/0x850 [nouveau]
Hardware name: LENOVO 20EQS64N0B/20EQS64N0B, BIOS N1EET82W (1.55 ) 12/18/2018
Workqueue: events_long drm_dp_mst_link_probe_work [drm_kms_helper]
...
nouveau 0000:01:00.0: gr: wait for idle timeout (en: 1, ctxsw: 0, busy: 1)
nouveau 0000:01:00.0: gr: wait for idle timeout (en: 1, ctxsw: 0, busy: 1)
nouveau 0000:01:00.0: fifo: fault 01 [WRITE] at 0000000000008000 engine 00 [GR] client 15 [HUB/SCC_NB] reason c4 [] on channel -1 [0000000000 unknown]
The GPU never manages to recover. Booting without loading nouveau causes
issues as well, since the GPU starts sending spurious interrupts that cause
other device's IRQs to get disabled by the kernel:
irq 16: nobody cared (try booting with the "irqpoll" option)
...
handlers:
[<000000007faa9e99>] i801_isr [i2c_i801]
Disabling IRQ #16
...
serio: RMI4 PS/2 pass-through port at rmi4-00.fn03
i801_smbus 0000:00:1f.4: Timeout waiting for interrupt!
i801_smbus 0000:00:1f.4: Transaction timeout
rmi4_f03 rmi4-00.fn03: rmi_f03_pt_write: Failed to write to F03 TX register (-110).
i801_smbus 0000:00:1f.4: Timeout waiting for interrupt!
i801_smbus 0000:00:1f.4: Transaction timeout
rmi4_physical rmi4-00: rmi_driver_set_irq_bits: Failed to change enabled interrupts!
This causes the touchpad and sometimes other things to get disabled.
Since this happens without nouveau, we can't fix this problem from nouveau
itself.
Add a PCI quirk for the specific P50 variant of this GPU. Make sure the
GPU is advertising NoReset- so we don't reset the GPU when the machine is
in Dedicated graphics mode (where the GPU being initialized by the BIOS is
normal and expected). Map the GPU MMIO space and read the magic 0x2240c
register, which will have bit 1 set if the device was POSTed during a
previous boot. Once we've confirmed all of this, reset the GPU and
re-disable it - bringing it back to a healthy state.
Link: https://bugzilla.kernel.org/show_bug.cgi?id=203003
Link: https://lore.kernel.org/lkml/20190212220230.1568-1-lyude@redhat.com
Signed-off-by: Lyude Paul <lyude@redhat.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Cc: nouveau@lists.freedesktop.org
Cc: dri-devel@lists.freedesktop.org
Cc: Karol Herbst <kherbst@redhat.com>
Cc: Ben Skeggs <skeggsb@gmail.com>
Cc: stable@vger.kernel.org
2019-02-13 01:02:30 +03:00
if ( ret < 0 )
pci_err ( pdev , " Failed to reset GPU: %d \n " , ret ) ;
}
iounmap ( map ) ;
out_disable :
pci_disable_device ( pdev ) ;
}
DECLARE_PCI_FIXUP_CLASS_FINAL ( PCI_VENDOR_ID_NVIDIA , 0x13b1 ,
PCI_CLASS_DISPLAY_VGA , 8 ,
quirk_reset_lenovo_thinkpad_p50_nvgpu ) ;
2019-12-19 22:20:06 +03:00
/*
* Device [ 1 b21 : 2142 ]
* When in D0 , PME # doesn ' t get asserted when plugging USB 3.0 device .
*/
static void pci_fixup_no_d0_pme ( struct pci_dev * dev )
{
pci_info ( dev , " PME# does not work under D0, disabling it \n " ) ;
dev - > pme_support & = ~ ( PCI_PM_CAP_PME_D0 > > PCI_PM_CAP_PME_SHIFT ) ;
}
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_ASMEDIA , 0x2142 , pci_fixup_no_d0_pme ) ;
PCI: Move Apex Edge TPU class quirk to fix BAR assignment
Some Google Apex Edge TPU devices have a class code of 0
(PCI_CLASS_NOT_DEFINED). This prevents the PCI core from assigning
resources for the Apex BARs because __dev_sort_resources() ignores
classless devices, host bridges, and IOAPICs.
On x86, firmware typically assigns those resources, so this was not a
problem. But on some architectures, firmware does *not* assign BARs, and
since the PCI core didn't do it either, the Apex device didn't work
correctly:
apex 0000:01:00.0: can't enable device: BAR 0 [mem 0x00000000-0x00003fff 64bit pref] not claimed
apex 0000:01:00.0: error enabling PCI device
f390d08d8b87 ("staging: gasket: apex: fixup undefined PCI class") added a
quirk to fix the class code, but it was in the apex driver, and if the
driver was built as a module, it was too late to help.
Move the quirk to the PCI core, where it will always run early enough that
the PCI core will assign resources if necessary.
Link: https://lore.kernel.org/r/CAEzXK1r0Er039iERnc2KJ4jn7ySNUOG9H=Ha8TD8XroVqiZjgg@mail.gmail.com
Fixes: f390d08d8b87 ("staging: gasket: apex: fixup undefined PCI class")
Reported-by: Luís Mendes <luis.p.mendes@gmail.com>
Debugged-by: Luís Mendes <luis.p.mendes@gmail.com>
Tested-by: Luis Mendes <luis.p.mendes@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Cc: Todd Poynor <toddpoynor@google.com>
2020-04-09 20:43:45 +03:00
2020-05-08 09:53:41 +03:00
/*
* Device [ 12 d8 : 0x400e ] and [ 12 d8 : 0x400f ]
* These devices advertise PME # support in all power states but don ' t
* reliably assert it .
*/
static void pci_fixup_no_pme ( struct pci_dev * dev )
{
pci_info ( dev , " PME# is unreliable, disabling it \n " ) ;
dev - > pme_support = 0 ;
}
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_PERICOM , 0x400e , pci_fixup_no_pme ) ;
DECLARE_PCI_FIXUP_FINAL ( PCI_VENDOR_ID_PERICOM , 0x400f , pci_fixup_no_pme ) ;
2020-06-06 21:01:58 +03:00
PCI: Move Apex Edge TPU class quirk to fix BAR assignment
Some Google Apex Edge TPU devices have a class code of 0
(PCI_CLASS_NOT_DEFINED). This prevents the PCI core from assigning
resources for the Apex BARs because __dev_sort_resources() ignores
classless devices, host bridges, and IOAPICs.
On x86, firmware typically assigns those resources, so this was not a
problem. But on some architectures, firmware does *not* assign BARs, and
since the PCI core didn't do it either, the Apex device didn't work
correctly:
apex 0000:01:00.0: can't enable device: BAR 0 [mem 0x00000000-0x00003fff 64bit pref] not claimed
apex 0000:01:00.0: error enabling PCI device
f390d08d8b87 ("staging: gasket: apex: fixup undefined PCI class") added a
quirk to fix the class code, but it was in the apex driver, and if the
driver was built as a module, it was too late to help.
Move the quirk to the PCI core, where it will always run early enough that
the PCI core will assign resources if necessary.
Link: https://lore.kernel.org/r/CAEzXK1r0Er039iERnc2KJ4jn7ySNUOG9H=Ha8TD8XroVqiZjgg@mail.gmail.com
Fixes: f390d08d8b87 ("staging: gasket: apex: fixup undefined PCI class")
Reported-by: Luís Mendes <luis.p.mendes@gmail.com>
Debugged-by: Luís Mendes <luis.p.mendes@gmail.com>
Tested-by: Luis Mendes <luis.p.mendes@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Cc: Todd Poynor <toddpoynor@google.com>
2020-04-09 20:43:45 +03:00
static void apex_pci_fixup_class ( struct pci_dev * pdev )
{
pdev - > class = ( PCI_CLASS_SYSTEM_OTHER < < 8 ) | pdev - > class ;
}
DECLARE_PCI_FIXUP_CLASS_HEADER ( 0x1ac1 , 0x089a ,
PCI_CLASS_NOT_DEFINED , 8 , apex_pci_fixup_class ) ;
