pcie_port_device_remove currently calls the remove method of port
drivers twice. Ouch!
We are calling device_for_each_child multiple times for no apparent
reason.
So make it simple. Place put_device and device_unregister into
remove_iter, and throw out the rest. Only call device_for_each_child
once.
The code is simpler and actually works!
Signed-off-by: Eric W. Biederman <ebiederm@aristanetworks.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
This closes http://bugzilla.kernel.org/show_bug.cgi?id=10893
which is a showstopper for X development on alpha.
The generic HAVE_PCI_MMAP code (drivers/pci-sysfs.c) is not
very useful since we have to deal with three different types
of MMIO address spaces: sparse and dense mappings for old
ev4/ev5 machines and "normal" 1:1 MMIO space (bwx) for ev56 and
later.
Also "write combine" mappings are meaningless on alpha - roughly
speaking, alpha does write combining, IO reordering and other
optimizations by default, unless user splits IO accesses
with memory barriers.
I think the cleanest way to deal with resource files on alpha
is to convert the default no-op pci_create_resource_files() and
pci_remove_resource_files() for !HAVE_PCI_MMAP case into __weak
functions and override them with alpha specific ones.
Another alpha hook is needed for "legacy_" resource files
to handle sparse addressing (pci_adjust_legacy_attr).
With the "standard" resourceN files on ev56/ev6 libpciaccess
works "out of the box". Handling of resourceN_sparse/resourceN_dense
files on older machines obviously requires some userland work.
Sparse/dense stuff has been tested on sx164 (pca56/pyxis, normally
uses bwx IO) with the kernel hacked into "cia compatible" mode.
Signed-off-by: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
drivers/pci/hotplug/fakephp.c: In function 'pci_rescan_bus':
drivers/pci/hotplug/fakephp.c:271: warning: passing argument 1 of 'pci_bus_assign_resources' discards qualifiers from pointer target type
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
There is code in msix_capability_init() which, when the requested number
of MSI-X couldn't be allocated, calculates how many MSI-X /could/ be
allocated and returns that to the driver. That allows the driver to then
make a second request, with a number of MSIs that should succeed.
The current code requires the arch code to setup as many msi_descs as it
can, and then return to the generic code. On some platforms the arch
code may already know how many MSI-X it can allocate, before it sets up
any of the msi_descs.
So change the logic such that if the arch code returns a positive error
code, that is taken to be the number of MSI-X that could be allocated.
If the error code is negative we still calculate the number available
using the old method.
Because it's a little subtle, make sure the error return code from
arch_setup_msi_irq() is always negative. That way only implementations
of arch_setup_msi_irqs() need to be careful about returning a positive
error code.
Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
With for (busnr = 0; busnr <= end; busnr++) { ... } busnr reaches end + 1
after the loop. So fix the "no busses available" check to look for just
busnr > end rather than >=.
Signed-off-by: Roel Kluin <roel.kluin@gmail.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
- Rename pci_osc_control_set() to acpi_pci_osc_control_set() according
to the other API names in drivers/acpi/pci_root.c.
- Move _OSC related definitions to include/linux/acpi.h because _OSC
related API is implemented in drivers/acpi/pci_root.c now.
Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Reviewed-by: Andrew Patterson <andrew.patterson@hp.com>
Tested-by: Andrew Patterson <andrew.patterson@hp.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Move PCI _OSC management code from drivers/pci/pci-acpi.c to
drivers/acpi/pci_root.c. The benefits are
- We no longer need struct osc_data and its management code (contents
are moved to struct acpi_pci_root). This simplify the code, and we
no longer care about kmalloc() failure.
- We can make pci_acpi_osc_support() be a static function, which is
called only from drivers/acpi/pci_root.c.
Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Reviewed-by: Andrew Patterson <andrew.patterson@hp.com>
Tested-by: Andrew Patterson <andrew.patterson@hp.com>
Acked-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
For all devices need to do function level reset, currently we need wait for
at least 200ms, which can be too long if we have lots of devices...
The patch checked pending bit before msleep() to skip some unnecessary
sleeping interval.
Signed-off-by: Sheng Yang <sheng@linux.intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Let it stay as serial, since it doesn't have subdevice in the form of 0x00PS.
Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Convert usages of pr_debug to dev_dbg and add physical slot name.
Note that we use dev_dbg on the struct pci_bus and still manually
print out the PCI slot number (instead of calling dev_dbg on a
pci_dev) because a struct pci_bus with empty physical slots will
not have any pci_devs.
Reviewed-by: Andrew Patterson <andrew.patterson@hp.com>
Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
The cmd_busy field in struct controller takes only two values 0 or
1. So it should be one bit.
Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Current pciehp disables software notification of adapter presence
changed event and MRL changed event when slot is turned off. Because
of this, there is no way to detect those events on empty slots in the
current pciehp implementation.
According to the past discussion(*), this behavior was introduced to
prevent endless loop that could happen if pcie_isr() runs after power
fault is detected on a certain platform whose stickey power-fault bit
remains on till the slot is powered on again.
(*) http://sourceforge.net/mailarchive/message.php?msg_id=20051130135409.A14918%40unix-os.sc.intel.com
I think this endless loop can be avoided using one bit flag that
indicates power fault had been detected, instead of disabling software
notification of adapter present changed event and MRL changed event.
With this patch, we can enable software notification mechanism of
presence changed and MRL changed event on the empty slots again.
Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Fix possible endless loop in pcie_isr.
Currently, pcie_isr() (interrupt service routine of pciehp) can end up in an
endless loop if the Slot Status register is set again immediately after being
cleared. According to the past discussion (see below URL) this case can happen
if the power fault detected bit is set during handling.
http://sourceforge.net/mailarchive/message.php?msg_id=20051130135409.A14918%40unix-os.sc.intel.com
Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Error handling code following a kmalloc should free the allocated data.
Since the subsequent code that could provoke an error does not use the
allocated data, the allocation is just moved below it.
The semantic match that finds the problem is as follows:
(http://www.emn.fr/x-info/coccinelle/)
// <smpl>
@r exists@
local idexpression x;
statement S;
expression E;
identifier f,l;
position p1,p2;
expression *ptr != NULL;
@@
(
if ((x@p1 = \(kmalloc\|kzalloc\|kcalloc\)(...)) == NULL) S
|
x@p1 = \(kmalloc\|kzalloc\|kcalloc\)(...);
...
if (x == NULL) S
)
<... when != x
when != if (...) { <+...x...+> }
x->f = E
...>
(
return \(0\|<+...x...+>\|ptr\);
|
return@p2 ...;
)
@script:python@
p1 << r.p1;
p2 << r.p2;
@@
print "* file: %s kmalloc %s return %s" % (p1[0].file,p1[0].line,p2[0].line)
// </smpl>
Signed-off-by: Julia Lawall <julia@diku.dk>
Reviewed-by: Matthew Wilcox <willy@linux.intel.com>
Reviewed-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
According to kerneljanitors todo list all printk calls (beginning
a new line) should have an according KERN_* constant.
Those are the missing pieces here for the pci subsystem.
Signed-off-by: Frank Seidel <frank@f-seidel.de>
Reviewed-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Tested-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
When removing a bus, 'is_added' should be checked to make sure the
bus has been successfully added by pci_bus_add_child() who will sets
'is_added'.
Signed-off-by: Yu Zhao <yu.zhao@intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Weak functions aren't all they're cracked up to be. They lead to
incorrect binaries with some toolchains, they require us to have empty
functions we otherwise wouldn't, and the unused code is not elided
(as of gcc 4.3.2 anyway).
So replace the weak MSI arch hooks with the #define foo foo idiom. We no
longer need empty versions of arch_setup/teardown_msi_irq().
This is less source (by 1 line!), and results in smaller binaries too:
text data bss dec hex filename
9354300 1693916 678424 11726640 b2ef30 build/powerpc/vmlinux-before
9354052 1693852 678424 11726328 b2edf8 build/powerpc/vmlinux-after
Also smaller on x86_64 and arm (iop13xx).
Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
If MSI-X interrupt mode is used by the PCI Express port driver, too
many vectors are allocated and it is not ensured that the right
vectors will be used for the right services. Namely, the PCI Express
specification states that both PCI Express native PME and PCI Express
hotplug will always use the same MSI or MSI-X message for signalling
interrupts, which implies that the same vector will be used by both
of them. Also, the VC service does not use interrupts at all.
Moreover, is not clear which of the vectors allocated by
pci_enable_msix() in the current code will be used for PME and
hotplug and which of them will be used for AER if all of these
services are configured.
For these reasons, rework the allocation of interrupts for PCI
Express ports so that if MSI-X are enabled, the right vectors will be
used for the right purposes.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Reviewed-by: Hidetoshi Seto <seto.hidetoshi@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Introduce new function pci_msix_table_size() returning the size of
the MSI-X table of given PCI device or 0 if the device doesn't
support MSI-X.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Reviewed-by: Hidetoshi Seto <seto.hidetoshi@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
The PCI Express port driver uses 'struct pcie_port_service_id' for
matching port service devices and drivers, but this structure
contains fields that duplicate information from the port device
itself (vendor, device, subvendor, subdevice) and fields that are not
used by any existing port service driver (class, class_mask,
drvier_data). Also, both existing port service drivers (AER and
PCIe HP) don't even use the vendor and device fields for device
matching. Therefore 'struct pcie_port_service_id' can be removed
altogether and the only useful members of it (port_type, service) can
be introduced directly into the port service device and port service
driver structures. That simplifies the code quite a bit and reduces
its size.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
The second argument of the ->probe() callback in
struct pcie_port_service_driver is unnecessary and never used.
Remove it.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
The function pcie_portdrv_save_config() in portdrv_pci.c is not
necessary. Remove it.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
The PCI Express port driver calls pci_enable_device() before setting
up interrupts, which is wrong, because if there is an interrupt pin
configured for the port, pci_enable_device() will likely set up an
interrupt link for it. However, this shouldn't be done if either
MSI or MSI-X interrupt mode is chosen for the port.
The solution is to call pci_enable_device() after setting up
interrupts, because in that case the interrupt link won't be set up
if MSI or MSI-X are enabled.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
The PCI Express port driver should not attempt to register service
devices that require the ability to generate interrupts if generating
interrupts is not possible. Namely, if the port has no interrupt pin
configured and we cannot set up MSI or MSI-X for it, there is no way
it can generate interrupts and in such a case the port services that
rely on interrupts (PME, PCIe HP, AER) should not be enabled for it.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
PCI Express port driver extension, as defined by struct
pcie_port_device_ext in portdrv.h, is allocated and initialized, but
never used (it also is never freed). Extend it to hold the PCI Express
port type as well as the port interrupt mode, change its name and use it
to simplify the code in portdrv_core.c .
Additionally, remove the redundant interrupt_mode member of struct
pcie_device defined in include/linux/pcieport_if.h .
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Impact: invalid use of GFP_KERNEL in interrupt context
Queued invalidation and interrupt-remapping will get initialized with
interrupts disabled (while enabling interrupt-remapping). So use
GFP_ATOMIC instead of GFP_KERNEL for memory alloacations.
Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Impact: fix interrupt table entry leak
Fix the typo which was not clearing all the interrupt remapping table
entries corresponding to an irq.
Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Impact: cleanup/sanitization
Start from a sane state while enabling dma and interrupt-remapping, by
clearing the previous recorded faults and disabling previously
enabled queued invalidation and interrupt-remapping.
Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Impact: new interfaces (not yet used)
Routines for disabling queued invalidation and interrupt remapping.
Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Impact: interface augmentation (not yet used)
Enable fault handling flow for intr-remapping aswell. Fault handling
code now shared by both dma-remapping and intr-remapping.
Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Impact: code movement
Move page fault handling code to dmar.c
This will be shared both by DMA-remapping and Intr-remapping code.
Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Impact: fix potential deadlock on x2apic
fix "hard-safe -> hard-unsafe lock order detected" with irq_2_ir_lock
On x2apic enabled system:
[ INFO: hard-safe -> hard-unsafe lock order detected ]
2.6.27-03151-g4480f15b #1
------------------------------------------------------
swapper/1 [HC0[0]:SC0[0]:HE0:SE1] is trying to acquire:
(irq_2_ir_lock){--..}, at: [<ffffffff8038ebc0>] get_irte+0x2f/0x95
and this task is already holding:
(&irq_desc_lock_class){+...}, at: [<ffffffff802649ed>] setup_irq+0x67/0x281
which would create a new lock dependency:
(&irq_desc_lock_class){+...} -> (irq_2_ir_lock){--..}
but this new dependency connects a hard-irq-safe lock:
(&irq_desc_lock_class){+...}
... which became hard-irq-safe at:
[<ffffffffffffffff>] 0xffffffffffffffff
to a hard-irq-unsafe lock:
(irq_2_ir_lock){--..}
... which became hard-irq-unsafe at:
... [<ffffffff802547b5>] __lock_acquire+0x571/0x706
[<ffffffff8025499f>] lock_acquire+0x55/0x71
[<ffffffff8062f2c4>] _spin_lock+0x2c/0x38
[<ffffffff8038ee50>] alloc_irte+0x8a/0x14b
[<ffffffff8021f733>] setup_IO_APIC_irq+0x119/0x30e
[<ffffffff8090860e>] setup_IO_APIC+0x146/0x6e5
[<ffffffff809058fc>] native_smp_prepare_cpus+0x24e/0x2e9
[<ffffffff808f982c>] kernel_init+0x5a/0x176
[<ffffffff8020c289>] child_rip+0xa/0x11
[<ffffffffffffffff>] 0xffffffffffffffff
Fix this theoretical lock order issue by using spin_lock_irqsave() instead of
spin_lock()
Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Impact: cleanup
node_to_cpumask (and the blecherous node_to_cpumask_ptr which
contained a declaration) are replaced now everyone implements
cpumask_of_node.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
The PCIe port driver calls pci_enable_device() during probe but
never calls pci_disable_device() during remove.
Cc: stable@kernel.org
Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Matthew Wilcox <willy@linux.intel.com>
Prakash's system needs MSI disabled on some bridges, but not all.
This seems to be the minimal fix for 2.6.29, but should be replaced
during 2.6.30.
Signed-off-by: Prakash Punnoor <prakash@punnoor.de>
Signed-off-by: Matthew Wilcox <willy@linux.intel.com>
The RPA PCI hotplug driver calls EEH routines, so should depend on
EEH. Also PPC_PSERIES implies PPC64, so remove that.
Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Matthew Wilcox <willy@linux.intel.com>
Commit 47a8b0cc (Enable PCIe AER only after checking firmware
support) wants to walk the PCI bus in the remove path to disable
AER, and calls pci_walk_bus for downstream bridges.
Unfortunately, in the remove path, we remove devices and bridges
in a depth-first manner, starting with the furthest downstream
bridge and working our way backwards.
The furthest downstream bridges will not have a dev->subordinate,
and we hit a NULL deref in pci_walk_bus.
Check for dev->subordinate first before attempting to walk the
PCI hierarchy below us.
Acked-by: Andrew Patterson <andrew.patterson@hp.com>
Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Matthew Wilcox <willy@linux.intel.com>
This patch is intended to disable L0s ASPM link state for 82598 (ixgbe)
parts due to the fact that it is possible to corrupt TX data when coming
back out of L0s on some systems. The workaround had been added for 82575
(igb) previously, but did not use the ASPM api. This quirk uses the ASPM
api to prevent the ASPM subsystem from re-enabling the L0s state.
Instead of adding the fix in igb to the ixgbe driver as well it was
decided to move it into a pci quirk. It is necessary to move the fix out
of the driver and into a pci quirk in order to prevent the issue from
occuring prior to driver load to handle the possibility of the device being
passed to a VM via direct assignment.
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
CC: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Matthew Wilcox <willy@linux.intel.com>
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jbarnes/pci-2.6:
PCI: AMD 813x B2 devices do not need boot interrupt quirk
PCI: Enable PCIe AER only after checking firmware support
PCI: pciehp: Handle interrupts that happen during initialization.
PCI: don't enable too many HT MSI mappings
PCI: add some sysfs ABI docs
PCI quirk: enable MSI on 8132
Turns out that the new AMD 813x devices do not need the
quirk_disable_amd_813x_boot_interrupt quirk to be run on them. If it
is, no interrupts are seen on the PCI-X adapter.
From: Stefan Assmann <sassmann@novell.com>
Reported-by: Jamie Wellnitz <Jamie.Wellnitz@emulex.com>
Tested-by: Jamie Wellnitz <Jamie.Wellnitz@emulex.com>
Cc: stable <stable@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Jesse Barnes <jbarnes@hobbes.lan>
The PCIe port driver currently sets the PCIe AER error reporting bits for
any root or switch port without first checking to see if firmware will grant
control. This patch moves setting these bits to the AER service driver
aer_enable_port routine. The bits are then set for the root port and any
downstream switch ports after the check for firmware support (aer_osc_setup)
is made. The patch also unsets the bits in a similar fashion when the AER
service driver is unloaded.
Reviewed-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Andrew Patterson <andrew.patterson@hp.com>
Signed-off-by: Jesse Barnes <jbarnes@hobbes.lan>
Move the enabling of interrupts after all of the data structures
are setup so that we can safely run the interrupt handler as
soon as it is registered.
Reviewed-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Tested-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Eric W. Biederman <ebiederm@aristanetworks.com>
Signed-off-by: Jesse Barnes <jbarnes@hobbes.lan>
Prakash reported that his c51-mcp51 ondie sound card doesn't work with
MSI. But if he hacks out the HT-MSI quirk, MSI works fine.
So this patch reworks the nv_msi_ht_cap_quirk(). It will now only
enable ht_msi on own its root device, avoiding enabling it on devices
following that root dev.
Reported-by: Prakash Punnoor <prakash@punnoor.de>
Tested-by: Prakash Punnoor <prakash@punnoor.de>
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Signed-off-by: Jesse Barnes <jbarnes@hobbes.lan>
David reported that LSI SAS doesn't work with MSI. It turns out that
his BIOS doesn't enable it, but the HT MSI 8132 does support HT MSI.
Add quirk to enable it
Cc: stable@kernel.org
Reported-by: David Lang <david@lang.hm>
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
This is the cause of the DMA faults and disk corruption that people have
been seeing. Some chipsets neglect to report the RWBF "capability" --
the flag which says that we need to flush the chipset write-buffer when
changing the DMA page tables, to ensure that the change is visible to
the IOMMU.
Override that bit on the affected chipsets, and everything is happy
again.
Thanks to Chris and Bhavesh and others for helping to debug.
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Tested-by: Chris Wright <chrisw@sous-sol.org>
Reviewed-by: Bhavesh Davda <bhavesh@vmware.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This is the cause of the DMA faults and disk corruption that people have
been seeing. Some chipsets neglect to report the RWBF "capability" --
the flag which says that we need to flush the chipset write-buffer when
changing the DMA page tables, to ensure that the change is visible to
the IOMMU.
Override that bit on the affected chipsets, and everything is happy
again.
Thanks to Chris and Bhavesh and others for helping to debug.
Should resolve:
https://bugzilla.redhat.com/show_bug.cgi?id=479996http://bugzilla.kernel.org/show_bug.cgi?id=12578
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Tested-and-acked-by: Chris Wright <chrisw@sous-sol.org>
Reviewed-by: Bhavesh Davda <bhavesh@vmware.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Fix pci kernel-doc parameter missing notation, correct
function name, and fix typo:
Warning(linux-2.6.28-git10//drivers/pci/pci.c:1511): No description found for parameter 'exclusive'
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Hidetoshi Seto points out that commit
bffac3c593 has wrong values in the array.
Rather than correct the array, we can just use a bounds check and
perform the calculation specified in the comment. As a bonus, this will
not run off the end of the array if the device specifies an illegal
value in the MSI capability.
Signed-off-by: Matthew Wilcox <willy@linux.intel.com>
Signed-off-by: Hidetoshi Seto <seto.hidetoshi@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
* 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
ptrace, x86: fix the usage of ptrace_fork()
i8327: fix outb() parameter order
x86: fix math_emu register frame access
x86: math_emu info cleanup
x86: include correct %gs in a.out core dump
x86, vmi: put a missing paravirt_release_pmd in pgd_dtor
x86: find nr_irqs_gsi with mp_ioapic_routing
x86: add clflush before monitor for Intel 7400 series
x86: disable intel_iommu support by default
x86: don't apply __supported_pte_mask to non-present ptes
x86: fix grammar in user-visible BIOS warning
x86/Kconfig.cpu: make Kconfig help readable in the console
x86, 64-bit: print DMI info in the oops trace
When hardware detects any error with a descriptor from the invalidation
queue, it stops fetching new descriptors from the queue until software
clears the Invalidation Queue Error bit in the Fault Status register.
Following fix handles the IQE so the kernel won't be trapped in an
infinite loop.
Signed-off-by: Yu Zhao <yu.zhao@intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
This fix should be safe since iommu->agaw is only used in intel-iommu.c.
And this file is only compiled with DMAR=y.
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Due to recurring issues with DMAR support on certain platforms.
There's a number of filesystem corruption incidents reported:
https://bugzilla.redhat.com/show_bug.cgi?id=479996http://bugzilla.kernel.org/show_bug.cgi?id=12578
Provide a Kconfig option to change whether it is enabled by
default.
If disabled, it can still be reenabled by passing intel_iommu=on to the
kernel. Keep the .config option off by default.
Signed-off-by: Kyle McMartin <kyle@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Acked-By: David Woodhouse <David.Woodhouse@intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Currently, the PM core always attempts to manage devices with drivers
that use the new PM framework. In particular, it attempts to disable
the devices (which is unnecessary), to save their state (which may be
undesirable if the driver has done that already) and to put them into
low power states (again, this may be undesirable if the driver has
already put the device into a low power state). That need not be
the right thing to do, so make the core be more careful in this
respect.
Generally, there are the following categories of devices to consider:
* bridge devices without drivers
* non-bridge devices without drivers
* bridge devices with drivers
* non-bridge devices with drivers
and each of them should be handled differently.
For bridge devices without drivers the PCI PM core will save their
state on suspend and restore it (early) during resume, after putting
them into D0 if necessary. It will not attempt to do anything else
to these devices.
For non-bridge devices without drivers the PCI PM core will disable
them and save their state on suspend. During resume, it will put
them into D0, if necessary, restore their state (early) and reenable
them.
For bridge devices with drivers the PCI PM core will only save
their state on suspend if the driver hasn't done that already.
Still, the core will restore their state (early) during resume,
after putting them into D0, if necessary.
For non-bridge devices with drivers the PCI PM core will only save
their state on suspend if the driver hasn't done that already. Also,
if the state of the device hasn't been saved by the driver, the core
will attempt to put the device into a low power state. During
resume the core will restore the state of the device (early), after
putting it into D0, if necessary.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
pci_restore_standard_config() unconditionally changes current_state
to PCI_D0 after attempting to change the device's power state, but
it should rather read the actual current power state from the
device.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
It is a mistake to disable and enable PCI bridges and PCI Express
ports during suspend-resume, at least at the time when it is
currently done. Disabling them may lead to problems with accessing
devices behind them and they should be automatically enabled when
their standard config spaces are restored. Fix this by not attempting
to disable bridges during suspend and enable them during resume.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Simplify suspend and resume of the PCI Express port driver. It no
longer needs to save and restore the standard configuration space of the
device; this is now done by the PCI PM core layer.
This patch is reported to fix the regression tracked as
http://bugzilla.kernel.org/show_bug.cgi?id=12598
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Reported-and-tested-by: Parag Warudkar <parag.lkml@gmail.com>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Make pci_legacy_suspend() save the state of the device if it is
in PCI_UNKNOWN after its suspend callback has run and warn only if
the power state of the device has been changed by its suspend
callback.
Also, use WARN_ONCE(), which is more useful, in pci_legacy_suspend(),
so that the name of the offending function is printed.
Additionally, remove the unnecessary line of code setting
pci_dev->state_saved.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Reported-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Check if the standard configuration registers of a PCI device have
been saved during suspend before trying to restore them during
resume.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Reported-By: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Suspend to RAM is reported to break on some machines as a result of
attempting to put one of driverless PCI devices into a low power
state. Avoid that by not attepmting to power manage driverless
devices during suspend.
Fix up pci_pm_poweroff() after a previous incomplete fix for the same
thing during hibernation.
This patch is reported to fix the regression from 2.6.28 tracked as
http://bugzilla.kernel.org/show_bug.cgi?id=12605
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Reported-and-tested-by: Eric Sesterhenn <snakebyte@gmx.de>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
This patch makes the ROM reading code return an error to user space if
the size of the ROM read is equal to 0.
The patch also emits a warnings if the contents of the ROM are invalid,
and documents the effects of the "enable" file on ROM reading.
Signed-off-by: Timothy S. Nelson <wayland@wayland.id.au>
Acked-by: Alex Villacis-Lasso <a_villacis@palosanto.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
We only want to disable ASPM when the last function is removed from
the parent's device list. We determine this by checking to see if
the parent's device list is completely empty.
Unfortunately, we never hit that code because the parent is considered
an upstream port, and never had an ASPM link_state associated with it.
The early check for !link_state causes us to return early, we never
discover that our device list is empty, and thus we never remove the
downstream ports' link_state nodes.
Instead of checking to see if the parent's device list is empty, we can
check to see if we are the last device on the list, and if so, then we
know that we can clean up properly.
Cc: Shaohua Li <shaohua.li@intel.com>
Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
The dma ops unification enables X86 and IA64 to share intel_dma_ops so
we can make dma mapping functions static. This also remove unused
intel_map_single().
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
dma_mapping_error is used to see if dma_map_single and dma_map_page
succeed. IA64 VT-d dma_mapping_error always says that dma_map_single
is successful even though it could fail. Note that X86 VT-d works
properly in this regard.
This patch fixes IA64 VT-d dma_mapping_error by adding VT-d's own
dma_mapping_error() that works for both X86_64 and IA64. VT-d uses
zero as an error dma address so VT-d's dma_mapping_error returns 1 if
a passed dma address is zero (as x86's VT-d dma_mapping_error does
now).
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Some hardware exposes PCIE slots in such a way that they can be claimed
by either the acpiphp or pciehp driver. pciehp is the preferred driver
if the firmware allows the OS to claim control via the _OSC method so
should be loaded first - if it fails to bind (either due to a missing
_OSC method or the firmware refusing to hand off control) then we can
fall back to acpiphp or a vendor-specific driver.
This patch simply changes the link order to ensure that pciehp will be
initialised before acpiphp if both are statically built into the kernel.
Signed-off-by: Matthew Garrett <mjg@redhat.com>
Acked-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
For PCI devices, pci_bus_assign_resources() must be called to set up the
pci_device->resource array before pci_bus_add_devices() can be called, else
attempts to load drivers results in BAR collision errors where there are none.
This is not done in fakephp, so devices can be "unplugged" but scanning the
parent bus won't bring the devices back due to resource unallocation. Move the
pci_bus_add_device-calling logic into pci_rescan_bus and preface it with a call
to pci_bus_assign_resources so that we only have to (re)allocate resources once
per bus where a new device is found.
Signed-off-by: Darrick J. Wong <djwong@us.ibm.com>
Acked-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Add an msi_mask() function which returns the correct bitmask for the
number of MSI interrupts you have. This fixes an undefined bug in
msi_capability_init().
Signed-off-by: Matthew Wilcox <willy@linux.intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
pci_restore_standard_config() adds extra delay for PCI buses in
low power states (B2 or B3), but this is only correct for buses in
B2, because the buses in B3 are reset when they are put back into
B0. Thus we should wait for such buses to settle after the reset,
but it's not a good idea to wait that long (1.1 s) with interrupts
off.
On the other hand, we have never waited for buses in B2 and B3
during resume and it seems reasonable to go back to this well
tested behaviour.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Devices that have MSI-X enabled before suspend to RAM or hibernation
and that are in a low power state during resume will not be handled
correctly by pci_restore_standard_config(). Namely, it first calls
pci_restore_state() which calls pci_restore_msi_state(), which in turn
executes __pci_restore_msix_state() that accesses the device's memory
space to restore the contents of the MSI-X table. However, if the
device is in a low power state at this point, it's memory space is
not accessible.
The easiest way to fix this potential problem is to make
pci_restore_standard_config() call pci_restore_state() after
it has put the device into the full power state, D0. Fortunately,
all of this is done with interrupts off, so the change of ordering
should not cause any trouble.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Hibernation breaks on EeePC 701 as a result of attempting to put one
of its (driverless) devices into a low power state. Avoid that by
not attepmting to power manage driverless devices during hibernation.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Reported-and-tested-by: Alan Jenkins <alan-jenkins@tuffmail.co.uk>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
If one of device drivers refuses to suspend by returning error code
from its ->suspend() callback, the devices that have already been
suspended are resumed by executing their drivers' ->resume()
callbacks. Some of these callbacks expect the device's
configuration space to be restored if the device has been put into
D3 before they are called. Unfortunately, this mechanism has been
broken by recent changes moving the restoration of config spaces
of some devices (most importantly, USB controllers and HDA Intel)
into the resume callbacks executed with interrupts off. Obviously,
these callbacks are not invoked in the suspend error path and, as a
result, the system cannot be successfully brought back into the
working state in case of a suspend error. The same thing happens
in the hibernation error path right before putting the system into
S4.
Similarly, the suspend testing facility associated with the
/sys/power/pm_test file is broken, because it uses the very same
mechanism that is used in the suspend and hibernation error paths.
Fix the breakage by making the PCI core restore the configuration
spaces of PCI devices that haven't been restored already before
pci_pm_resume() is called for those devices by the PM core.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jbarnes/pci-2.6:
PCI hotplug: fix lock imbalance in pciehp
PCI PM: Restore standard config registers of all devices early
PCI/MSI: bugfix/utilize for msi_capability_init()
set_lock_status omits mutex_unlock in fail path. Add the omitted
unlock.
As a result a lockup caused by this can be triggered from userspace
by writing 1 to /sys/bus/pci/slots/.../lock often enough.
Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Reviewed-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
There is a problem in our handling of suspend-resume of PCI devices that
many of them have their standard config registers restored with
interrupts enabled and they are put into the full power state with
interrupts enabled as well. This may lead to the following scenario:
* an interrupt vector is shared between two or more devices
* one device is resumed earlier and generates an interrupt
* the interrupt handler of another device tries to handle it and
attempts to access the device the config space of which hasn't been
restored yet and/or which still is in a low power state
* the system crashes as a result
To prevent this from happening we should restore the standard
configuration registers of all devices with interrupts disabled and we
should put them into the D0 power state right after that.
Unfortunately, this cannot be done using the existing
pci_set_power_state(), because it can sleep. Also, to do it we have to
make sure that the config spaces of all devices were actually saved
during suspend.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
ACPI hotplug panic with current git head
http://lkml.org/lkml/2009/1/10/136
Rather than reverting the entire commit that causes the crash:
e8c331e963
"PCI hotplug: introduce functions for ACPI slot detection"
simply harden against it while the changes to
the hotplug code on this particularl machine are understood.
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Acked-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Len Brown <len.brown@intel.com>
This reverts commit 98e6e286d7, as Yinghai
Lu reports that it breaks kexec with at least the e1000 and e1000e
drivers. The reason is that the shutdown sequence puts the hardware
into D3 sleep, and the commit causes us to claim that it then is in D0
(running) state just because we don't understand the PM capabilities.
Which then later makes "pci_set_power_state()" not do anything, and the
device never wakes up properly and just returns 0xff to everything.
Reported-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: From: Rafael J. Wysocki <rjw@sisk.pl>
Cc: Jesse Barnes <jesse.barnes@intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
With some broken BIOSs when VT-d is enabled, the data structures are
filled incorrectly. This can cause a NULL pointer dereference in very
early boot.
Signed-off-by: Dirk Hohndel <hohndel@linux.intel.com>
Acked-by: Yu Zhao <yu.zhao@intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Impact: clean up sparseirq fallout on random.c
Ingo suggested to change some ifdef from SPARSE_IRQ to GENERIC_HARDIRQS
so we could some #ifdef later if all arch support genirq
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: Matt Mackall <mpm@selenic.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
* 'cpus4096-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
[IA64] fix typo in cpumask_of_pcibus()
x86: fix x86_32 builds for summit and es7000 arch's
cpumask: use work_on_cpu in acpi-cpufreq.c for read_measured_perf_ctrs
cpumask: use work_on_cpu in acpi-cpufreq.c for drv_read and drv_write
cpumask: use cpumask_var_t in acpi-cpufreq.c
cpumask: use work_on_cpu in acpi/cstate.c
cpumask: convert struct cpufreq_policy to cpumask_var_t
cpumask: replace CPUMASK_ALLOC etc with cpumask_var_t
x86: cleanup remaining cpumask_t ops in smpboot code
cpumask: update pci_bus_show_cpuaffinity to use new cpumask API
cpumask: update local_cpus_show to use new cpumask API
ia64: cpumask fix for is_affinity_mask_valid()
Put PM callbacks in drivers/pci/pci-driver.c in the order in which
they are executed which makes it much easier to follow the code.
No functional changes should result from this.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Pavel Machek <pavel@suse.cz>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
It should be quite clear that it generally makes sense to execute
the default PM callbacks (ie. the callbacks used for handling
suspend, hibernation and resume of PCI devices without drivers) for
all devices. Of course, the drivers that provide legacy PCI PM
support (ie. the ->suspend, ->suspend_late, ->resume_early
or ->resume hooks in the pci_driver structure), carry out these
operations too, so we can't do it for devices with such drivers.
Still, we can make the default PM callbacks run for devices with
drivers using the new framework (ie. implement the pm object), since
there are no such drivers at the moment.
This also simplifies the code and makes it smaller.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Pavel Machek <pavel@suse.cz>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Use the observation that the power state of a PCI device can be
loaded into its pci_dev structure as soon as pci_pm_init() is run for
it and make that happen.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Pavel Machek <pavel@suse.cz>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
The size of drivers/pci/pci-driver.c can be reduced quite a bit
if pci_fixup_device() is called from the legacy PM callbacks, so make
it happen.
No functional changes should result from this.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Pavel Machek <pavel@suse.cz>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Rename two functions and rearrange code in drivers/pci/pci-driver.c
so that it's easier to follow. In particular, separate invocations
of the legacy callbacks from the rest of the new callbacks' code.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
It generally is better to avoid accessing devices behind bridges that
may not be in the D0 power state, because in that case the bridges'
secondary buses may not be accessible. For this reason, during the
early phase of resume (ie. with interrupts disabled), before
restoring the standard config registers of a device, check the power
state of the bridge the device is behind and postpone the restoration
of the device's config space, as well as any other operations that
would involve accessing the device, if that state is not D0.
In such cases the restoration of the device's config space will be
retried during the "normal" phase of resume (ie. with interrupts
enabled), so that the bridge can be put into D0 before that happens.
Also, save standard configuration registers of PCI devices during the
"normal" phase of suspend (ie. with interrupts enabled), so that the
bridges the devices are behind can be put into low power states (we
don't put bridges into low power states at the moment, but we may
want to do it in the future and it seems reasonable to design for
that).
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Move pci_has_legacy_pm_support() closer to the functions that
call it.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Pavel Machek <pavel@suse.cz>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
PCI devices without drivers can be put into low power states during
suspend with the help of pci_prepare_to_sleep() and prevented from
generating wake-up events during resume with the help of
pci_enable_wake(). However, it's better not to put bridges into
low power states during suspend, because that might result in entire
bus segments being powered off.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Pavel Machek <pavel@suse.cz>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
PCI devices without drivers are not disabled during suspend and
hibernation, but they are enabled during resume, with the help of
pci_reenable_device(), so there is an unbalanced execution of
pcibios_enable_device() in the resume code path.
To correct this introduce function pci_disable_enabled_device()
that will disable the argument device, if it is enabled when the
function is being run, without updating the device's pci_dev
structure and use it in the suspend code path to balance the
pci_reenable_device() executed during resume.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Pavel Machek <pavel@suse.cz>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
pci_fixup_device() is called too early in pci_pm_poweroff() and too
late in pci_pm_restore(). Moreover, pci_pm_restore_noirq() calls
pci_fixup_device() twice and in a wrong way. Fix that.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Pavel Machek <pavel@suse.cz>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
The cpu_relax() function can be a noop on certain architectures like
IA-64 when CPU threads are disabled, so use msleep instead during link
retraining busy/wait loop.
Introduce define LINK_RETRAIN_TIMEOUT instead of hard-coding timeout in
pcie_aspm_configure_common_clock.
Use time_after() to avoid jiffy wraparound when checking for expired
timeout.
After timeout expires, recheck link status register link training bit
instead of checking for expired timeout to avoid possible false
positive.
Note that Matthew Wilcox came up with the first rough version of this
patch.
Reviewed-by: Matthew Wilcox <willy@linux.intel.com>
Signed-off-by: Andrew Patterson <andrew.patterson@hp.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Add kerneldoc comments to the reamining functions in
drivers/pci/pcie/portdrv_core.c .
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Rearrange code in drivers/pci/pcie/portdrv_bus.c and
drivers/pci/pcie/portdrv_core.c so that related functions and data
structures are closer together.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
There is a problem with the suspend and resume of PCI Express port
service devices that the ->suspend() and ->resume() routines of each
service device are called twice in each suspend-resume cycle, which
is obviously wrong.
The scenario is that first, the PCI Express port driver calls
suspend and resume routines of each port service driver from its
pcie_portdrv_suspend() and pcie_portdrv_resume() callbacks,
respectively (which is correct), and second, the pcie_port_bus_type
driver calls them from its ->suspend() and ->resume() callbacks
(which is not correct, because it doesn't happen at the right time).
The solution is to remove the ->suspend() and ->resume() callbacks
from pcie_port_bus_type and the associated functions.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Add kerneldoc comments to some functions in
drivers/pci/pcie/portdrv_core.c, since the code in there is not
easy to follow without any additional description.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
During an online device reset it may be useful to disable bus-mastering.
pci_disable_device() does that, and far more besides, so is not suitable
for an online reset.
Add pci_clear_master() which does just this.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Reviewed-by: Matthew Wilcox <willy@linux.intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
func is checked not to be NULL a few lines before.
A simplified version of the semantic patch that makes this change is as
follows: (http://www.emn.fr/x-info/coccinelle/)
// <smpl>
@r exists@
local idexpression x;
expression E;
position p1,p2;
@@
if (x@p1 == NULL || ...) { ... when forall
return ...; }
... when != \(x=E\|x--\|x++\|--x\|++x\|x-=E\|x+=E\|x|=E\|x&=E\|&x\)
(
x@p2 == NULL
|
x@p2 != NULL
)
// another path to the test that is not through p1?
@s exists@
local idexpression r.x;
position r.p1,r.p2;
@@
... when != x@p1
(
x@p2 == NULL
|
x@p2 != NULL
)
@fix depends on !s@
position r.p1,r.p2;
expression x,E;
statement S1,S2;
@@
(
- if ((x@p2 != NULL) || ...)
S1
|
- if ((x@p2 == NULL) && ...) S1
|
- BUG_ON(x@p2 == NULL);
)
// </smpl>
Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Clean up register definitions related to PCI Express Hot plug.
- Add register definitions into include/linux/pci_regs.h, and use
them instead of pciehp's locally definied register definitions.
- Remove pciehp's locally defined register definitions
- Remove unused register definitions in pciehp.
- Some minor cleanups.
Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Bit 10 in Link Status register used to be defined as Training Error in
the PCI Express 1.0a specification. But it was removed by Training Error
ECN and is no longer defined. So pciehp must ignore the value read from
it.
Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
In a PCIe hierarchy with a switch present, if the link state of an
endpoint device is changed, we must check the whole hierarchy from the
endpoint device to root port, and for each link in the hierarchy, the new
link state should be configured. Previously, the implementation checked
the state but forgot to configure the links between root port to switch.
Fixes Novell bz #448987.
Signed-off-by: Shaohua Li <shaohua.li@intel.com>
Tested-by: Andrew Patterson <andrew.patterson@hp.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Since pci_bus has a struct device, use dev_printk directly instead
of faking it by hand.
Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
The VPD on all devices may not be 32K. Unfortunately, there is no
generic way to find the size, so this adds a simple API hook
to reset it.
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Change PCI VPD API which was only used by sysfs to something usable
in drivers.
* move iteration over multiple words to the low level
* use conventional types for arguments
* add exportable wrapper
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Accessing the VPD area can take a long time. The existing
VPD access code fails consistently on my hardware. There are comments
in the SysKonnect vendor driver that it can take up to 13ms per word.
Change the access routines to:
* use a mutex rather than spinning with IRQ's disabled and lock held
* have a much longer timeout
* call cond_resched while spinning
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Reviewed-by: Matthew Wilcox <willy@linux.intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
This patch adds pci_common_swizzle(), which swizzles INTx values all the
way up to a root bridge.
This common implementation can replace several architecture-specific
ones. This should someday be combined with pci_get_interrupt_pin(),
but I left it separate for now to make reviewing easier.
Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Some ACPI related PCI hotplug code can be shared among PCI hotplug
drivers. This patch introduces the following functions in
drivers/pci/hotplug/acpi_pcihp.c to share the code, and changes
acpiphp and pciehp to use them.
- int acpi_pci_detect_ejectable(struct pci_bus *pbus)
This checks if the specified PCI bus has ejectable slots.
- int acpi_pci_check_ejectable(struct pci_bus *pbus, acpi_handle handle)
This checks if the specified handle is ejectable ACPI PCI slot. The
'pbus' parameter is needed to check if 'handle' is PCI related ACPI
object.
This patch also introduces the following inline function in
include/linux/pci-acpi.h, which is useful to get ACPI handle of the
PCI bridge from struct pci_bus of the bridge's secondary bus.
- static inline acpi_handle acpi_pci_get_bridge_handle(struct pci_bus *pbus)
This returns ACPI handle of the PCI bridge which generates PCI bus
specified by 'pbus'.
Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
ACPI based hot-pluggable PCIe slot detection logic was added to
prevent the problem non hot-pluggable PCIe slot was detected as
hot-pluggable. The slot detection logic can be selected through
'pciehp_detect_mode', but it would be better if it is selected
automatically.
This patch adds 'auto' option for 'pciehp_detect_mode'. When it is
specified, pciehp judges which 'acpi' or 'pcie' should be used. It
seems that the physical slot number is duplicated among some slots on
most of the platforms with the above-mentioned problem. So 'auto' mode
uses this information to judge which 'acpi' or 'pcie' should be
used. That is, if duplicated physical slot numbers are detected,
'acpi' mode is used. This method is not perfect, but it's realistic.
Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
There is a problem that some non hot-pluggable PCIe slots are detected
as hot-pluggable by pciehp on some platforms. The immediate cause of
this problem is that hot-plug capable bit in the Slot Capabilities
register is set even for non hot-pluggable slots on those platforms.
It seems a BIOS/hardware problem, but we need workaround about that.
Some of those platforms define hot-pluggable PCIe slots on ACPI
namespace properly, while hot-plug capable bit in the Slot
Capabilities register is set improperly. So using ACPI namespace
information in pciehp to detect PCIe hot-pluggable slots would be a
workaround.
This patch adds 'pciehp_detect_mode' module option. When 'acpi' is
specified, pciehp uses ACPI namespace information to detect PCIe
hot-pluggable slots.
Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
This uses work_on_cpu(), rather than altering the cpumask of the
thread which we happen to be.
Note the cleanups:
1) I've removed the CONFIG_NUMA test, since dev_to_node() returns -1
for !CONFIG_NUMA anyway and the compiler will eliminate it.
2) No need to reset mempolicy to default (a bad idea anyway) since
work_on_cpu is run from a workqueue.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Suspend-resume of PCI Express ports has recently been moved into
_suspend_late() and _resume_early() callbacks, but some functions
executed from there should not be called with interrupts disabled,
eg. pci_enable_device(). For this reason, split the suspend-resume
of PCI Express ports into parts to be executed with interrupts
disabled and with interrupts enabled.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Currently, PCI devices without the PM capability that are power
manageable by the platform (eg. ACPI) are not handled correctly
by pci_set_power_state(), because their current_state field is not
updated to reflect the new power state of the device. Fix this by
making pci_update_current_state() accept additional argument
representing the power state of the device as set by the platform.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
When PCI devices are initialized, we check whether they support PCI PM
caps and set the device can_wakeup flag if so. However, some devices
may have platform provided wakeup events rather than PCI PME signals, so
we need to set can_wakeup in that case too. Doing so should allow
wakeups from many more devices, especially on cost constrained systems.
Reported-by: Alan Stern <stern@rowland.harvard.edu>
Tested-by: Joseph Chan <JosephChan@via.com.tw>
Acked-by: "Rafael J. Wysocki" <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
This patch splits a new function, pci_bus_add_child(), from
pci_bus_add_devices(). The new function can be used to register PCI
buses to the device core.
Signed-off-by: Yu Zhao <yu.zhao@intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Cleanup pci_bus_add_devices() by negating the conditional and
continuing, rather than having a single conditional take up the whole
body.
Signed-off-by: Yu Zhao <yu.zhao@intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Add a function to map a given resource number to a corresponding
register so drivers can get the offset and type of device specific BARs.
Signed-off-by: Yu Zhao <yu.zhao@intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Allow pci_alloc_child_bus() to allocate buses without bridge devices.
Some SR-IOV devices can occupy more than one bus number, but there is no
explicit bridges because that have internal routing mechanism.
Signed-off-by: Yu Zhao <yu.zhao@intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Export __pci_read_base() so it can be used by whole PCI subsystem.
Signed-off-by: Yu Zhao <yu.zhao@intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Remove the unnecessary number of resources condition checks because
the pci_update_resource() will check availability of the resources.
Signed-off-by: Yu Zhao <yu.zhao@intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
This patch moves all definitions of the PCI resource names to an 'enum',
and also replaces some hard-coded resource variables with symbol
names. This change eases introduction of device specific resources.
Reviewed-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Yu Zhao <yu.zhao@intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Eric W. Biederman
Ivan Kokshaysky
Andrew Morton
Michael Ellerman
Roel Kluin
Kenji Kaneshige
Sheng Yang
Jiri Slaby
Alex Chiang
Julia Lawall
Frank Seidel
Yu Zhao
Rafael J. Wysocki
Kay Sievers
Harvey Harrison
Ingo Molnar
Suresh Siddha
Rusty Russell
Prakash Punnoor
Alexander Duyck
Linus Torvalds
Stefan Assmann
Andrew Patterson
Yinghai Lu
David Woodhouse
Tony Battersby
Randy Dunlap
Matthew Wilcox
Joerg Roedel
Kyle McMartin
Timothy S. Nelson
FUJITA Tomonori
Matthew Garrett
Darrick J. Wong
Matthew Wilcox
Hidetoshi Seto
James Bottomley
Heiko Carstens
Dirk Hohndel
Len Brown
Jaswinder Singh Rajput
Ben Hutchings
Shaohua Li
Bjorn Helgaas
Stephen Hemminger
Jesse Barnes