Age | Commit message (Collapse) | Author |
|
The generic PCI configuration space accessors are globally serialized via
pci_lock. On larger systems this causes massive lock contention when the
configuration space has to be accessed frequently. One such access pattern
is the Intel Uncore performance counter unit.
Provide a kernel config option which can be selected by an architecture
when the low level PCI configuration space accessors in the architecture
use their own serialization or can operate completely lockless.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Bjorn Helgaas <helgaas@kernel.org>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: linux-pci@vger.kernel.org
Link: http://lkml.kernel.org/r/20170316215057.205961140@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
|
|
Callers normally treat the config space accessors as returning PCBIOS_*
error codes, not Linux error codes (or they don't look at them at all). We
have pcibios_err_to_errno() in case the error code needs to be translated.
Fixes: 4b1038834739 ("PCI: Don't attempt config access to disconnected devices")
Signed-off-by: Brian Norris <briannorris@chromium.org>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Keith Busch <keith.busch@intel.com>
|
|
* pci/misc:
PCI: Change pci_host_common_probe() visibility
PCI: Fix typo pci_cfg_access_lock() comment
PCI: Include pci.h for struct pci_ops definition
|
|
A PCI/PCI-X to PCI Express bridge, sometimes referred to as a "reverse
bridge", is a bridge with conventional PCI or PCI-X on its primary side and
a PCI Express Port on its secondary (downstream) side.
That PCIe Port is a Downstream Port and could be connected to a slot, just
like a Root Port or a Switch Downstream Port. Make pcie_downstream_port()
return true for them, so we can access the Slot registers in the PCIe
capability.
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
|
|
There is no pci_cfg_access_unlocked(). I think the author meant
pci_cfg_access_unlock().
Signed-off-by: Brian Norris <briannorris@chromium.org>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
|
|
If we've detected the PCI device is disconnected, there is no need to
attempt to access its config space since we know the operation will fail.
Make all the config reads and writes return -ENODEV error immediately when
in such a state.
If a caller requests a config read to a disconnected device, return a data
value of all 1's. This is the same as what hardware is expected to return
when accessing a removed device, but software can do this faster without
relying on hardware.
Tested-by: Krishna Dhulipala <krishnad@fb.com>
Signed-off-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Wei Zhang <wzhang@fb.com>
|
|
Replace the inline PCI device config read and write accessors with exported
functions. This is preparing for these functions to make use of private
data.
Tested-by: Krishna Dhulipala <krishnad@fb.com>
Signed-off-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Wei Zhang <wzhang@fb.com>
|
|
<linux/sched.h> into <linux/sched/signal.h>
Fix up affected files that include this signal functionality via sched.h.
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
|
|
* pci/vpd:
PCI: Increase VPD access timeout to 125ms
|
|
The PCI core uses a fixed 50ms timeout when waiting for VPD accesses to
complete. When an access does not complete within this period, a warning
is logged and an error returned to the caller.
While this default timeout is valid for most hardware, some devices can
experience longer access delays under certain circumstances. For example,
one of the IBM CXL Flash devices can take up to ~120ms in a worst-case
scenario. These types of devices can benefit from an extended timeout.
To support devices with a longer access delay, increase the timeout in
pci_vpd_wait() to 125ms. The PCI specification is silent with respect to
VPD delays, therefore there is no concern for violating a threshold.
Tested-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>
Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
|
|
pci_lock is an IRQ-safe spinlock that protects all accesses to PCI
configuration space (see PCI_OP_READ() and PCI_OP_WRITE() in pci/access.c).
The pci_cfg_access_unlock() path acquires pci_lock, then p->pi_lock (inside
wake_up_all()). According to lockdep, there is a possible path involving
snbep_uncore_pci_read_counter() that could acquire them in the reverse
order: acquiring p->pi_lock, then pci_lock, which could result in a
deadlock. Lockdep details are in the bugzilla below.
Avoid the possible deadlock by dropping pci_lock before waking up any
config access waiters.
Link: https://bugzilla.kernel.org/show_bug.cgi?id=192901
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
|
|
Hardware that supports only 32-bit config writes is not spec-compliant.
For example, if software performs a 16-bit write, we must do a 32-bit read,
merge in the 16 bits we intend to write, followed by a 32-bit write. If
the 16 bits we *don't* intend to write happen to have any RW1C (write-one-
to-clear) bits set, we just inadvertently cleared something we shouldn't
have.
Add a rate-limited warning when we do sub-32 bit config writes. Remove
similar probe-time warnings from some of the affected host bridge drivers.
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Enthusiastically-Acked-by: Russell King <rmk+kernel@armlinux.org.uk>
Acked-by: Shawn Lin <shawn.lin@rock-chips.com> # rockchip
Acked-by: Thierry Reding <treding@nvidia.com>
|
|
After 104daa71b396 ("PCI: Determine actual VPD size on first access"), the
PCI core computes the valid VPD size by parsing the VPD starting at offset
0x0. We don't attempt to read past that valid size because that causes
some devices to crash.
However, some devices do have data past that valid size. For example,
Chelsio adapters contain two VPD structures, and the driver needs both of
them.
Add pci_set_vpd_size(). If a driver knows it is safe to read past the end
of the VPD data structure at offset 0, it can use pci_set_vpd_size() to
allow access to as much data as it needs.
[bhelgaas: changelog, split patches, rename to pci_set_vpd_size() and
return int (not ssize_t)]
Fixes: 104daa71b396 ("PCI: Determine actual VPD size on first access")
Tested-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Casey Leedom <leedom@chelsio.com>
Signed-off-by: Hariprasad Shenai <hariprasad@chelsio.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
|
|
Use usleep_range() instead of udelay() while waiting for a VPD access to
complete. This is not a performance path, so no need to hog the CPU.
Rationale for usleep_range() parameters:
We clear PCI_VPD_ADDR_F for a read (or set it for a write), then wait for
the device to change it. For a device that updates PCI_VPD_ADDR between
our config write and subsequent config read, we won't sleep at all and
can get the device's maximum rate.
Sleeping a minimum of 10 usec per 4-byte access limits throughput to
about 400Kbytes/second. VPD is small (32K bytes at most), and most
devices use only a fraction of that.
We back off exponentially up to 1024 usec per iteration. If we reach
1024, we've already waited up to 1008 usec (16 + 32 + ... + 512), so if
we miss an update and wait an extra 1024 usec, we can still get about
1/2 of the device's maximum rate.
Tested-by: Shane Seymour <shane.seymour@hpe.com>
Tested-by: Babu Moger <babu.moger@oracle.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
|
|
We only support one flavor of VPD, so there's no need to complicate things
by having a "generic" struct pci_vpd and a more specific struct
pci_vpd_pci22.
Fold struct pci_vpd_pci22 directly into struct pci_vpd.
[bhelgaas: remove NULL check before kfree of dev->vpd (per kfreeaddr.cocci)]
Tested-by: Shane Seymour <shane.seymour@hpe.com>
Tested-by: Babu Moger <babu.moger@oracle.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
|
|
There's only one kind of VPD, so we don't need to qualify it as "the
version described by PCI spec rev 2.2."
Rename the following symbols to remove unnecessary "pci22":
PCI_VPD_PCI22_SIZE -> PCI_VPD_MAX_SIZE
pci_vpd_pci22_size() -> pci_vpd_size()
pci_vpd_pci22_wait() -> pci_vpd_wait()
pci_vpd_pci22_read() -> pci_vpd_read()
pci_vpd_pci22_write() -> pci_vpd_write()
pci_vpd_pci22_ops -> pci_vpd_ops
pci_vpd_pci22_init() -> pci_vpd_init()
Tested-by: Shane Seymour <shane.seymour@hpe.com>
Tested-by: Babu Moger <babu.moger@oracle.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
|
|
The struct pci_vpd_ops.release function pointer is always
pci_vpd_pci22_release(), so there's no need for the flexibility of a
function pointer.
Inline the pci_vpd_pci22_release() body into pci_vpd_release() and remove
pci_vpd_pci22_release() and the struct pci_vpd_ops.release function
pointer.
Tested-by: Shane Seymour <shane.seymour@hpe.com>
Tested-by: Babu Moger <babu.moger@oracle.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
|
|
Move pci_vpd_release() so it's next to the other VPD functions. This puts
it next to pci_vpd_pci22_init(), which allocates the space freed by
pci_vpd_release().
Tested-by: Shane Seymour <shane.seymour@hpe.com>
Tested-by: Babu Moger <babu.moger@oracle.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
|
|
pci_read_vpd() and pci_write_vpd() were stranded in the middle of config
accessor functions. Move them close to the other VPD code in the file.
No functional change.
Tested-by: Shane Seymour <shane.seymour@hpe.com>
Tested-by: Babu Moger <babu.moger@oracle.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
|
|
PCI-2.2 VPD entries have a maximum size of 32k, but might actually be
smaller than that. To figure out the actual size one has to read the VPD
area until the 'end marker' is reached.
Per spec, reading outside of the VPD space is "not allowed." In practice,
it may cause simple read errors or even crash the card. To make matters
worse not every PCI card implements this properly, leaving us with no 'end'
marker or even completely invalid data.
Try to determine the size of the VPD data when it's first accessed. If no
valid data can be read an I/O error will be returned when reading or
writing the sysfs attribute.
As the amount of VPD data is unknown initially the size of the sysfs
attribute will always be set to '0'.
[bhelgaas: changelog, use 0/1 (not false/true) for bitfield, tweak
pci_vpd_pci22_read() error checking]
Tested-by: Shane Seymour <shane.seymour@hpe.com>
Tested-by: Babu Moger <babu.moger@oracle.com>
Signed-off-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Cc: Alexander Duyck <alexander.duyck@gmail.com>
|
|
Make struct pci_vpd_pci22.busy a 1-bit field instead of a bool. We intend
to add another flag, and two bitfields are cheaper than two bools.
Tested-by: Shane Seymour <shane.seymour@hpe.com>
Tested-by: Babu Moger <babu.moger@oracle.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
|
|
Fix all whitespace issues (missing or needed whitespace) in all files in
drivers/pci. Code is compiled with allyesconfig before and after code
changes and objects are recorded and checked with objdiff and they are not
changed after this commit.
Signed-off-by: Bogicevic Sasa <brutallesale@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
|
|
932c435caba8 ("PCI: Add dev_flags bit to access VPD through function 0")
added PCI_DEV_FLAGS_VPD_REF_F0. Previously, we set the flag on every
non-zero function of quirked devices. If a function turned out to be
different from function 0, i.e., it had a different class, vendor ID, or
device ID, the flag remained set but we didn't make VPD accessible at all.
Flip this around so we only set PCI_DEV_FLAGS_VPD_REF_F0 for functions that
are identical to function 0, and allow regular VPD access for any other
functions.
[bhelgaas: changelog, stable tag]
Fixes: 932c435caba8 ("PCI: Add dev_flags bit to access VPD through function 0")
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Bjorn Helgaas <helgaas@kernel.org>
Acked-by: Myron Stowe <myron.stowe@redhat.com>
Acked-by: Mark Rustad <mark.d.rustad@intel.com>
CC: stable@vger.kernel.org
|
|
Commit 932c435caba8 ("PCI: Add dev_flags bit to access VPD through function
0") passes PCI_SLOT(devfn) for the devfn parameter of pci_get_slot().
Generally this works because we're fairly well guaranteed that a PCIe
device is at slot address 0, but for the general case, including
conventional PCI, it's incorrect. We need to get the slot and then convert
it back into a devfn.
Fixes: 932c435caba8 ("PCI: Add dev_flags bit to access VPD through function 0")
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Bjorn Helgaas <helgaas@kernel.org>
Acked-by: Myron Stowe <myron.stowe@redhat.com>
Acked-by: Mark Rustad <mark.d.rustad@intel.com>
CC: stable@vger.kernel.org
|
|
Add a dev_flags bit, PCI_DEV_FLAGS_VPD_REF_F0, to access VPD through
function 0 to provide VPD access on other functions. This is for hardware
devices that provide copies of the same VPD capability registers in
multiple functions. Because the kernel expects that each function has its
own registers, both the locking and the state tracking are affected by VPD
accesses to different functions.
On such devices for example, if a VPD write is performed on function 0,
*any* later attempt to read VPD from any other function of that device will
hang. This has to do with how the kernel tracks the expected value of the
F bit per function.
Concurrent accesses to different functions of the same device can not only
hang but also corrupt both read and write VPD data.
When hangs occur, typically the error message:
vpd r/w failed. This is likely a firmware bug on this device.
will be seen.
Never set this bit on function 0 or there will be an infinite recursion.
Signed-off-by: Mark Rustad <mark.d.rustad@intel.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-by: Alexander Duyck <alexander.h.duyck@redhat.com>
CC: stable@vger.kernel.org
|
|
As used in the PCIe spec, "Downstream Port" includes both Root Ports and
Switch Downstream Ports. We sometimes checked for PCI_EXP_TYPE_DOWNSTREAM
when we should have checked for PCI_EXP_TYPE_ROOT_PORT or
PCI_EXP_TYPE_DOWNSTREAM.
For a Root Port without a slot, the effect of this was that using
pcie_capability_read_word() to read PCI_EXP_SLTSTA returned zero instead of
showing the Presence Detect State bit hardwired to one as the PCIe Spec,
r3.0, sec 7.8, requires. (This read is completed in software because
previous PCIe spec versions didn't require PCI_EXP_SLTSTA to exist at all.)
Nothing in the kernel currently depends on this (pciehp only reads
PCI_EXP_SLTSTA on ports with slots), so this is a cleanup and not a
functional change.
Add a pcie_downstream_port() helper function and use it.
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
|
|
Many PCI controllers' configuration space accesses are memory-mapped and
vary only in address calculation and access checks. There are 2 main
access methods: a decoded address space such as ECAM or a single address
and data register similar to x86. This implementation can support both
cases as well as be used in cases that need additional pre- or post-access
handling.
Add a new pci_ops member, map_bus, which can do access checks and any
necessary setup. It returns the address to use for the configuration space
access. The access types supported are 32-bit only accesses or correct
byte, word, or dword sized accesses.
Tested-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Rob Herring <robh@kernel.org>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Thierry Reding <treding@nvidia.com>
|
|
Previously we applied _HPX type 2 record Link Control register settings
only to bridges with a subordinate bus. But it's better to apply them to
all devices with a link because if the subordinate bus has not been
allocated yet, we won't apply settings to the device.
Use pcie_cap_has_lnkctl() to determine whether the device has a Link
Control register instead of looking at dev->subordinate.
[bhelgaas: changelog]
Fixes: 6cd33649fa83 ("PCI: Add pci_configure_device() during enumeration")
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
|
|
Merge quoted strings that are broken across lines into a single entity.
The compiler merges them anyway, but checkpatch complains about it, and
merging them makes it easier to grep for strings.
No functional change.
[bhelgaas: changelog, do the same for everything under drivers/pci]
Signed-off-by: Ryan Desfosses <ryan@desfo.org>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
|
|
The PCI user-space config accessors pci_user_{read,write}_config_*() return
negative error numbers, which were introduced by commit 34e3207205ef
("PCI: handle positive error codes"). That patch converted all positive
error numbers from platform-specific PCI config accessors to -EINVAL, which
means the callers don't know anything about the specific cause of the
failure.
The patch fixes the issue by converting the positive PCIBIOS_* error values
to generic negative error numbers with pcibios_err_to_errno().
[bhelgaas: changelog]
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-by: Greg Thelen <gthelen@google.com>
|
|
My philosophy is unused code is dead code. And dead code is subject to bit
rot and is a likely source of bugs. Use it or lose it.
This reverts db5679437a2b ("PCI: add interface to set visible size of
VPD"), removing this interface:
pci_vpd_truncate()
[bhelgaas: split to separate patch, also remove prototype from pci.h]
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
|
|
pcie_cap_has_devctl() does nothing, so remove it. Simplicity over
consistency in this case. No functional change.
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-By: Jiang Liu <jiang.liu@huawei.com>
|
|
Previously we allowed callers to access Slot Capabilities, Status, and
Control for Root Ports even if the Root Port did not implement a slot.
This seems dubious because the spec only requires these registers if a
slot is implemented.
It's true that even Root Ports without slots must have *space* for these
slot registers, because the Root Capabilities, Status, and Control
registers are after the slot registers in the capability. However,
for a v1 PCIe Capability, the *semantics* of the slot registers are
undefined unless a slot is implemented.
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-By: Jiang Liu <jiang.liu@huawei.com>
|
|
Previously we relied on the PCIe r3.0, sec 7.8, spec language that says
"For Functions that do not implement the [Link, Slot, Root] registers,
these spaces must be hardwired to 0b," which means that for v2 PCIe
capabilities, we don't need to check the device type at all.
But it's simpler if we don't need to check the capability version at all,
and I think the spec is explicit enough about which registers are required
for which types that we can remove the version checks.
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-By: Jiang Liu <jiang.liu@huawei.com>
|
|
Every PCIe device has a link, except Root Complex Integrated Endpoints
and Root Complex Event Collectors. Previously we didn't give access
to PCIe capability link-related registers for Upstream Ports, Downstream
Ports, and Bridges, so attempts to read PCI_EXP_LNKCTL incorrectly
returned zero. See PCIe spec r3.0, sec 7.8 and 1.3.2.3.
Reference: http://lkml.kernel.org/r/979A8436335E3744ADCD3A9F2A2B68A52AD136BE@SJEXCHMB10.corp.ad.broadcom.com
Reported-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-By: Jiang Liu <jiang.liu@huawei.com>
|
|
PCI_EXP_FLAGS_TYPE is a mask, not an offset. Fix it.
Previously, pcie_capability_read_word(..., PCI_EXP_FLAGS, ...) would
fail.
[bhelgaas: tweak changelog]
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
CC: stable@vger.kernel.org # v3.7+
|
|
Use PCI Express Capability access functions to simplify device
Capabilities Register usages.
Signed-off-by: Myron Stowe <myron.stowe@redhat.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
|
|
The PCI Express Capability (PCIe spec r3.0, sec 7.8) comes in two
versions, v1 and v2. In v1 Capability structures (PCIe spec r1.0 and
r1.1), some fields are optional, so the structure size depends on the
device type.
This patch adds functions to access this capability so drivers don't
have to be aware of the differences between v1 and v2. Note that these
new functions apply only to the "PCI Express Capability," not to any of
the other "PCI Express Extended Capabilities" (AER, VC, ACS, MFVC, etc.)
Function pcie_capability_read_word/dword() reads the PCIe Capabilities
register and returns the value in the reference parameter "val". If
the PCIe Capabilities register is not implemented on the PCIe device,
"val" is set to 0.
Function pcie_capability_write_word/dword() writes the value to the
specified PCIe Capability register.
Function pcie_capability_clear_and_set_word/dword() sets and/or clears bits
of a PCIe Capability register.
[bhelgaas: changelog, drop "pci_" prefixes, don't export
pcie_capability_reg_implemented()]
Signed-off-by: Jiang Liu <jiang.liu@huawei.com>
Signed-off-by: Yijing Wang <wangyijing@huawei.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
|
|
VFIO PCI support will make use of these for user-initiated
PCI config accesses.
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
|
|
These new PCI services allow to probe for 2.3-compliant INTx masking
support and then use the feature from PCI interrupt handlers. The
services are properly synchronized with concurrent config space access
via sysfs or on device reset.
This enables generic PCI device drivers like uio_pci_generic or KVM's
device assignment to implement the necessary kernel-side IRQ handling
without any knowledge about device-specific interrupt status and control
registers.
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
|
|
pci_block_user_cfg_access was designed for the use case that a single
context, the IPR driver, temporarily delays user space accesses to the
config space via sysfs. This assumption became invalid by the time
pci_dev_reset was added as locking instance. Today, if you run two loops
in parallel that reset the same device via sysfs, you end up with a
kernel BUG as pci_block_user_cfg_access detect the broken assumption.
This reworks the pci_block_user_cfg_access to a sleeping service
pci_cfg_access_lock and an atomic-compatible variant called
pci_cfg_access_trylock. The former not only blocks user space access as
before but also waits if access was already locked. The latter service
just returns false in this case, allowing the caller to resolve the
conflict instead of raising a BUG.
Adaptions of the ipr driver were originally written by Brian King.
Acked-by: Brian King <brking@linux.vnet.ibm.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
|
|
Callers expect pci_user_{read,write}_config_*() to indicate errors by
returning negative values. Prior to this change, the indicated routines
could return positive error codes (e.g. PCIBIOS_BAD_REGISTER_NUMBER)
which callers would mistakenly interpret as success.
This change converts any non-zero return from the mentioned routines
into unambiguous negative value return codes.
Signed-off-by: Greg Thelen <gthelen@google.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
|
|
pci_vpd_pci22_write() calls pci_vpd_pci22_wait() after writing
PCI_VPD_DATA and PCI_VPD_ADDR to wait for the VPD operation to complete.
The result pci_vpd_pci22_wait() was not checked for error.
This change checks for error.
Signed-off-by: Greg Thelen <gthelen@google.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
|
|
pci_read/write_vpd() can fail due to a timeout. Usually the command
times out because of firmware issues (incorrect vpd length, etc.) on the
PCI card. Currently, the timeout occurs silently.
Output a message to the user indicating that they should check with
their vendor for new firmware.
Reviewed-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Prarit Bhargava <prarit@redhat.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
|
|
pci_lock must be a real spinlock in preempt-rt. Convert it to
raw_spinlock. No change for !RT kernels.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
|
|
implicit slab.h inclusion from percpu.h
percpu.h is included by sched.h and module.h and thus ends up being
included when building most .c files. percpu.h includes slab.h which
in turn includes gfp.h making everything defined by the two files
universally available and complicating inclusion dependencies.
percpu.h -> slab.h dependency is about to be removed. Prepare for
this change by updating users of gfp and slab facilities include those
headers directly instead of assuming availability. As this conversion
needs to touch large number of source files, the following script is
used as the basis of conversion.
http://userweb.kernel.org/~tj/misc/slabh-sweep.py
The script does the followings.
* Scan files for gfp and slab usages and update includes such that
only the necessary includes are there. ie. if only gfp is used,
gfp.h, if slab is used, slab.h.
* When the script inserts a new include, it looks at the include
blocks and try to put the new include such that its order conforms
to its surrounding. It's put in the include block which contains
core kernel includes, in the same order that the rest are ordered -
alphabetical, Christmas tree, rev-Xmas-tree or at the end if there
doesn't seem to be any matching order.
* If the script can't find a place to put a new include (mostly
because the file doesn't have fitting include block), it prints out
an error message indicating which .h file needs to be added to the
file.
The conversion was done in the following steps.
1. The initial automatic conversion of all .c files updated slightly
over 4000 files, deleting around 700 includes and adding ~480 gfp.h
and ~3000 slab.h inclusions. The script emitted errors for ~400
files.
2. Each error was manually checked. Some didn't need the inclusion,
some needed manual addition while adding it to implementation .h or
embedding .c file was more appropriate for others. This step added
inclusions to around 150 files.
3. The script was run again and the output was compared to the edits
from #2 to make sure no file was left behind.
4. Several build tests were done and a couple of problems were fixed.
e.g. lib/decompress_*.c used malloc/free() wrappers around slab
APIs requiring slab.h to be added manually.
5. The script was run on all .h files but without automatically
editing them as sprinkling gfp.h and slab.h inclusions around .h
files could easily lead to inclusion dependency hell. Most gfp.h
inclusion directives were ignored as stuff from gfp.h was usually
wildly available and often used in preprocessor macros. Each
slab.h inclusion directive was examined and added manually as
necessary.
6. percpu.h was updated not to include slab.h.
7. Build test were done on the following configurations and failures
were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my
distributed build env didn't work with gcov compiles) and a few
more options had to be turned off depending on archs to make things
build (like ipr on powerpc/64 which failed due to missing writeq).
* x86 and x86_64 UP and SMP allmodconfig and a custom test config.
* powerpc and powerpc64 SMP allmodconfig
* sparc and sparc64 SMP allmodconfig
* ia64 SMP allmodconfig
* s390 SMP allmodconfig
* alpha SMP allmodconfig
* um on x86_64 SMP allmodconfig
8. percpu.h modifications were reverted so that it could be applied as
a separate patch and serve as bisection point.
Given the fact that I had only a couple of failures from tests on step
6, I'm fairly confident about the coverage of this conversion patch.
If there is a breakage, it's likely to be something in one of the arch
headers which should be easily discoverable easily on most builds of
the specific arch.
Signed-off-by: Tejun Heo <tj@kernel.org>
Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
|
|
pci_bus_set_ops changes pci_ops associated with a pci_bus. This can be
used by debug tools such as PCIE AER error injection to fake some PCI
configuration registers.
Acked-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
|
|
Add drivers/pci/*.c source files to DocBook/kernel-api.tmpl
and update those pci/*.c source files that need kernel-doc fixes.
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
|
|
pci_vpd_truncate() should check for dev->vpd->attr, otherwise this might
happen:
sky2 driver version 1.22
Unable to handle kernel paging request for data at address 0x0000000c
Faulting instruction address: 0xc01836fc
Oops: Kernel access of bad area, sig: 11 [#1]
[...]
NIP [c01836fc] pci_vpd_truncate+0x38/0x40
LR [c029be18] sky2_probe+0x14c/0x518
Call Trace:
[ef82bde0] [c029bda4] sky2_probe+0xd8/0x518 (unreliable)
[ef82be20] [c018a11c] local_pci_probe+0x24/0x34
[ef82be30] [c018a14c] pci_call_probe+0x20/0x30
[ef82be50] [c018a330] __pci_device_probe+0x64/0x78
[ef82be60] [c018a44c] pci_device_probe+0x30/0x58
[ef82be80] [c01aa270] really_probe+0x78/0x1a0
[ef82bea0] [c01aa460] __driver_attach+0xa4/0xa8
[ef82bec0] [c01a96ac] bus_for_each_dev+0x60/0x9c
[ef82bef0] [c01aa0b4] driver_attach+0x24/0x34
[ef82bf00] [c01a9e08] bus_add_driver+0x12c/0x1cc
[ef82bf20] [c01aa87c] driver_register+0x6c/0x110
[ef82bf30] [c018a770] __pci_register_driver+0x4c/0x9c
[ef82bf50] [c03782c8] sky2_init_module+0x30/0x40
[ef82bf60] [c0001dbc] do_one_initcall+0x34/0x1a0
[ef82bfd0] [c0362240] do_initcalls+0x38/0x58
This happens with CONFIG_SKY2=y, and "ip=on" kernel command line, so
pci_vpd_truncate() is called before late_initcall(pci_sysfs_init),
therefore ->attr isn't yet initialized.
Acked-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.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>
|