drivers/vfio/pci/vfio_pci_nvlink2.c cannot be compiled for in-tree
building.
CC drivers/vfio/pci/vfio_pci_nvlink2.o
In file included from drivers/vfio/pci/trace.h:102,
from drivers/vfio/pci/vfio_pci_nvlink2.c:29:
./include/trace/define_trace.h:89:42: fatal error: ./trace.h: No such file or directory
#include TRACE_INCLUDE(TRACE_INCLUDE_FILE)
^
compilation terminated.
make[1]: *** [scripts/Makefile.build;277: drivers/vfio/pci/vfio_pci_nvlink2.o] Error 1
To fix the build error, let's tell include/trace/define_trace.h the
location of drivers/vfio/pci/trace.h
Fixes: 7f92891778 ("vfio_pci: Add NVIDIA GV100GL [Tesla V100 SXM2] subdriver")
Reported-by: Laura Abbott <labbott@redhat.com>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
- Replace global vfio-pci lock with per bus lock to allow concurrent
open and release (Alex Williamson)
- Declare mdev function as static (Paolo Cretaro)
- Convert char to u8 in mdev/mtty sample driver (Nathan Chancellor)
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.14 (GNU/Linux)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=HOtX
-----END PGP SIGNATURE-----
Merge tag 'vfio-v4.21-rc1' of git://github.com/awilliam/linux-vfio
Pull VFIO updates from Alex Williamson:
- Replace global vfio-pci lock with per bus lock to allow concurrent
open and release (Alex Williamson)
- Declare mdev function as static (Paolo Cretaro)
- Convert char to u8 in mdev/mtty sample driver (Nathan Chancellor)
* tag 'vfio-v4.21-rc1' of git://github.com/awilliam/linux-vfio:
vfio-mdev/samples: Use u8 instead of char for handle functions
vfio/mdev: add static modifier to add_mdev_supported_type
vfio/pci: Parallelize device open and release
POWER9 Witherspoon machines come with 4 or 6 V100 GPUs which are not
pluggable PCIe devices but still have PCIe links which are used
for config space and MMIO. In addition to that the GPUs have 6 NVLinks
which are connected to other GPUs and the POWER9 CPU. POWER9 chips
have a special unit on a die called an NPU which is an NVLink2 host bus
adapter with p2p connections to 2 to 3 GPUs, 3 or 2 NVLinks to each.
These systems also support ATS (address translation services) which is
a part of the NVLink2 protocol. Such GPUs also share on-board RAM
(16GB or 32GB) to the system via the same NVLink2 so a CPU has
cache-coherent access to a GPU RAM.
This exports GPU RAM to the userspace as a new VFIO device region. This
preregisters the new memory as device memory as it might be used for DMA.
This inserts pfns from the fault handler as the GPU memory is not onlined
until the vendor driver is loaded and trained the NVLinks so doing this
earlier causes low level errors which we fence in the firmware so
it does not hurt the host system but still better be avoided; for the same
reason this does not map GPU RAM into the host kernel (usual thing for
emulated access otherwise).
This exports an ATSD (Address Translation Shootdown) register of NPU which
allows TLB invalidations inside GPU for an operating system. The register
conveniently occupies a single 64k page. It is also presented to
the userspace as a new VFIO device region. One NPU has 8 ATSD registers,
each of them can be used for TLB invalidation in a GPU linked to this NPU.
This allocates one ATSD register per an NVLink bridge allowing passing
up to 6 registers. Due to the host firmware bug (just recently fixed),
only 1 ATSD register per NPU was actually advertised to the host system
so this passes that alone register via the first NVLink bridge device in
the group which is still enough as QEMU collects them all back and
presents to the guest via vPHB to mimic the emulated NPU PHB on the host.
In order to provide the userspace with the information about GPU-to-NVLink
connections, this exports an additional capability called "tgt"
(which is an abbreviated host system bus address). The "tgt" property
tells the GPU its own system address and allows the guest driver to
conglomerate the routing information so each GPU knows how to get directly
to the other GPUs.
For ATS to work, the nest MMU (an NVIDIA block in a P9 CPU) needs to
know LPID (a logical partition ID or a KVM guest hardware ID in other
words) and PID (a memory context ID of a userspace process, not to be
confused with a linux pid). This assigns a GPU to LPID in the NPU and
this is why this adds a listener for KVM on an IOMMU group. A PID comes
via NVLink from a GPU and NPU uses a PID wildcard to pass it through.
This requires coherent memory and ATSD to be available on the host as
the GPU vendor only supports configurations with both features enabled
and other configurations are known not to work. Because of this and
because of the ways the features are advertised to the host system
(which is a device tree with very platform specific properties),
this requires enabled POWERNV platform.
The V100 GPUs do not advertise any of these capabilities via the config
space and there are more than just one device ID so this relies on
the platform to tell whether these GPUs have special abilities such as
NVLinks.
Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Acked-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
VFIO regions already support region capabilities with a limited set of
fields. However the subdriver might have to report to the userspace
additional bits.
This adds an add_capability() hook to vfio_pci_regops.
Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Acked-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
So far we only allowed mapping of MMIO BARs to the userspace. However
there are GPUs with on-board coherent RAM accessible via side
channels which we also want to map to the userspace. The first client
for this is NVIDIA V100 GPU with NVLink2 direct links to a POWER9
NPU-enabled CPU; such GPUs have 16GB RAM which is coherently mapped
to the system address space, we are going to export these as an extra
PCI region.
We already support extra PCI regions and this adds support for mapping
them to the userspace.
Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Acked-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
In commit 61d792562b ("vfio-pci: Use mutex around open, release, and
remove") a mutex was added to freeze the refcnt for a device so that
we can handle errors and perform bus resets on final close. However,
bus resets can be rather slow and a global mutex here is undesirable.
Evaluating the potential locking granularity, a per-device mutex
provides the best resolution but with multiple devices on a bus all
released concurrently, they'll race to acquire each other's mutex,
likely resulting in no reset at all if we use trylock. We therefore
lock at the granularity of the bus/slot reset as we're only attempting
a single reset for this group of devices anyway. This allows much
greater scaling as we're bounded in the number of devices protected by
a single reflck object.
Reported-by: Christian Ehrhardt <christian.ehrhardt@canonical.com>
Tested-by: Christian Ehrhardt <christian.ehrhardt@canonical.com>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
The SR-IOV spec requires that VFs must report zero for the INTx pin
register as VFs are precluded from INTx support. It's much easier for
the host kernel to understand whether a device is a VF and therefore
whether a non-zero pin register value is bogus than it is to do the
same in userspace. Override the INTx count for such devices and
virtualize the pin register to provide a consistent view of the device
to the user.
As this is clearly a spec violation, warn about it to support hardware
validation, but also provide a known whitelist as it doesn't do much
good to continue complaining if the hardware vendor doesn't plan to
fix it.
Known devices with this issue: 8086:270c
Tested-by: Gage Eads <gage.eads@intel.com>
Reviewed-by: Ashok Raj <ashok.raj@intel.com>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Free allocated vdev->msi_perm in error path.
Signed-off-by: Li Qiang <liq3ea@gmail.com>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
-----BEGIN PGP SIGNATURE-----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=2Dmn
-----END PGP SIGNATURE-----
Merge tag 'pci-v4.19-changes' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci
Pull pci updates from Bjorn Helgaas:
- Decode AER errors with names similar to "lspci" (Tyler Baicar)
- Expose AER statistics in sysfs (Rajat Jain)
- Clear AER status bits selectively based on the type of recovery (Oza
Pawandeep)
- Honor "pcie_ports=native" even if HEST sets FIRMWARE_FIRST (Alexandru
Gagniuc)
- Don't clear AER status bits if we're using the "Firmware-First"
strategy where firmware owns the registers (Alexandru Gagniuc)
- Use sysfs_match_string() to simplify ASPM sysfs parsing (Andy
Shevchenko)
- Remove unnecessary includes of <linux/pci-aspm.h> (Bjorn Helgaas)
- Defer DPC event handling to work queue (Keith Busch)
- Use threaded IRQ for DPC bottom half (Keith Busch)
- Print AER status while handling DPC events (Keith Busch)
- Work around IDT switch ACS Source Validation erratum (James
Puthukattukaran)
- Emit diagnostics for all cases of PCIe Link downtraining (Links
operating slower than they're capable of) (Alexandru Gagniuc)
- Skip VFs when configuring Max Payload Size (Myron Stowe)
- Reduce Root Port Max Payload Size if necessary when hot-adding a
device below it (Myron Stowe)
- Simplify SHPC existence/permission checks (Bjorn Helgaas)
- Remove hotplug sample skeleton driver (Lukas Wunner)
- Convert pciehp to threaded IRQ handling (Lukas Wunner)
- Improve pciehp tolerance of missed events and initially unstable
links (Lukas Wunner)
- Clear spurious pciehp events on resume (Lukas Wunner)
- Add pciehp runtime PM support, including for Thunderbolt controllers
(Lukas Wunner)
- Support interrupts from pciehp bridges in D3hot (Lukas Wunner)
- Mark fall-through switch cases before enabling -Wimplicit-fallthrough
(Gustavo A. R. Silva)
- Move DMA-debug PCI init from arch code to PCI core (Christoph
Hellwig)
- Fix pci_request_irq() usage of IRQF_ONESHOT when no handler is
supplied (Heiner Kallweit)
- Unify PCI and DMA direction #defines (Shunyong Yang)
- Add PCI_DEVICE_DATA() macro (Andy Shevchenko)
- Check for VPD completion before checking for timeout (Bert Kenward)
- Limit Netronome NFP5000 config space size to work around erratum
(Jakub Kicinski)
- Set IRQCHIP_ONESHOT_SAFE for PCI MSI irqchips (Heiner Kallweit)
- Document ACPI description of PCI host bridges (Bjorn Helgaas)
- Add "pci=disable_acs_redir=" parameter to disable ACS redirection for
peer-to-peer DMA support (we don't have the peer-to-peer support yet;
this is just one piece) (Logan Gunthorpe)
- Clean up devm_of_pci_get_host_bridge_resources() resource allocation
(Jan Kiszka)
- Fixup resizable BARs after suspend/resume (Christian König)
- Make "pci=earlydump" generic (Sinan Kaya)
- Fix ROM BAR access routines to stay in bounds and check for signature
correctly (Rex Zhu)
- Add DMA alias quirk for Microsemi Switchtec NTB (Doug Meyer)
- Expand documentation for pci_add_dma_alias() (Logan Gunthorpe)
- To avoid bus errors, enable PASID only if entire path supports
End-End TLP prefixes (Sinan Kaya)
- Unify slot and bus reset functions and remove hotplug knowledge from
callers (Sinan Kaya)
- Add Function-Level Reset quirks for Intel and Samsung NVMe devices to
fix guest reboot issues (Alex Williamson)
- Add function 1 DMA alias quirk for Marvell 88SS9183 PCIe SSD
Controller (Bjorn Helgaas)
- Remove Xilinx AXI-PCIe host bridge arch dependency (Palmer Dabbelt)
- Remove Aardvark outbound window configuration (Evan Wang)
- Fix Aardvark bridge window sizing issue (Zachary Zhang)
- Convert Aardvark to use pci_host_probe() to reduce code duplication
(Thomas Petazzoni)
- Correct the Cadence cdns_pcie_writel() signature (Alan Douglas)
- Add Cadence support for optional generic PHYs (Alan Douglas)
- Add Cadence power management ops (Alan Douglas)
- Remove redundant variable from Cadence driver (Colin Ian King)
- Add Kirin MSI support (Xiaowei Song)
- Drop unnecessary root_bus_nr setting from exynos, imx6, keystone,
armada8k, artpec6, designware-plat, histb, qcom, spear13xx (Shawn
Guo)
- Move link notification settings from DesignWare core to individual
drivers (Gustavo Pimentel)
- Add endpoint library MSI-X interfaces (Gustavo Pimentel)
- Correct signature of endpoint library IRQ interfaces (Gustavo
Pimentel)
- Add DesignWare endpoint library MSI-X callbacks (Gustavo Pimentel)
- Add endpoint library MSI-X test support (Gustavo Pimentel)
- Remove unnecessary GFP_ATOMIC from Hyper-V "new child" allocation
(Jia-Ju Bai)
- Add more devices to Broadcom PAXC quirk (Ray Jui)
- Work around corrupted Broadcom PAXC config space to enable SMMU and
GICv3 ITS (Ray Jui)
- Disable MSI parsing to work around broken Broadcom PAXC logic in some
devices (Ray Jui)
- Hide unconfigured functions to work around a Broadcom PAXC defect
(Ray Jui)
- Lower iproc log level to reduce console output during boot (Ray Jui)
- Fix mobiveil iomem/phys_addr_t type usage (Lorenzo Pieralisi)
- Fix mobiveil missing include file (Lorenzo Pieralisi)
- Add mobiveil Kconfig/Makefile support (Lorenzo Pieralisi)
- Fix mvebu I/O space remapping issues (Thomas Petazzoni)
- Use generic pci_host_bridge in mvebu instead of ARM-specific API
(Thomas Petazzoni)
- Whitelist VMD devices with fast interrupt handlers to avoid sharing
vectors with slow handlers (Keith Busch)
* tag 'pci-v4.19-changes' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci: (153 commits)
PCI/AER: Don't clear AER bits if error handling is Firmware-First
PCI: Limit config space size for Netronome NFP5000
PCI/MSI: Set IRQCHIP_ONESHOT_SAFE for PCI-MSI irqchips
PCI/VPD: Check for VPD access completion before checking for timeout
PCI: Add PCI_DEVICE_DATA() macro to fully describe device ID entry
PCI: Match Root Port's MPS to endpoint's MPSS as necessary
PCI: Skip MPS logic for Virtual Functions (VFs)
PCI: Add function 1 DMA alias quirk for Marvell 88SS9183
PCI: Check for PCIe Link downtraining
PCI: Add ACS Redirect disable quirk for Intel Sunrise Point
PCI: Add device-specific ACS Redirect disable infrastructure
PCI: Convert device-specific ACS quirks from NULL termination to ARRAY_SIZE
PCI: Add "pci=disable_acs_redir=" parameter for peer-to-peer support
PCI: Allow specifying devices using a base bus and path of devfns
PCI: Make specifying PCI devices in kernel parameters reusable
PCI: Hide ACS quirk declarations inside PCI core
PCI: Delay after FLR of Intel DC P3700 NVMe
PCI: Disable Samsung SM961/PM961 NVMe before FLR
PCI: Export pcie_has_flr()
PCI: mvebu: Drop bogus comment above mvebu_pcie_map_registers()
...
We expect to receive PFs with SR-IOV disabled, however some host
drivers leave SR-IOV enabled at unbind. This puts us in a state where
we can potentially assign both the PF and the VF, leading to both
functionality as well as security concerns due to lack of managing the
SR-IOV state as well as vendor dependent isolation from the PF to VF.
If we were to attempt to actively disable SR-IOV on driver probe, we
risk VF bound drivers blocking, potentially risking live lock
scenarios. Therefore simply refuse to bind to PFs with SR-IOV enabled
with a warning message indicating the issue. Users can resolve this
by re-binding to the host driver and disabling SR-IOV before
attempting to use the device with vfio-pci.
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Now that the old implementation of pci_reset_bus() is gone, replace
pci_try_reset_bus() with pci_reset_bus().
Compared to the old implementation, new code will fail immmediately with
-EAGAIN if object lock cannot be obtained.
Signed-off-by: Sinan Kaya <okaya@codeaurora.org>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Drivers are expected to call pci_try_reset_slot() or pci_try_reset_bus() by
querying if a system supports hotplug or not. A survey showed that most
drivers don't do this and we are leaking hotplug capability to the user.
Hide pci_try_slot_reset() from drivers and embed into pci_try_bus_reset().
Change pci_try_reset_bus() parameter from struct pci_bus to struct pci_dev.
Signed-off-by: Sinan Kaya <okaya@codeaurora.org>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
info.index can be indirectly controlled by user-space, hence leading
to a potential exploitation of the Spectre variant 1 vulnerability.
This issue was detected with the help of Smatch:
drivers/vfio/pci/vfio_pci.c:734 vfio_pci_ioctl()
warn: potential spectre issue 'vdev->region'
Fix this by sanitizing info.index before indirectly using it to index
vdev->region
Notice that given that speculation windows are large, the policy is
to kill the speculation on the first load and not worry if it can be
completed with a dependent load/store [1].
[1] https://marc.info/?l=linux-kernel&m=152449131114778&w=2
Cc: stable@vger.kernel.org
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Allow the code which provides extensions to support direct assignment
of Intel IGD (GVT-d) to be compiled out of the kernel if desired. The
config option for this was previously automatically enabled on X86,
therefore the default remains Y. This simply provides the option to
disable it even for X86.
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
The ioeventfd here is actually irqfd handling of an ioeventfd such as
supported in KVM. A user is able to pre-program a device write to
occur when the eventfd triggers. This is yet another instance of
eventfd-irqfd triggering between KVM and vfio. The impetus for this
is high frequency writes to pages which are virtualized in QEMU.
Enabling this near-direct write path for selected registers within
the virtualized page can improve performance and reduce overhead.
Specifically this is initially targeted at NVIDIA graphics cards where
the driver issues a write to an MMIO register within a virtualized
region in order to allow the MSI interrupt to re-trigger.
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
The iowriteXX/ioreadXX functions assume little endian hardware and
convert to little endian on a write and from little endian on a read.
We currently do our own explicit conversion to negate this. Instead,
add some endian dependent defines to avoid all byte swaps. There
should be no functional change other than big endian systems aren't
penalized with wasted swaps.
Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
This creates a common helper that we'll use for ioeventfd setup.
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
This reverts commit 2170dd0431
The intent of commit 2170dd0431 ("vfio-pci: Mask INTx if a device is
not capabable of enabling it") was to disallow the user from seeing
that the device supports INTx if the platform is incapable of enabling
it. The detection of this case however incorrectly includes devices
which natively do not support INTx, such as SR-IOV VFs, and further
discussions reveal gaps even for the target use case.
Reported-by: Arjun Vynipadath <arjun@chelsio.com>
Fixes: 2170dd0431 ("vfio-pci: Mask INTx if a device is not capabable of enabling it")
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
By default VFIO disables mapping of MSIX BAR to the userspace as
the userspace may program it in a way allowing spurious interrupts;
instead the userspace uses the VFIO_DEVICE_SET_IRQS ioctl.
In order to eliminate guessing from the userspace about what is
mmapable, VFIO also advertises a sparse list of regions allowed to mmap.
This works fine as long as the system page size equals to the MSIX
alignment requirement which is 4KB. However with a bigger page size
the existing code prohibits mapping non-MSIX parts of a page with MSIX
structures so these parts have to be emulated via slow reads/writes on
a VFIO device fd. If these emulated bits are accessed often, this has
serious impact on performance.
This allows mmap of the entire BAR containing MSIX vector table.
This removes the sparse capability for PCI devices as it becomes useless.
As the userspace needs to know for sure whether mmapping of the MSIX
vector containing data can succeed, this adds a new capability -
VFIO_REGION_INFO_CAP_MSIX_MAPPABLE - which explicitly tells the userspace
that the entire BAR can be mmapped.
This does not touch the MSIX mangling in the BAR read/write handlers as
we are doing this just to enable direct access to non MSIX registers.
Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
[aw - fixup whitespace, trim function name]
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
The vfio_info_add_capability() helper requires the caller to pass a
capability ID, which it then uses to fill in header fields, assuming
hard coded versions. This makes for an awkward and rigid interface.
The only thing we want this helper to do is allocate sufficient
space in the caps buffer and chain this capability into the list.
Reduce it to that simple task.
Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Acked-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Reviewed-by: Kirti Wankhede <kwankhede@nvidia.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
At the moment VFIO rightfully assumes that INTx is supported if
the interrupt pin is not set to zero in the device config space.
However if that is not the case (the pin is not zero but pdev->irq is),
vfio_intx_enable() fails.
In order to prevent the userspace from trying to enable INTx when we know
that it cannot work, let's mask the PCI_INTERRUPT_PIN register.
Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
MRRS defines the maximum read request size a device is allowed to
make. Drivers will often increase this to allow more data transfer
with a single request. Completions to this request are bound by the
MPS setting for the bus. Aside from device quirks (none known), it
doesn't seem to make sense to set an MRRS value less than MPS, yet
this is a likely scenario given that user drivers do not have a
system-wide view of the PCI topology. Virtualize MRRS such that the
user can set MRRS >= MPS, but use MPS as the floor value that we'll
write to hardware.
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
With virtual PCI-Express chipsets, we now see userspace/guest drivers
trying to match the physical MPS setting to a virtual downstream port.
Of course a lone physical device surrounded by virtual interconnects
cannot make a correct decision for a proper MPS setting. Instead,
let's virtualize the MPS control register so that writes through to
hardware are disallowed. Userspace drivers like QEMU assume they can
write anything to the device and we'll filter out anything dangerous.
Since mismatched MPS can lead to AER and other faults, let's add it
to the kernel side rather than relying on userspace virtualization to
handle it.
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Root complex integrated endpoints do not have a link and therefore may
use a smaller PCIe capability in config space than we expect when
building our config map. Add a case for these to avoid reporting an
erroneous overlap.
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Device lock bites again; if a device .remove() callback races a user
calling ioctl(VFIO_GROUP_GET_DEVICE_FD), the unbind request will hold
the device lock, but the user ioctl may have already taken a vfio_device
reference. In the case of a PCI device, the initial open will attempt
to reset the device, which again attempts to get the device lock,
resulting in deadlock. Use the trylock PCI reset interface and return
error on the open path if reset fails due to lock contention.
Link: https://lkml.org/lkml/2017/7/25/381
Reported-by: Wen Congyang <wencongyang2@huawei.com>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
XXV710 has the same broken INTx behavior as the rest of the X/XL710
series, the interrupt status register is not wired to report pending
INTx interrupts, thus we never associate the interrupt to the device.
Extend the device IDs to include these so that we hide that the
device supports INTx at all to the user.
Reported-by: Stefan Assmann <sassmann@redhat.com>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Acked-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Here, pci_iomap can fail, handle this case release selected
pci regions and return -ENOMEM.
Signed-off-by: Arvind Yadav <arvind.yadav.cs@gmail.com>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Using ancient compilers (gcc-4.5 or older) on ARM, we get a link
failure with the vfio-pci driver:
ERROR: "__aeabi_lcmp" [drivers/vfio/pci/vfio-pci.ko] undefined!
The reason is that the compiler tries to do a comparison of
a 64-bit range. This changes it to convert to a 32-bit number
explicitly first, as newer compilers do for themselves.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1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=ut0k
-----END PGP SIGNATURE-----
Merge tag 'pci-v4.10-changes' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci
Pull PCI updates from Bjorn Helgaas:
"PCI changes:
- add support for PCI on ARM64 boxes with ACPI. We already had this
for theoretical spec-compliant hardware; now we're adding quirks
for the actual hardware (Cavium, HiSilicon, Qualcomm, X-Gene)
- add runtime PM support for hotplug ports
- enable runtime suspend for Intel UHCI that uses platform-specific
wakeup signaling
- add yet another host bridge registration interface. We hope this is
extensible enough to subsume the others
- expose device revision in sysfs for DRM
- to avoid device conflicts, make sure any VF BAR updates are done
before enabling the VF
- avoid unnecessary link retrains for ASPM
- allow INTx masking on Mellanox devices that support it
- allow access to non-standard VPD for Chelsio devices
- update Broadcom iProc support for PAXB v2, PAXC v2, inbound DMA,
etc
- update Rockchip support for max-link-speed
- add NVIDIA Tegra210 support
- add Layerscape LS1046a support
- update R-Car compatibility strings
- add Qualcomm MSM8996 support
- remove some uninformative bootup messages"
* tag 'pci-v4.10-changes' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci: (115 commits)
PCI: Enable access to non-standard VPD for Chelsio devices (cxgb3)
PCI: Expand "VPD access disabled" quirk message
PCI: pciehp: Remove loading message
PCI: hotplug: Remove hotplug core message
PCI: Remove service driver load/unload messages
PCI/AER: Log AER IRQ when claiming Root Port
PCI/AER: Log errors with PCI device, not PCIe service device
PCI/AER: Remove unused version macros
PCI/PME: Log PME IRQ when claiming Root Port
PCI/PME: Drop unused support for PMEs from Root Complex Event Collectors
PCI: Move config space size macros to pci_regs.h
x86/platform/intel-mid: Constify mid_pci_platform_pm
PCI/ASPM: Don't retrain link if ASPM not possible
PCI: iproc: Skip check for legacy IRQ on PAXC buses
PCI: pciehp: Leave power indicator on when enabling already-enabled slot
PCI: pciehp: Prioritize data-link event over presence detect
PCI: rcar: Add gen3 fallback compatibility string for pcie-rcar
PCI: rcar: Use gen2 fallback compatibility last
PCI: rcar-gen2: Use gen2 fallback compatibility last
PCI: rockchip: Move the deassert of pm/aclk/pclk after phy_init()
..
Move PCI configuration space size macros (PCI_CFG_SPACE_SIZE and
PCI_CFG_SPACE_EXP_SIZE) from drivers/pci/pci.h to
include/uapi/linux/pci_regs.h so they can be used by more drivers and
eliminate duplicate definitions.
[bhelgaas: Expand comment to include PCI-X details]
Signed-off-by: Wang Sheng-Hui <shhuiw@foxmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
As of commit d97ffe2368 ("PCI: Fix return value from
pci_user_{read,write}_config_*()") it's unnecessary to call
pcibios_err_to_errno() to fixup the return value from these functions.
pcibios_err_to_errno() already does simple passthrough of -errno values,
therefore no functional change is expected.
[aw: changelog]
Signed-off-by: Cao jin <caoj.fnst@cn.fujitsu.com>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Update msix_sparse_mmap_cap() to use vfio_info_add_capability()
Update region type capability to use vfio_info_add_capability()
Signed-off-by: Kirti Wankhede <kwankhede@nvidia.com>
Signed-off-by: Neo Jia <cjia@nvidia.com>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
The VFIO_DEVICE_SET_IRQS ioctl did not sufficiently sanitize
user-supplied integers, potentially allowing memory corruption. This
patch adds appropriate integer overflow checks, checks the range bounds
for VFIO_IRQ_SET_DATA_NONE, and also verifies that only single element
in the VFIO_IRQ_SET_DATA_TYPE_MASK bitmask is set.
VFIO_IRQ_SET_ACTION_TYPE_MASK is already correctly checked later in
vfio_pci_set_irqs_ioctl().
Furthermore, a kzalloc is changed to a kcalloc because the use of a
kzalloc with an integer multiplication allowed an integer overflow
condition to be reached without this patch. kcalloc checks for overflow
and should prevent a similar occurrence.
Signed-off-by: Vlad Tsyrklevich <vlad@tsyrklevich.net>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Simplify the interrupt setup by using the new PCI layer helpers.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
The MSI/X shutdown path can gratuitously enable INTx, which is not
something we want to happen if we're dealing with broken INTx device.
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
We use a BAR restore trick to try to detect when a user has performed
a device reset, possibly through FLR or other backdoors, to put things
back into a working state. This is important for backdoor resets, but
we can actually just virtualize the "front door" resets provided via
PCIe and AF FLR. Set these bits as virtualized + writable, allowing
the default write to set them in vconfig, then we can simply check the
bit, perform an FLR of our own, and clear the bit. We don't actually
have the granularity in PCI to specify the type of reset we want to
do, but generally devices don't implement both PCIe and AF FLR and
we'll favor these over other types of reset, so we should generally
lineup. We do test whether the device provides the requested FLR type
to stay consistent with hardware capabilities though.
This seems to fix several instance of devices getting into bad states
with userspace drivers, like dpdk, running inside a VM.
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Reviewed-by: Greg Rose <grose@lightfleet.com>
There are multiple cases in vfio_pci_set_ctx_trigger_single() where
we assume we can safely read from our data pointer without actually
checking whether the user has passed any data via the count field.
VFIO_IRQ_SET_DATA_NONE in particular is entirely broken since we
attempt to pull an int32_t file descriptor out before even checking
the data type. The other data types assume the data pointer contains
one element of their type as well.
In part this is good news because we were previously restricted from
doing much sanitization of parameters because it was missed in the
past and we didn't want to break existing users. Clearly DATA_NONE
is completely broken, so it must not have any users and we can fix
it up completely. For DATA_BOOL and DATA_EVENTFD, we'll just
protect ourselves, returning error when count is zero since we
previously would have oopsed.
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Reported-by: Chris Thompson <the_cartographer@hotmail.com>
Cc: stable@vger.kernel.org
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Current vfio-pci implementation disallows to mmap
sub-page(size < PAGE_SIZE) MMIO BARs because these BARs' mmio
page may be shared with other BARs. This will cause some
performance issues when we passthrough a PCI device with
this kind of BARs. Guest will be not able to handle the mmio
accesses to the BARs which leads to mmio emulations in host.
However, not all sub-page BARs will share page with other BARs.
We should allow to mmap the sub-page MMIO BARs which we can
make sure will not share page with other BARs.
This patch adds support for this case. And we try to add a
dummy resource to reserve the remainder of the page which
hot-add device's BAR might be assigned into. But it's not
necessary to handle the case when the BAR is not page aligned.
Because we can't expect the BAR will be assigned into the same
location in a page in guest when we passthrough the BAR. And
it's hard to access this BAR in userspace because we have
no way to get the BAR's location in a page.
Signed-off-by: Yongji Xie <xyjxie@linux.vnet.ibm.com>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
The size of the VPD area is not necessarily 4-byte aligned, so a
pci_vpd_read() might return less than 4 bytes. Zero our buffer and
accept anything other than an error. Intel X710 NICs exercise this.
Fixes: 4e1a635552 ("vfio/pci: Use kernel VPD access functions")
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Both the INTx and MSI/X disable paths do an eventfd_ctx_put() for the
trigger eventfd before calling vfio_virqfd_disable() any potential
mask and unmask eventfds. This opens a use-after-free race where an
inopportune irqfd can reference the freed signalling eventfd. Reorder
to avoid this possibility.
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
PCI-Express spec says that reading 4 bytes at offset 100h should return
zero if there is no extended capability so VFIO reads this dword to
know if there are extended capabilities.
However it is not always possible to access the extended space so
generic PCI code in pci_cfg_space_size_ext() checks if
pci_read_config_dword() can read beyond 100h and if the check fails,
it sets the config space size to 100h.
VFIO does its own extended capabilities check by reading at offset 100h
which may produce 0xffffffff which VFIO treats as the extended config
space presense and calls vfio_ecap_init() which fails to parse
capabilities (which is expected) but right before the exit, it writes
zero at offset 100h which is beyond the buffer allocated for
vdev->vconfig (which is 256 bytes) which leads to random memory
corruption.
This makes VFIO only check for the extended capabilities if
the discovered config size is more than 256 bytes.
Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
If a device is reset without the memory or i/o bits enabled in the
command register we may not detect it, potentially leaving the device
without valid BAR programming. Add an additional test to check the
BARs on each write to the command register.
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
INTx masking has two components, the first is that we need the ability
to prevent the device from continuing to assert INTx. This is
provided via the DisINTx bit in the command register and is the only
thing we can really probe for when testing if INTx masking is
supported. The second component is that the device needs to indicate
if INTx is asserted via the interrupt status bit in the device status
register. With these two features we can generically determine if one
of the devices we own is asserting INTx, signal the user, and mask the
interrupt while the user services the device.
Generally if one or both of these components is broken we resort to
APIC level interrupt masking, which requires an exclusive interrupt
since we have no way to determine the source of the interrupt in a
shared configuration. This often makes it difficult or impossible to
configure the system for userspace use of the device, for an interrupt
mode that the user may not need.
One possible configuration of broken INTx masking is that the DisINTx
support is fully functional, but the interrupt status bit never
signals interrupt assertion. In this case we do have the ability to
prevent the device from asserting INTx, but lack the ability to
identify the interrupt source. For this case we can simply pretend
that the device lacks INTx support entirely, keeping DisINTx set on
the physical device, virtualizing this bit for the user, and
virtualizing the interrupt pin register to indicate no INTx support.
We already support virtualization of the DisINTx bit and already
virtualize the interrupt pin for platforms without INTx support. By
tying these components together, setting DisINTx on open and reset,
and identifying devices broken in this particular way, we can provide
support for them w/o the handicap of APIC level INTx masking.
Intel i40e (XL710/X710) 10/20/40GbE NICs have been identified as being
broken in this specific way. We leave the vfio-pci.nointxmask option
as a mechanism to bypass this support, enabling INTx on the device
with all the requirements of APIC level masking.
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Cc: John Ronciak <john.ronciak@intel.com>
Cc: Jesse Brandeburg <jesse.brandeburg@intel.com>
Various enablers for assignment of Intel graphics devices and future
support of vGPU devices (Alex Williamson). This includes
- Handling the vfio type1 interface as an API rather than a specific
implementation, allowing multiple type1 providers.
- Capability chains, similar to PCI device capabilities, that allow
extending ioctls. Extensions here include device specific regions
and sparse mmap descriptions. The former is used to expose non-PCI
regions for IGD, including the OpRegion (particularly the Video
BIOS Table), and read only PCI config access to the host and LPC
bridge as drivers often depend on identifying those devices.
Sparse mmaps here are used to describe the MSIx vector table,
which vfio has always protected from mmap, but never had an API to
explicitly define that protection. In future vGPU support this is
expected to allow the description of PCI BARs that may mix direct
access and emulated access within a single region.
- The ability to expose the shadow ROM as an option ROM as IGD use
cases may rely on the ROM even though the physical device does not
make use of a PCI option ROM BAR.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.14 (GNU/Linux)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=/lJn
-----END PGP SIGNATURE-----
Merge tag 'vfio-v4.6-rc1' of git://github.com/awilliam/linux-vfio
Pull VFIO updates from Alex Williamson:
"Various enablers for assignment of Intel graphics devices and future
support of vGPU devices (Alex Williamson). This includes
- Handling the vfio type1 interface as an API rather than a specific
implementation, allowing multiple type1 providers.
- Capability chains, similar to PCI device capabilities, that allow
extending ioctls. Extensions here include device specific regions
and sparse mmap descriptions. The former is used to expose non-PCI
regions for IGD, including the OpRegion (particularly the Video
BIOS Table), and read only PCI config access to the host and LPC
bridge as drivers often depend on identifying those devices.
Sparse mmaps here are used to describe the MSIx vector table, which
vfio has always protected from mmap, but never had an API to
explicitly define that protection. In future vGPU support this is
expected to allow the description of PCI BARs that may mix direct
access and emulated access within a single region.
- The ability to expose the shadow ROM as an option ROM as IGD use
cases may rely on the ROM even though the physical device does not
make use of a PCI option ROM BAR"
* tag 'vfio-v4.6-rc1' of git://github.com/awilliam/linux-vfio:
vfio/pci: return -EFAULT if copy_to_user fails
vfio/pci: Expose shadow ROM as PCI option ROM
vfio/pci: Intel IGD host and LCP bridge config space access
vfio/pci: Intel IGD OpRegion support
vfio/pci: Enable virtual register in PCI config space
vfio/pci: Add infrastructure for additional device specific regions
vfio: Define device specific region type capability
vfio/pci: Include sparse mmap capability for MSI-X table regions
vfio: Define sparse mmap capability for regions
vfio: Add capability chain helpers
vfio: Define capability chains
vfio: If an IOMMU backend fails, keep looking
vfio/pci: Fix unsigned comparison overflow
Calling return copy_to_user(...) in an ioctl will not
do the right thing if there's a pagefault:
copy_to_user returns the number of bytes not copied
in this case.
Fix up vfio to do
return copy_to_user(...)) ?
-EFAULT : 0;
everywhere.
Cc: stable@vger.kernel.org
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
