linux.git/drivers/vfio/pci, branch v5.4.156

vfio/pci: Handle concurrent vma faults

2021-07-14T14:53:47+00:00

[ Upstream commit 6a45ece4c9af473555f01f0f8b97eba56e3c7d0d ]

io_remap_pfn_range() will trigger a BUG_ON if it encounters a
populated pte within the mapping range.  This can occur because we map
the entire vma on fault and multiple faults can be blocked behind the
vma_lock.  This leads to traces like the one reported below.

We can use our vma_list to test whether a given vma is mapped to avoid
this issue.

[ 1591.733256] kernel BUG at mm/memory.c:2177!
[ 1591.739515] Internal error: Oops - BUG: 0 [#1] PREEMPT SMP
[ 1591.747381] Modules linked in: vfio_iommu_type1 vfio_pci vfio_virqfd vfio pv680_mii(O)
[ 1591.760536] CPU: 2 PID: 227 Comm: lcore-worker-2 Tainted: G O 5.11.0-rc3+ #1
[ 1591.770735] Hardware name:  , BIOS HixxxxFPGA 1P B600 V121-1
[ 1591.778872] pstate: 40400009 (nZcv daif +PAN -UAO -TCO BTYPE=--)
[ 1591.786134] pc : remap_pfn_range+0x214/0x340
[ 1591.793564] lr : remap_pfn_range+0x1b8/0x340
[ 1591.799117] sp : ffff80001068bbd0
[ 1591.803476] x29: ffff80001068bbd0 x28: 0000042eff6f0000
[ 1591.810404] x27: 0000001100910000 x26: 0000001300910000
[ 1591.817457] x25: 0068000000000fd3 x24: ffffa92f1338e358
[ 1591.825144] x23: 0000001140000000 x22: 0000000000000041
[ 1591.832506] x21: 0000001300910000 x20: ffffa92f141a4000
[ 1591.839520] x19: 0000001100a00000 x18: 0000000000000000
[ 1591.846108] x17: 0000000000000000 x16: ffffa92f11844540
[ 1591.853570] x15: 0000000000000000 x14: 0000000000000000
[ 1591.860768] x13: fffffc0000000000 x12: 0000000000000880
[ 1591.868053] x11: ffff0821bf3d01d0 x10: ffff5ef2abd89000
[ 1591.875932] x9 : ffffa92f12ab0064 x8 : ffffa92f136471c0
[ 1591.883208] x7 : 0000001140910000 x6 : 0000000200000000
[ 1591.890177] x5 : 0000000000000001 x4 : 0000000000000001
[ 1591.896656] x3 : 0000000000000000 x2 : 0168044000000fd3
[ 1591.903215] x1 : ffff082126261880 x0 : fffffc2084989868
[ 1591.910234] Call trace:
[ 1591.914837]  remap_pfn_range+0x214/0x340
[ 1591.921765]  vfio_pci_mmap_fault+0xac/0x130 [vfio_pci]
[ 1591.931200]  __do_fault+0x44/0x12c
[ 1591.937031]  handle_mm_fault+0xcc8/0x1230
[ 1591.942475]  do_page_fault+0x16c/0x484
[ 1591.948635]  do_translation_fault+0xbc/0xd8
[ 1591.954171]  do_mem_abort+0x4c/0xc0
[ 1591.960316]  el0_da+0x40/0x80
[ 1591.965585]  el0_sync_handler+0x168/0x1b0
[ 1591.971608]  el0_sync+0x174/0x180
[ 1591.978312] Code: eb1b027f 540000c0 f9400022 b4fffe02 (d4210000)

Fixes: 11c4cd07ba11 ("vfio-pci: Fault mmaps to enable vma tracking")
Reported-by: Zeng Tao 
Suggested-by: Zeng Tao 
Link: https://lore.kernel.org/r/162497742783.3883260.3282953006487785034.stgit@omen
Signed-off-by: Alex Williamson 
Signed-off-by: Sasha Levin

vfio/pci: zap_vma_ptes() needs MMU

2021-06-10T11:37:03+00:00

[ Upstream commit 2a55ca37350171d9b43d561528f23d4130097255 ]

zap_vma_ptes() is only available when CONFIG_MMU is set/enabled.
Without CONFIG_MMU, vfio_pci.o has build errors, so make
VFIO_PCI depend on MMU.

riscv64-linux-ld: drivers/vfio/pci/vfio_pci.o: in function `vfio_pci_mmap_open':
vfio_pci.c:(.text+0x1ec): undefined reference to `zap_vma_ptes'
riscv64-linux-ld: drivers/vfio/pci/vfio_pci.o: in function `.L0 ':
vfio_pci.c:(.text+0x165c): undefined reference to `zap_vma_ptes'

Fixes: 11c4cd07ba11 ("vfio-pci: Fault mmaps to enable vma tracking")
Signed-off-by: Randy Dunlap 
Reported-by: kernel test robot 
Cc: Alex Williamson 
Cc: Cornelia Huck 
Cc: kvm@vger.kernel.org
Cc: Jason Gunthorpe 
Cc: Eric Auger 
Message-Id: <20210515190856.2130-1-rdunlap@infradead.org>
Signed-off-by: Alex Williamson 
Signed-off-by: Sasha Levin

vfio/pci: Fix error return code in vfio_ecap_init()

2021-06-10T11:37:03+00:00

[ Upstream commit d1ce2c79156d3baf0830990ab06d296477b93c26 ]

The error code returned from vfio_ext_cap_len() is stored in 'len', not
in 'ret'.

Fixes: 89e1f7d4c66d ("vfio: Add PCI device driver")
Reported-by: Hulk Robot 
Signed-off-by: Zhen Lei 
Reviewed-by: Max Gurtovoy 
Message-Id: <20210515020458.6771-1-thunder.leizhen@huawei.com>
Signed-off-by: Alex Williamson 
Signed-off-by: Sasha Levin

vfio/pci: Add missing range check in vfio_pci_mmap

2021-04-21T10:56:16+00:00

commit 909290786ea335366e21d7f1ed5812b90f2f0a92 upstream.

When mmaping an extra device region verify that the region index
derived from the mmap offset is valid.

Fixes: a15b1883fee1 ("vfio_pci: Allow mapping extra regions")
Cc: stable@vger.kernel.org
Signed-off-by: Christian A. Ehrhardt 
Message-Id: <20210412214124.GA241759@lisa.in-ulm.de>
Reviewed-by: David Gibson 
Reviewed-by: Cornelia Huck 
Signed-off-by: Alex Williamson 
Signed-off-by: Greg Kroah-Hartman

vfio/nvlink: Add missing SPAPR_TCE_IOMMU depends

2021-04-07T12:47:43+00:00

commit e0146a108ce4d2c22b9510fd12268e3ee72a0161 upstream.

Compiling the nvlink stuff relies on the SPAPR_TCE_IOMMU otherwise there
are compile errors:

 drivers/vfio/pci/vfio_pci_nvlink2.c:101:10: error: implicit declaration of function 'mm_iommu_put' [-Werror,-Wimplicit-function-declaration]
                            ret = mm_iommu_put(data->mm, data->mem);

As PPC only defines these functions when the config is set.

Previously this wasn't a problem by chance as SPAPR_TCE_IOMMU was the only
IOMMU that could have satisfied IOMMU_API on POWERNV.

Fixes: 179209fa1270 ("vfio: IOMMU_API should be selected")
Reported-by: kernel test robot 
Signed-off-by: Jason Gunthorpe 
Message-Id: <0-v1-83dba9768fc3+419-vfio_nvlink2_kconfig_jgg@nvidia.com>
Signed-off-by: Alex Williamson 
Signed-off-by: Greg Kroah-Hartman

vfio/pci: Move dummy_resources_list init in vfio_pci_probe()

2021-01-06T13:48:36+00:00

[ Upstream commit 16b8fe4caf499ae8e12d2ab1b1324497e36a7b83 ]

In case an error occurs in vfio_pci_enable() before the call to
vfio_pci_probe_mmaps(), vfio_pci_disable() will  try to iterate
on an uninitialized list and cause a kernel panic.

Lets move to the initialization to vfio_pci_probe() to fix the
issue.

Signed-off-by: Eric Auger 
Fixes: 05f0c03fbac1 ("vfio-pci: Allow to mmap sub-page MMIO BARs if the mmio page is exclusive")
CC: Stable  # v4.7+
Signed-off-by: Alex Williamson 
Signed-off-by: Sasha Levin

vfio/pci/nvlink2: Do not attempt NPU2 setup on POWER8NVL NPU

2020-12-30T10:51:30+00:00

commit d22f9a6c92de96304c81792942ae7c306f08ac77 upstream.

We execute certain NPU2 setup code (such as mapping an LPID to a device
in NPU2) unconditionally if an Nvlink bridge is detected. However this
cannot succeed on POWER8NVL machines as the init helpers return an error
other than ENODEV which means the device is there is and setup failed so
vfio_pci_enable() fails and pass through is not possible.

This changes the two NPU2 related init helpers to return -ENODEV if
there is no "memory-region" device tree property as this is
the distinction between NPU and NPU2.

Tested on
- POWER9 pvr=004e1201, Ubuntu 19.04 host, Ubuntu 18.04 vm,
  NVIDIA GV100 10de:1db1 driver 418.39
- POWER8 pvr=004c0100, RHEL 7.6 host, Ubuntu 16.10 vm,
  NVIDIA P100 10de:15f9 driver 396.47

Fixes: 7f92891778df ("vfio_pci: Add NVIDIA GV100GL [Tesla V100 SXM2] subdriver")
Cc: stable@vger.kernel.org # 5.0
Signed-off-by: Alexey Kardashevskiy 
Signed-off-by: Alex Williamson 
Signed-off-by: Greg Kroah-Hartman

vfio-pci: Use io_remap_pfn_range() for PCI IO memory

2020-12-30T10:51:16+00:00

[ Upstream commit 7b06a56d468b756ad6bb43ac21b11e474ebc54a0 ]

commit f8f6ae5d077a ("mm: always have io_remap_pfn_range() set
pgprot_decrypted()") allows drivers using mmap to put PCI memory mapped
BAR space into userspace to work correctly on AMD SME systems that default
to all memory encrypted.

Since vfio_pci_mmap_fault() is working with PCI memory mapped BAR space it
should be calling io_remap_pfn_range() otherwise it will not work on SME
systems.

Fixes: 11c4cd07ba11 ("vfio-pci: Fault mmaps to enable vma tracking")
Signed-off-by: Jason Gunthorpe 
Acked-by: Peter Xu 
Tested-by: Tom Lendacky 
Signed-off-by: Alex Williamson 
Signed-off-by: Sasha Levin

vfio/pci: Bypass IGD init in case of -ENODEV

2020-11-18T18:20:25+00:00

[ Upstream commit e4eccb853664de7bcf9518fb658f35e748bf1f68 ]

Bypass the IGD initialization when -ENODEV returns,
that should be the case if opregion is not available for IGD
or within discrete graphics device's option ROM,
or host/lpc bridge is not found.

Then use of -ENODEV here means no special device resources found
which needs special care for VFIO, but we still allow other normal
device resource access.

Cc: Zhenyu Wang 
Cc: Xiong Zhang 
Cc: Hang Yuan 
Cc: Stuart Summers 
Signed-off-by: Fred Gao 
Signed-off-by: Alex Williamson 
Signed-off-by: Sasha Levin

vfio/pci: Clear token on bypass registration failure

2020-10-29T08:57:56+00:00

[ Upstream commit 852b1beecb6ff9326f7ca4bc0fe69ae860ebdb9e ]

The eventfd context is used as our irqbypass token, therefore if an
eventfd is re-used, our token is the same.  The irqbypass code will
return an -EBUSY in this case, but we'll still attempt to unregister
the producer, where if that duplicate token still exists, results in
removing the wrong object.  Clear the token of failed producers so
that they harmlessly fall out when unregistered.

Fixes: 6d7425f109d2 ("vfio: Register/unregister irq_bypass_producer")
Reported-by: guomin chen 
Tested-by: guomin chen 
Signed-off-by: Alex Williamson 
Signed-off-by: Sasha Levin