linux-stable.git/drivers/iommu, branch v5.3

iommu/amd: Fix race in increase_address_space()

2019-09-06T08:55:51+00:00

After the conversion to lock-less dma-api call the
increase_address_space() function can be called without any
locking. Multiple CPUs could potentially race for increasing
the address space, leading to invalid domain->mode settings
and invalid page-tables. This has been happening in the wild
under high IO load and memory pressure.

Fix the race by locking this operation. The function is
called infrequently so that this does not introduce
a performance regression in the dma-api path again.

Reported-by: Qian Cai 
Fixes: 256e4621c21a ('iommu/amd: Make use of the generic IOVA allocator')
Signed-off-by: Joerg Roedel

iommu/amd: Flush old domains in kdump kernel

2019-09-06T08:34:30+00:00

When devices are attached to the amd_iommu in a kdump kernel, the old device
table entries (DTEs), which were copied from the crashed kernel, will be
overwritten with a new domain number.  When the new DTE is written, the IOMMU
is told to flush the DTE from its internal cache--but it is not told to flush
the translation cache entries for the old domain number.

Without this patch, AMD systems using the tg3 network driver fail when kdump
tries to save the vmcore to a network system, showing network timeouts and
(sometimes) IOMMU errors in the kernel log.

This patch will flush IOMMU translation cache entries for the old domain when
a DTE gets overwritten with a new domain number.

Signed-off-by: Stuart Hayes 
Fixes: 3ac3e5ee5ed5 ('iommu/amd: Copy old trans table from old kernel')
Signed-off-by: Joerg Roedel

iommu/vt-d: Remove global page flush support

2019-09-03T13:01:27+00:00

Global pages support is removed from VT-d spec 3.0. Since global pages G
flag only affects first-level paging structures and because DMA request
with PASID are only supported by VT-d spec. 3.0 and onward, we can
safely remove global pages support.

For kernel shared virtual address IOTLB invalidation, PASID
granularity and page selective within PASID will be used. There is
no global granularity supported. Without this fix, IOTLB invalidation
will cause invalid descriptor error in the queued invalidation (QI)
interface.

Fixes: 1c4f88b7f1f9 ("iommu/vt-d: Shared virtual address in scalable mode")
Reported-by: Sanjay K Kumar 
Signed-off-by: Jacob Pan 
Signed-off-by: Joerg Roedel

Revert "iommu/vt-d: Avoid duplicated pci dma alias consideration"

2019-08-30T14:26:12+00:00

This reverts commit 557529494d79f3f1fadd486dd18d2de0b19be4da.

Commit 557529494d79f ("iommu/vt-d: Avoid duplicated pci dma alias
consideration") aimed to address a NULL pointer deference issue
happened when a thunderbolt device driver returned unexpectedly.

Unfortunately, this change breaks a previous pci quirk added by
commit cc346a4714a59 ("PCI: Add function 1 DMA alias quirk for
Marvell devices"), as the result, devices like Marvell 88SE9128
SATA controller doesn't work anymore.

We will continue to try to find the real culprit mentioned in
557529494d79f, but for now we should revert it to fix current
breakage.

Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=204627
Cc: Stijn Tintel 
Cc: Petr Vandrovec 
Reported-by: Stijn Tintel 
Reported-by: Petr Vandrovec 
Signed-off-by: Lu Baolu 
Signed-off-by: Joerg Roedel

dma-direct: fix zone selection after an unaddressable CMA allocation

2019-08-20T22:14:10+00:00

The new dma_alloc_contiguous hides if we allocate CMA or regular
pages, and thus fails to retry a ZONE_NORMAL allocation if the CMA
allocation succeeds but isn't addressable.  That means we either fail
outright or dip into a small zone that might not succeed either.

Thanks to Hillf Danton for debugging this issue.

Fixes: b1d2dc009dec ("dma-contiguous: add dma_{alloc,free}_contiguous() helpers")
Reported-by: Tobias Klausmann 
Signed-off-by: Christoph Hellwig 
Tested-by: Tobias Klausmann

Merge tag 'dma-mapping-5.3-4' of git://git.infradead.org/users/hch/dma-mapping

2019-08-14T17:31:11+00:00

Pull dma-mapping fixes from Christoph Hellwig:

 - fix the handling of the bus_dma_mask in dma_get_required_mask, which
   caused a regression in this merge window (Lucas Stach)

 - fix a regression in the handling of DMA_ATTR_NO_KERNEL_MAPPING (me)

 - fix dma_mmap_coherent to not cause page attribute mismatches on
   coherent architectures like x86 (me)

* tag 'dma-mapping-5.3-4' of git://git.infradead.org/users/hch/dma-mapping:
  dma-mapping: fix page attributes for dma_mmap_*
  dma-direct: don't truncate dma_required_mask to bus addressing capabilities
  dma-direct: fix DMA_ATTR_NO_KERNEL_MAPPING

dma-mapping: fix page attributes for dma_mmap_*

2019-08-10T17:52:45+00:00

All the way back to introducing dma_common_mmap we've defaulted to mark
the pages as uncached.  But this is wrong for DMA coherent devices.
Later on DMA_ATTR_WRITE_COMBINE also got incorrect treatment as that
flag is only treated special on the alloc side for non-coherent devices.

Introduce a new dma_pgprot helper that deals with the check for coherent
devices so that only the remapping cases ever reach arch_dma_mmap_pgprot
and we thus ensure no aliasing of page attributes happens, which makes
the powerpc version of arch_dma_mmap_pgprot obsolete and simplifies the
remaining ones.

Note that this means arch_dma_mmap_pgprot is a bit misnamed now, but
we'll phase it out soon.

Fixes: 64ccc9c033c6 ("common: dma-mapping: add support for generic dma_mmap_* calls")
Reported-by: Shawn Anastasio 
Reported-by: Gavin Li 
Signed-off-by: Christoph Hellwig 
Acked-by: Catalin Marinas  # arm64

iommu/vt-d: Fix possible use-after-free of private domain

2019-08-09T15:35:25+00:00

Multiple devices might share a private domain. One real example
is a pci bridge and all devices behind it. When remove a private
domain, make sure that it has been detached from all devices to
avoid use-after-free case.

Cc: Ashok Raj 
Cc: Jacob Pan 
Cc: Kevin Tian 
Cc: Alex Williamson 
Fixes: 942067f1b6b97 ("iommu/vt-d: Identify default domains replaced with private")
Signed-off-by: Lu Baolu 
Tested-by: Alex Williamson 
Signed-off-by: Joerg Roedel

iommu/vt-d: Detach domain before using a private one

2019-08-09T15:35:25+00:00

When the default domain of a group doesn't work for a device,
the iommu driver will try to use a private domain. The domain
which was previously attached to the device must be detached.

Cc: Ashok Raj 
Cc: Jacob Pan 
Cc: Kevin Tian 
Cc: Alex Williamson 
Fixes: 942067f1b6b97 ("iommu/vt-d: Identify default domains replaced with private")
Reported-by: Alex Williamson 
Link: https://lkml.org/lkml/2019/8/2/1379
Signed-off-by: Lu Baolu 
Tested-by: Alex Williamson 
Signed-off-by: Joerg Roedel

iommu/dma: Handle SG length overflow better

2019-08-09T15:31:39+00:00

Since scatterlist dimensions are all unsigned ints, in the relatively
rare cases where a device's max_segment_size is set to UINT_MAX, then
the "cur_len + s_length <= max_len" check in __finalise_sg() will always
return true. As a result, the corner case of such a device mapping an
excessively large scatterlist which is mergeable to or beyond a total
length of 4GB can lead to overflow and a bogus truncated dma_length in
the resulting segment.

As we already assume that any single segment must be no longer than
max_len to begin with, this can easily be addressed by reshuffling the
comparison.

Fixes: 809eac54cdd6 ("iommu/dma: Implement scatterlist segment merging")
Reported-by: Nicolin Chen 
Tested-by: Nicolin Chen 
Signed-off-by: Robin Murphy 
Signed-off-by: Joerg Roedel