On Tue, 2024-08-06 at 15:45 +0200, Niklas Schnelle wrote:
> This fixes a crash when surprise hot-unplugging a PCI device. This crash
> happens because during hot-unplug __iommu_group_set_domain_nofail()
> attaching the default domain fails when the platform no longer
> recognizes the device as it has already been removed and we end up with
> a NULL domain pointer and UAF. This is exactly the case referred to in
> the second comment in __iommu_device_set_domain() and just as stated
> there if we can instead attach the blocking domain the UAF is prevented
> as this can handle the already removed device. Implement the blocking
> domain to use this handling. This would still leave us with a warning
> for a failed attach. As failing to attach when the device is no longer
> present is expected behavior turn this into an explicit -ENODEV error
> and don't warn for it. Also change the error return for a NULL zdev to
> -EIO as we don't want to ignore this case that would be a serious bug.
>
> Fixes: c76c067e488c ("s390/pci: Use dma-iommu layer")
> Signed-off-by: Niklas Schnelle <schnelle@xxxxxxxxxxxxx>
> ---
> Note: I somewhat suspect this to be related to the following discussion
> or at least we have seen the same backtraces in reports that we suspect
> to be caused by the issue fixed with this patch. In the case I was able
> to reproduce with vfio-pci pass-through to a KVM guest I got a different
> trace though.
Forgot the link:
https://lore.kernel.org/all/8743264a-9700-4227-a556-5f931c720211@xxxxxxxxxx/
>
> Organizational note: I'll be on vacation starting Thursday. Matt will
> then take over and sent new revisions as necessary.
> ---
> drivers/iommu/iommu.c | 7 ++++---
---8<---
