iommu/amd: serialize sequence allocation under concurrent TLB invalidations

Browse source

With concurrent TLB invalidations, completion wait randomly gets timed out
because cmd_sem_val was incremented outside the IOMMU spinlock, allowing
CMD_COMPL_WAIT commands to be queued out of sequence and breaking the
ordering assumption in wait_on_sem().
Move the cmd_sem_val increment under iommu->lock so completion sequence
allocation is serialized with command queuing.
And remove the unnecessary return.

Fixes: d2a0cac105 ("iommu/amd: move wait_on_sem() out of spinlock")

Tested-by: Srikanth Aithal <sraithal@amd.com>
Reported-by: Srikanth Aithal <sraithal@amd.com>
Signed-off-by: Ankit Soni <Ankit.Soni@amd.com>
Reviewed-by: Vasant Hegde <vasant.hegde@amd.com>
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>

...

This commit is contained in:

Ankit Soni 2026-01-22 15:30:38 +00:00 • committed by Joerg Roedel

parent 5b0530bb16

commit 9e249c4841

3 changed files with 14 additions and 8 deletions

Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL

Download patch file Download diff file Expand all files Collapse all files

2

drivers/iommu/amd/amd_iommu_types.h

View file

2

drivers/iommu/amd/init.c

View file

18

drivers/iommu/amd/iommu.c

View file