[PATCH rfcv1 5/8] iommu/arm-smmu-v3: Pre-allocate a per-master invalidation array

Jason Gunthorpe jgg at nvidia.com
Mon Sep 8 08:36:04 PDT 2025


On Sat, Sep 06, 2025 at 12:45:58AM -0700, Nicolin Chen wrote:
 
> One thing that I noticed is that group mutex alone isn't enough,
> because there can be two arm_smmu_build_invs() calls during the
> same attach_dev callback. And the second one would overwrite.

Well the group mutex is what makes it non-concurrent. Two things using
it sequentially within the same thread are a different issue

Jason



More information about the linux-arm-kernel mailing list