Re: [PATCH v3 0/3] iommu/amd: Invalidate IRT cache for DMA aliases

[Magnus Kalland]

> I tested the patch with lockdep (CONFIG_PROVE_LOCKING=y) enabled and
> observed the following lockdep warnings in the kernel logs.

Thanks Dheeraj for the testing.

I believe we can make the dte_lock also a raw spinlock. What do you think?

I'll send a v4 addressing this lockdep issue and also with a fix for a
potential future lock-order inversion in iommu_flush_irt_and_complete.

Magnus