On 27/05/2021 04:13, Alexey Kardashevskiy wrote:

On 27/05/2021 00:45, Frederic Barrat wrote:

quoted

This reverts commit 3c0468d4451eb6b4f6604370639f163f9637a479.

That commit was breaking alignment guarantees for the DMA address when
allocating coherent mappings, as described in
Documentation/core-api/dma-api-howto.rst

It was also noticed by Mellanox' driver:
[ 1515.763621] mlx5_core c002:01:00.0: 
mlx5_frag_buf_alloc_node:146:(pid 13402): unexpected map alignment: 
0x0800000000c61000, page_shift=16
[ 1515.763635] mlx5_core c002:01:00.0: mlx5_cqwq_create:181:(pid
13402): mlx5_frag_buf_alloc_node() failed, -12

Signed-off-by: Frederic Barrat <redacted>

Should it be

Fixes: 3c0468d4451e ("powerpc/kernel/iommu: Align size for 
IOMMU_PAGE_SIZE() to save TCEs")

?

I had been wondering the same... I can see many revert commits with and 
without a "Fixes:" line. Since here the offending commit was merged in 
the latest merge window, I was thinking Fixes doesn't really bring 
anything, it will all stay internal to v5.13 development. I'd be happy 
to hear of the expected way of handling it. I'm guessing a big part of 
the answer is whether the tooling looking for a "Fixes" line is also 
looking for reverts.

   Fred

Anyway,

Reviewed-by: Alexey Kardashevskiy <redacted>