On 5/2/22 18:27, Kirill A. Shutemov wrote:
Again, Dave and Andi already commented you should use vmap() to avoid breakingI would rather convert it to use DMA API for memory allocation. It will
up the direct-mapping. Please use vmap() instead.
https://lore.kernel.org/all/ce0feeec-a949-35f8-3010-b0d69acbbc2e@xxxxxxxxxxxxxxx/
Will review the rest later.
tap into swiotlb buffer that already converted and there's no need to
touch direct mapping. Both allocation and freeing such memory is cheaper
because of that.
Sathya, I don't quite understand why you are so forcefully declining to
incorporate review feedback on this point. I gave very specific
feedback about the kind of mapping you need and that you should avoid
fragmenting the direct map if at all possible.
Why is this code still fragmenting the direct map?