RE: [PATCH v5 6/6] swiotlb: Remove pointless stride adjustment for allocations >= PAGE_SIZE

[Michael Kelley]

From: Christoph Hellwig <hch@xxxxxx> Sent: Thursday, February 29, 2024 5:34 AM
> 
> On Thu, Feb 29, 2024 at 07:36:05AM +0000, Michael Kelley wrote:
> > If there *is* a requirement for page alignment of page-size-or-greater
> > requests, even when alloc_align_mask and min_align_mask are zero,
> > we need to think about how to do that correctly, as that requirement
> > is no longer met after Patch 1 of this series.
> 
> It has been historical behavior that all dma allocations are page
> aligned (including in fact smaller than page sized ones that get
> rounded up to a page size).  The documentation actually (incorretly)
> states an even stronger guarantee:
> 
> "The CPU virtual address and the DMA address are both
> guaranteed to be aligned to the smallest PAGE_SIZE order which
> is greater than or equal to the requested size.  This invariant
> exists (for example) to guarantee that if you allocate a chunk
> which is smaller than or equal to 64 kilobytes, the extent of the
> buffer you receive will not cross a 64K boundary."

Any thoughts on how that historical behavior should apply if
the DMA min_align_mask is non-zero, or the alloc_align_mask
parameter to swiotbl_tbl_map_single() is non-zero? As currently
used, alloc_align_mask is page aligned if the IOMMU granule is
>= PAGE_SIZE. But a non-zero min_align_mask could mandate
returning a buffer that is not page aligned. Perhaps do the
historical behavior only if alloc_align_mask and min_align_mask
are both zero?

Michael