On 13.03.2024 18:27, Andrew Cooper wrote:
> --- a/xen/drivers/passthrough/x86/iommu.c
> +++ b/xen/drivers/passthrough/x86/iommu.c
> @@ -641,7 +641,7 @@ struct page_info *iommu_alloc_pgtable(struct domain_iommu
> *hd,
> if ( contig_mask )
> {
> /* See pt-contig-markers.h for a description of the marker scheme. */
> - unsigned int i, shift = find_first_set_bit(contig_mask);
> + unsigned int i, shift = ffsl(contig_mask) - 1;
The need for subtracting 1 is why personally I dislike ffs() / ffsl() (and
why I think find_first_set_bit() and __ffs() (but no __ffsl()) were
introduced).
But what I first of all would like to have clarification on is what your
(perhaps just abstract at this point) plans are wrt ffz() / ffzl().
Potential side-by-side uses would be odd now, and would continue to be odd
if the difference in bit labeling was retained. Since we're switching to
a consolidated set of basic helpers, such an anomaly would better not
survive imo.
Jan
