Replace this pattern in iommu.c:
cmpxchg64{,_local}(*ptr, 0, new) != 0
.. with the simpler and faster:
!try_cmpxchg64{,_local}(*ptr, &tmp, new)
The x86 CMPXCHG instruction returns success in the ZF flag, so this change
saves a compare after the CMPXCHG.
No functional change intended.
Signed-off-by: Uros Bizjak <[email protected]>
Cc: David Woodhouse <[email protected]>
Cc: Lu Baolu <[email protected]>
Cc: Joerg Roedel <[email protected]>
Cc: Will Deacon <[email protected]>
Cc: Robin Murphy <[email protected]>
---
drivers/iommu/intel/iommu.c | 9 +++++----
1 file changed, 5 insertions(+), 4 deletions(-)
diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
index 50eb9aed47cc..311073494f50 100644
--- a/drivers/iommu/intel/iommu.c
+++ b/drivers/iommu/intel/iommu.c
@@ -865,7 +865,7 @@ static struct dma_pte *pfn_to_dma_pte(struct dmar_domain *domain,
break;
if (!dma_pte_present(pte)) {
- uint64_t pteval;
+ uint64_t pteval, tmp;
tmp_page = alloc_pgtable_page(domain->nid, gfp);
@@ -877,7 +877,8 @@ static struct dma_pte *pfn_to_dma_pte(struct dmar_domain *domain,
if (domain->use_first_level)
pteval |= DMA_FL_PTE_XD | DMA_FL_PTE_US | DMA_FL_PTE_ACCESS;
- if (cmpxchg64(&pte->val, 0ULL, pteval))
+ tmp = 0ULL;
+ if (!try_cmpxchg64(&pte->val, &tmp, pteval))
/* Someone else set it while we were thinking; use theirs. */
free_pgtable_page(tmp_page);
else
@@ -2128,8 +2129,8 @@ __domain_mapping(struct dmar_domain *domain, unsigned long iov_pfn,
/* We don't need lock here, nobody else
* touches the iova range
*/
- tmp = cmpxchg64_local(&pte->val, 0ULL, pteval);
- if (tmp) {
+ tmp = 0ULL;
+ if (!try_cmpxchg64_local(&pte->val, &tmp, pteval)) {
static int dumps = 5;
pr_crit("ERROR: DMA PTE for vPFN 0x%lx already set (to %llx not %llx)\n",
iov_pfn, tmp, (unsigned long long)pteval);
--
2.42.0
On Sun, Apr 14, 2024 at 06:23:11PM +0200, Uros Bizjak wrote:
> Replace this pattern in iommu.c:
>
> cmpxchg64{,_local}(*ptr, 0, new) != 0
>
> ... with the simpler and faster:
>
> !try_cmpxchg64{,_local}(*ptr, &tmp, new)
>
> The x86 CMPXCHG instruction returns success in the ZF flag, so this change
> saves a compare after the CMPXCHG.
>
> No functional change intended.
>
> Signed-off-by: Uros Bizjak <[email protected]>
> Cc: David Woodhouse <[email protected]>
> Cc: Lu Baolu <[email protected]>
> Cc: Joerg Roedel <[email protected]>
> Cc: Will Deacon <[email protected]>
> Cc: Robin Murphy <[email protected]>
> ---
> drivers/iommu/intel/iommu.c | 9 +++++----
> 1 file changed, 5 insertions(+), 4 deletions(-)
Reviewed-by: Jason Gunthorpe <[email protected]>
Jason
On 4/15/24 12:23 AM, Uros Bizjak wrote:
> Replace this pattern in iommu.c:
>
> cmpxchg64{,_local}(*ptr, 0, new) != 0
>
> .. with the simpler and faster:
>
> !try_cmpxchg64{,_local}(*ptr, &tmp, new)
>
> The x86 CMPXCHG instruction returns success in the ZF flag, so this change
> saves a compare after the CMPXCHG.
>
> No functional change intended.
>
> Signed-off-by: Uros Bizjak<[email protected]>
> Cc: David Woodhouse<[email protected]>
> Cc: Lu Baolu<[email protected]>
> Cc: Joerg Roedel<[email protected]>
> Cc: Will Deacon<[email protected]>
> Cc: Robin Murphy<[email protected]>
> ---
> drivers/iommu/intel/iommu.c | 9 +++++----
> 1 file changed, 5 insertions(+), 4 deletions(-)
Patch has been queued for iommu/vt-d.
Best regards,
baolu