Skip to contentSkip to conversation

Minor fixups for hugetlb fault path

PATCH Applied Oscar Salvador 3 participants 5 messages May 9, 2024

Subsystems (2)
lore
Share & tools
lore

Oscar Salvador

author maintainer
#0 lore raw

cover Minor fixups for hugetlb fault path

Hi,

this series contains a couple of fixups for hugetlb_fault and hugetlb_wp respectively, where a VM_FAULT_SET_HINDEX call was missing.

I did not bother with a Fixes tag because the missing piece here is that we will not report to userspace the right extension of the faulty area by adjusting struct kernel_siginfo.si_addr_lsb, but I do not consider that to be a big issue because I assume that userspace already knows the size of the mapping anyway.

Oscar Salvador (2):
mm/hugetlb: Add missing VM_FAULT_SET_HINDEX in hugetlb_fault mm/hugetlb: Add missing VM_FAULT_SET_HINDEX in hugetlb_wp

 mm/hugetlb.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
Signature · 1 line
2.44.0
author maintainer
#1 lore raw

PATCH 1/2 mm/hugetlb: Add missing VM_FAULT_SET_HINDEX in hugetlb_fault

commit af19487f00f3 ("mm: make PTE_MARKER_SWAPIN_ERROR more general") added the code to handle pte_markers in hugetlb faulting path.
In case of an UFFD_POISON event, a PTE_MARKER_POISONED will be created and we will return VM_FAULT_HWPOISON_LARGE upon detecting that in the fault path. Add the missing VM_FAULT_SET_HINDEX, so the right si_addr_lsb will be passed to userspace to report the extension of the faulty area.

---
 mm/hugetlb.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
mm/hugetlb.c +2 -1
index 33d175add044..262456daf327 100644
@@ -6485,7 +6485,8 @@ vm_fault_t hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma,
 				pte_marker_get(pte_to_swp_entry(vmf.orig_pte));
 
 			if (marker & PTE_MARKER_POISONED) {
-				ret = VM_FAULT_HWPOISON_LARGE;
+				ret = VM_FAULT_HWPOISON_LARGE |
+				      VM_FAULT_SET_HINDEX(hstate_index(h));
 				goto out_mutex;
 			}
 		}
Signature · 1 line
2.44.0
author maintainer
#2 lore raw

PATCH 2/2 mm/hugetlb: Add missing VM_FAULT_SET_HINDEX in hugetlb_wp

In reply to Oscar Salvador

commit 1cb9dc4b475c ("mm: hwpoison: support recovery from HugePage copy-on-write faults") added support to use the mc variants when coping hugetlb pages on CoW faults.
Add the missing VM_FAULT_SET_HINDEX, so the right si_addr_lsb will be passed to userspace to report the extension of the faulty area.

---
 mm/hugetlb.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
mm/hugetlb.c +1 -1
index 262456daf327..6be78e7d4f6e 100644
@@ -6067,7 +6067,7 @@ static vm_fault_t hugetlb_wp(struct folio *pagecache_folio,
 		goto out_release_all;
 
 	if (copy_user_large_folio(new_folio, old_folio, vmf->real_address, vma)) {
-		ret = VM_FAULT_HWPOISON_LARGE;
+		ret = VM_FAULT_HWPOISON_LARGE | VM_FAULT_SET_HINDEX(hstate_index(h));
 		goto out_release_all;
 	}
 	__folio_mark_uptodate(new_folio);
Signature · 1 line
2.44.0
Acked-by on the v1 series
#3 lore raw
In reply to Oscar Salvador
Quoted 10 lines · Oscar Salvador

Hi,

this series contains a couple of fixups for hugetlb_fault and hugetlb_wp respectively, where a VM_FAULT_SET_HINDEX call was missing.

I did not bother with a Fixes tag because the missing piece here is that we will not report to userspace the right extension of the faulty area by adjusting struct kernel_siginfo.si_addr_lsb, but I do not consider that to be a big issue because I assume that userspace already knows the size of the mapping anyway.

Signature · 1 line
Peter Xu
#4 lore raw
Quoted 7 lines · Peter Xu
Quoted 5 lines · Oscar Salvador

Hi,

this series contains a couple of fixups for hugetlb_fault and hugetlb_wp respectively, where a VM_FAULT_SET_HINDEX call was missing.

Quoted 7 lines
Quoted 5 lines

I did not bother with a Fixes tag because the missing piece here is that we will not report to userspace the right extension of the faulty area by adjusting struct kernel_siginfo.si_addr_lsb, but I do not consider that to be a big issue because I assume that userspace already knows the size of the mapping anyway.

Looks correct to me as well. Thanks!

Quoted 4 lines

--
Peter Xu