The patch titled
Subject: mm: handle some PMD faults under the VMA lock
has been added to the -mm mm-unstable branch. Its filename is
mm-handle-some-pmd-faults-under-the-vma-lock.patch
This patch will shortly appear at
https://git.kernel.org/pub/scm/linux/kernel/git/akpm/25-new.git/tree/patches/mm-handle-some-pmd-faults-under-the-vma-lock.patch
This patch will later appear in the mm-unstable branch at
git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm
Before you just go and hit "reply", please:
a) Consider who else should be cc'ed
b) Prefer to cc a suitable mailing list as well
c) Ideally: find the original patch on the mailing list and do a
reply-to-all to that, adding suitable additional cc's
*** Remember to use Documentation/process/submit-checklist.rst when testing your code ***
The -mm tree is included into linux-next via the mm-everything
branch at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm
and is updated there every 2-3 working days
------------------------------------------------------
From: "Matthew Wilcox (Oracle)" <willy@xxxxxxxxxxxxx>
Subject: mm: handle some PMD faults under the VMA lock
Date: Mon, 24 Jul 2023 19:54:05 +0100
Push the VMA_LOCK check down from __handle_mm_fault() to
handle_pte_fault(). Once again, we refuse to call ->huge_fault() with the
VMA lock held, but we will wait for a PMD migration entry with the VMA
lock held, handle NUMA migration and set the accessed bit. We were
already doing this for anonymous VMAs, so it should be safe.
Link: https://lkml.kernel.org/r/20230724185410.1124082-6-willy@xxxxxxxxxxxxx
Signed-off-by: Matthew Wilcox (Oracle) <willy@xxxxxxxxxxxxx>
Cc: Arjun Roy <arjunroy@xxxxxxxxxx>
Cc: Eric Dumazet <edumazet@xxxxxxxxxx>
Cc: Punit Agrawal <punit.agrawal@xxxxxxxxxxxxx>
Cc: Suren Baghdasaryan <surenb@xxxxxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
---
mm/memory.c | 39 +++++++++++++++++++++++++--------------
1 file changed, 25 insertions(+), 14 deletions(-)
--- a/mm/memory.c~mm-handle-some-pmd-faults-under-the-vma-lock
+++ a/mm/memory.c
@@ -4947,36 +4947,47 @@ out_map:
static inline vm_fault_t create_huge_pmd(struct vm_fault *vmf)
{
- if (vma_is_anonymous(vmf->vma))
+ struct vm_area_struct *vma = vmf->vma;
+ if (vma_is_anonymous(vma))
return do_huge_pmd_anonymous_page(vmf);
- if (vmf->vma->vm_ops->huge_fault)
- return vmf->vma->vm_ops->huge_fault(vmf, PE_SIZE_PMD);
+ if (vma->vm_ops->huge_fault) {
+ if (vmf->flags & FAULT_FLAG_VMA_LOCK) {
+ vma_end_read(vma);
+ return VM_FAULT_RETRY;
+ }
+ return vma->vm_ops->huge_fault(vmf, PE_SIZE_PMD);
+ }
return VM_FAULT_FALLBACK;
}
/* `inline' is required to avoid gcc 4.1.2 build error */
static inline vm_fault_t wp_huge_pmd(struct vm_fault *vmf)
{
+ struct vm_area_struct *vma = vmf->vma;
const bool unshare = vmf->flags & FAULT_FLAG_UNSHARE;
vm_fault_t ret;
- if (vma_is_anonymous(vmf->vma)) {
+ if (vma_is_anonymous(vma)) {
if (likely(!unshare) &&
- userfaultfd_huge_pmd_wp(vmf->vma, vmf->orig_pmd))
+ userfaultfd_huge_pmd_wp(vma, vmf->orig_pmd))
return handle_userfault(vmf, VM_UFFD_WP);
return do_huge_pmd_wp_page(vmf);
}
- if (vmf->vma->vm_flags & (VM_SHARED | VM_MAYSHARE)) {
- if (vmf->vma->vm_ops->huge_fault) {
- ret = vmf->vma->vm_ops->huge_fault(vmf, PE_SIZE_PMD);
+ if (vma->vm_flags & (VM_SHARED | VM_MAYSHARE)) {
+ if (vma->vm_ops->huge_fault) {
+ if (vmf->flags & FAULT_FLAG_VMA_LOCK) {
+ vma_end_read(vma);
+ return VM_FAULT_RETRY;
+ }
+ ret = vma->vm_ops->huge_fault(vmf, PE_SIZE_PMD);
if (!(ret & VM_FAULT_FALLBACK))
return ret;
}
}
/* COW or write-notify handled on pte level: split pmd. */
- __split_huge_pmd(vmf->vma, vmf->pmd, vmf->address, false, NULL);
+ __split_huge_pmd(vma, vmf->pmd, vmf->address, false, NULL);
return VM_FAULT_FALLBACK;
}
@@ -5047,6 +5058,11 @@ static vm_fault_t handle_pte_fault(struc
{
pte_t entry;
+ if ((vmf->flags & FAULT_FLAG_VMA_LOCK) && !vma_is_anonymous(vmf->vma)) {
+ vma_end_read(vmf->vma);
+ return VM_FAULT_RETRY;
+ }
+
if (unlikely(pmd_none(*vmf->pmd))) {
/*
* Leave __pte_alloc() until later: because vm_ops->fault may
@@ -5186,11 +5202,6 @@ retry_pud:
if (pud_trans_unstable(vmf.pud))
goto retry_pud;
- if ((flags & FAULT_FLAG_VMA_LOCK) && !vma_is_anonymous(vma)) {
- vma_end_read(vma);
- return VM_FAULT_RETRY;
- }
-
if (pmd_none(*vmf.pmd) &&
hugepage_vma_check(vma, vm_flags, false, true, true)) {
ret = create_huge_pmd(&vmf);
_
Patches currently in -mm which might be from willy@xxxxxxxxxxxxx are
rmap-pass-the-folio-to-__page_check_anon_rmap.patch
highmem-add-memcpy_to_folio-and-memcpy_from_folio.patch
affs-convert-affs_symlink_read_folio-to-use-the-folio.patch
affs-convert-data-read-and-write-to-use-folios.patch
migrate-use-folio_set_bh-instead-of-set_bh_page.patch
ntfs3-convert-ntfs_get_block_vbo-to-use-a-folio.patch
jbd2-use-a-folio-in-jbd2_journal_write_metadata_buffer.patch
buffer-remove-set_bh_page.patch
zswap-make-zswap_store-take-a-folio.patch
memcg-convert-get_obj_cgroup_from_page-to-get_obj_cgroup_from_folio.patch
swap-remove-some-calls-to-compound_head-in-swap_readpage.patch
zswap-make-zswap_load-take-a-folio.patch
mm-remove-config_per_vma_lock-ifdefs.patch
mm-allow-per-vma-locks-on-file-backed-vmas.patch
mm-move-fault_flag_vma_lock-check-from-handle_mm_fault.patch
mm-handle-pud-faults-under-the-vma-lock.patch
mm-handle-some-pmd-faults-under-the-vma-lock.patch
mm-move-fault_flag_vma_lock-check-down-in-handle_pte_fault.patch
mm-move-fault_flag_vma_lock-check-down-from-do_fault.patch
mm-run-the-fault-around-code-under-the-vma-lock.patch
mm-handle-swap-and-numa-pte-faults-under-the-vma-lock.patch
mm-handle-faults-that-merely-update-the-accessed-bit-under-the-vma-lock.patch
