mm/migrate: Fix wrongly apply write bit after mkdirty on sparc64
Subsystems (4)
PATCH 1/1 mm/migrate: Fix wrongly apply write bit after mkdirty on sparc64
Nick Bowler reported another sparc64 breakage after the young/dirty persistent work for page migration (per "Link:" below). That's after a similar report [2].
It turns out page migration was overlooked, and it wasn't failing before because page migration was not enabled in the initial report test environment.
David proposed another way [2] to fix this from sparc64 side, but that patch didn't land somehow. Neither did I check whether there's any other arch that has similar issues.
Let's fix it for now as simple as moving the write bit handling to be after dirty, like what we did before.
Note: this is based on mm-unstable, because the breakage was since 6.1 and we're at a very late stage of 6.2 (-rc8), so I assume for this specific case we should target this at 6.3.
[1] https://lore.kernel.org/all/[email protected]/
[2] https://lore.kernel.org/all/[email protected]/
--- mm/huge_memory.c | 6 ++++-- mm/migrate.c | 2 ++ 2 files changed, 6 insertions(+), 2 deletions(-)
mm/huge_memory.c +4 -2
@@ -3274,8 +3274,6 @@ void remove_migration_pmd(struct page_vma_mapped_walk *pvmw, struct page *new) pmde =mk_huge_pmd (new,READ_ONCE (vma->vm_page_prot));if (pmd_swp_soft_dirty (*pvmw->pmd)) pmde =pmd_mksoft_dirty (pmde); -if (is_writable_migration_entry (entry)) - pmde =maybe_pmd_mkwrite (pmde, vma);if (pmd_swp_uffd_wp (*pvmw->pmd)) pmde =pmd_mkuffd_wp (pmde);if (!is_migration_entry_young (entry))
@@ -3283,6 +3281,10 @@ void remove_migration_pmd(struct page_vma_mapped_walk *pvmw, struct page *new)/* NOTE: this may contain setting soft-dirty on some archs */ if (PageDirty (new) &&is_migration_entry_dirty (entry)) pmde =pmd_mkdirty (pmde); +if (is_writable_migration_entry (entry)) + pmde =maybe_pmd_mkwrite (pmde, vma); +else + pmde =pmd_wrprotect (pmde);if (PageAnon (new)) {rmap_t rmap_flags =RMAP_COMPOUND ;
mm/migrate.c +2 -0
@@ -225,6 +225,8 @@ static bool remove_migration_pte(struct folio *folio, pte =maybe_mkwrite (pte, vma); else if (pte_swp_uffd_wp(*pvmw.pte)) pte =pte_mkuffd_wp (pte); + else + pte =pte_wrprotect (pte);if (folio_test_anon (folio) && !is_readable_migration_entry (entry)) rmap_flags |=RMAP_EXCLUSIVE ;
Signature · 1 line
2.39.1
Quoted 9 lines · Peter Xu
Nick Bowler reported another sparc64 breakage after the young/dirty persistent work for page migration (per "Link:" below). That's after a similar report [2].
It turns out page migration was overlooked, and it wasn't failing before because page migration was not enabled in the initial report test environment.
David proposed another way [2] to fix this from sparc64 side, but that patch didn't land somehow.
I pinged another time and asked Andrew to pick it up eventually.
Signature · 3 lines
Thanks, David / dhildenb
Andrew Morton
maintainerQuoted 5 lines · On Thu, 16 Feb 2023 16:37:19 +0100 David Hildenbrand <[email protected]> wrote:
Quoted 4 lines · Peter Xu
Nick Bowler reported another sparc64 breakage after the young/dirty persistent work for page migration (per "Link:" below). That's after a similar report [2].
Quoted 8 lines
Quoted 5 lines
It turns out page migration was overlooked, and it wasn't failing before because page migration was not enabled in the initial report test environment.
David proposed another way [2] to fix this from sparc64 side, but that patch didn't land somehow.
I pinged another time and asked Andrew to pick it up eventually.
urgh, sorry, I'd have filtered that out as a sparc patch.
Quoted 3 lines · Peter Xu
Nick Bowler reported another sparc64 breakage after the young/dirty persistent work for page migration (per "Link:" below). That's after a similar report [2].
Thx for handling this.
Quoted 10 lines
[...]
Note: this is based on mm-unstable, because the breakage was since 6.1 and we're at a very late stage of 6.2 (-rc8), so I assume for this specific case we should target this at 6.3.
[1] https://lore.kernel.org/all/[email protected]/
[2] https://lore.kernel.org/all/[email protected]/
Not that it matters much, but feel free to use this instead:
Then things don't depend on me (in case I ever get help with my cat herding job). And it also make it even more obvious that this patch fixes a regression to anyone who handles it downstream.
Quoted 1 line
Fixes 2e3468778dbe ("mm: remember young/dirty bit for page migrations")
That's a commit from 6.1, hence this should likely have:
[no, a fixes tag alone does not suffice, see docs]
Quoted 3 lines
Ciao, Thorsten (wearing his 'the Linux kernel's regression tracker' hat) --
Everything you wanna know about Linux kernel regression tracking:
https://linux-regtracking.leemhuis.info/about/#tldr
If I did something stupid, please tell me, as explained on that page.
Quoted 18 lines · Linux regression tracking (Thorsten Leemhuis)
Quoted 3 lines · Peter Xu
Nick Bowler reported another sparc64 breakage after the young/dirty persistent work for page migration (per "Link:" below). That's after a similar report [2].
Thx for handling this.
Quoted 10 lines
[...]
Note: this is based on mm-unstable, because the breakage was since 6.1 and we're at a very late stage of 6.2 (-rc8), so I assume for this specific case we should target this at 6.3.
[1] https://lore.kernel.org/all/[email protected]/
[2] https://lore.kernel.org/all/[email protected]/
Quoted 7 lines
Not that it matters much, but feel free to use this instead:
Then things don't depend on me (in case I ever get help with my cat herding job). And it also make it even more obvious that this patch fixes a regression to anyone who handles it downstream.
Sure.
Quoted 8 lines
Quoted 2 lines
Fixes 2e3468778dbe ("mm: remember young/dirty bit for page migrations")That's a commit from 6.1, hence this should likely have:
Cc [email protected] # 6.1.y[no, a fixes tag alone does not suffice, see docs]
Oops, I just forgot that. We definitely need to cc:stable. I'll make sure it's there if there's a new version, or I hope Andrew could help to attach otherwise.
Thanks,
Signature · 1 line
Peter Xu
Quoted 61 lines · Peter Xu
Nick Bowler reported another sparc64 breakage after the young/dirty persistent work for page migration (per "Link:" below). That's after a similar report [2].
It turns out page migration was overlooked, and it wasn't failing before because page migration was not enabled in the initial report test environment.
David proposed another way [2] to fix this from sparc64 side, but that patch didn't land somehow. Neither did I check whether there's any other arch that has similar issues.
Let's fix it for now as simple as moving the write bit handling to be after dirty, like what we did before.
Note: this is based on mm-unstable, because the breakage was since 6.1 and we're at a very late stage of 6.2 (-rc8), so I assume for this specific case we should target this at 6.3.
[1] https://lore.kernel.org/all/[email protected]/
[2] https://lore.kernel.org/all/[email protected]/Fixes 2e3468778dbe ("mm: remember young/dirty bit for page migrations")Link https://lore.kernel.org/all/CADyTPExpEqaJiMGoV+Z6xVgL50ZoMJg49B10LcZ=8eg19u34BA@mail.gmail.com/ open hereReported-by Nick Bowler <[email protected]>Tested-by Nick Bowler <[email protected]>Signed-off-by Peter Xu <[email protected]>--- mm/huge_memory.c | 6 ++++-- mm/migrate.c | 2 ++ 2 files changed, 6 insertions(+), 2 deletions(-)mm/huge_memory.c +4
-2@@ -3274,8 +3274,6 @@ void remove_migration_pmd(struct page_vma_mapped_walk *pvmw, struct page *new) pmde =mk_huge_pmd (new,READ_ONCE (vma->vm_page_prot));if (pmd_swp_soft_dirty (*pvmw->pmd)) pmde =pmd_mksoft_dirty (pmde); -if (is_writable_migration_entry (entry)) - pmde =maybe_pmd_mkwrite (pmde, vma);if (pmd_swp_uffd_wp (*pvmw->pmd)) pmde =pmd_mkuffd_wp (pmde);if (!is_migration_entry_young (entry))@@ -3283,6 +3281,10 @@ void remove_migration_pmd(struct page_vma_mapped_walk *pvmw, struct page *new)/* NOTE: this may contain setting soft-dirty on some archs */ if (PageDirty (new) &&is_migration_entry_dirty (entry)) pmde =pmd_mkdirty (pmde); +if (is_writable_migration_entry (entry)) + pmde =maybe_pmd_mkwrite (pmde, vma); +else + pmde =pmd_wrprotect (pmde);if (PageAnon (new)) {rmap_t rmap_flags =RMAP_COMPOUND ;mm/migrate.c +0
-0
mm/migrate.c static bool remove_migration_pte(struct folio *folio, · line 232
Quoted 8 lines
mm/migrate.c +2
-0@@ -225,6 +225,8 @@ static bool remove_migration_pte(struct folio *folio, pte =maybe_mkwrite (pte, vma); else if (pte_swp_uffd_wp(*pvmw.pte)) pte =pte_mkuffd_wp (pte); + else + pte =pte_wrprotect (pte);if (folio_test_anon (folio) && !is_readable_migration_entry (entry)) rmap_flags |=RMAP_EXCLUSIVE ;
I'd really rather focus on fixing the root cause instead, anyhow if my patch won't make it:
Signature · 3 lines
Thanks, David / dhildenb