Here is v2 of "enabling memory error handling on 1GB hugepage" patchset.
Major updates:
- (patch 3/9) I made pud_huge() and follow_huge_pud() aware of non-present
pud entry (based on Miaohe's comment).
- (patch 4/9 and patch 5/9) I extended the mechanism to save raw error
info to support multiple error subpages in a hugepage. Additionally,
I added a "unreliable" flag which prevents freeing hwpoison hugetlb
if any raw error info is lost.
- (patch 1/9 and 2/9) During testing some common cases for 1GB hugepage,
I found a few issues in existing code, so this series starts
with fixing them.
The remaining patches should have only minor updates since v1.
Patch dependency:
- "mm/memory-failure: disable unpoison once hw error happens"
(actually the conflict is not logical one, but adding MF_SIMULATED to
mf_flags conflicts with patch 6/9.)
v1: https://lore.kernel.org/linux-mm/20220602050631.771414-1-naoya.horiguchi@xxxxxxxxx/T/#u
Thanks,
Naoya Horiguchi
---
Summary:
Naoya Horiguchi (9):
mm/hugetlb: remove checking hstate_is_gigantic() in return_unused_surplus_pages()
mm/hugetlb: separate path for hwpoison entry in copy_hugetlb_page_range()
mm/hugetlb: make pud_huge() and huge_pud() aware of non-present pud entry
mm, hwpoison, hugetlb: support saving mechanism of raw error pages
mm, hwpoison: make unpoison aware of raw error info in hwpoisoned hugepage
mm, hwpoison: set PG_hwpoison for busy hugetlb pages
mm, hwpoison: make __page_handle_poison returns int
mm, hwpoison: skip raw hwpoison page in freeing 1GB hugepage
mm, hwpoison: enable memory error handling on 1GB hugepage
arch/x86/mm/hugetlbpage.c | 3 +-
include/linux/hugetlb.h | 13 ++++
include/linux/mm.h | 2 +-
include/linux/swapops.h | 9 +++
include/ras/ras_event.h | 1 -
mm/hugetlb.c | 78 ++++++++++++++--------
mm/memory-failure.c | 163 +++++++++++++++++++++++++++++++++++++---------
7 files changed, 209 insertions(+), 60 deletions(-)
