Hi everyone, I updated the patch for the issue reported by Muchun Song in [1]. I separated into two patches (to reduce the size of core fix for stable): patch 1/2 focuses on preventing VM_BUG_ON_PAGE() in the reported race. Patch 2/2 includes some code improvement like calling retry code, revising the function comment, and some refactoring. Any comment and testing would be appreciated. Thanks, Naoya Horiguchi [1] https://lore.kernel.org/linux-mm/20210421060259.67554-1-songmuchun@xxxxxxxxxxxxx/T/ --- Summary: Naoya Horiguchi (2): mm,hwpoison: fix race with compound page allocation mm,hwpoison: make get_hwpoison_page call get_any_page() mm/memory-failure.c | 169 +++++++++++++++++++++++++++++----------------------- 1 file changed, 96 insertions(+), 73 deletions(-)
