To be honest, I have a question on how this works.
As the comment says, we need to call pmd_read_atomic before using
pte_offset_map_lock to avoid data corruption.
For example, in function swapin_walk_pmd_entry:
pmd_none_or_trans_huge_or_clear_bad(pmd)
pmd_read_atomic(pmd)
The function involved should be pte_offset_map_lock and we never have
function pmd_offset_map_lock defined.
Fixes: 26c191788f18 ("mm: pmd_read_atomic: fix 32bit PAE pmd walk vs
pmd_populate SMP race conditio")
Signed-off-by: Wei Yang
---
Hope my understanding is correct.
---
arch/x86/include
2 matches
Mail list logo
