Messages in this thread | | | Date | Mon, 3 May 2010 20:13:40 +0200 | From | Andrea Arcangeli <> | Subject | Re: [PATCH 1/2] mm: Take all anon_vma locks in anon_vma_lock |
| |
> > Btw, Mel's patch doesn't really match the description of 2/2. 2/2 says > > that all pages must always be findable in rmap. Mel's patch seems to > > explicitly say "we want to ignore that thing that is busy for execve". Are > > we just avoiding a BUG_ON()? Is perhaps the BUG_ON() buggy? > > I have no good answer to this question. > > Mel? Andrea?
If try_to_unmap is allowed to establish the migration pte, then such pte has to remain reachable through rmap_walk at all times after that, or migration_entry_wait will crash because it notices the page has been migrated already (PG_lock not set) but there is still a migration pte established. (remove_migration_pte like split_huge_page isn't allowed to fail finding all migration ptes mapping a page during the rmap walk)
It's not false positive BUG_ON if that's what you mean, removing the BUG_ON would still lead to infinite hang waiting on a migration pte that shouldn't be there anymore.
| |