>> Um, silly question: has the mem_map[MAP_NR(page)].buffers != NULL
>> case gone away, as I don't see where it's created, and thus the
>> "race" isn't an issue? Or am I missing something?
David miller replied in no time flat (Sat, 26 Jul 1997 14:00:23 MDT)
> fs/buffer.c:grow_buffers()
Yes, but isn't that just for buffer-cache pages, so you know when the
page can be freed? Will filemap_write_page ever be called on such
pages? It's only called from filemap_swapout and filemap_sync_pte,
as far as I can tell (mm/filemap.c).
(Again, thanks to everyone for helping clarify this mysterious code.
I hope that I'm informing some people by having this conversation
on linux-kernel.)
-- -Colin