Re: [patch 2/3] mm: fault vs invalidate/truncate race fix

[Nick Piggin <nickpiggin@yahoo.com.au>]

Andrew Morton wrote:

>On Sat,  7 Oct 2006 15:06:21 +0200 (CEST)
>Nick Piggin <npiggin@suse.de> wrote:
>
>
>>Fix the race between invalidate_inode_pages and do_no_page.
>>
>
>- In do_no_page() there's a `goto retry' where we appear to have
>  forgotten to (conditionally) unlock the page.
>

Hmm, the retry should be gone - it was only there for the
seqlock-ish truncate race code.

>- In do_no_page() the COW-break code seem to have forgotten to
>  (conditionally) unlock the page which it just COWed?
>

It keeps the 'nopage_page' around and unlocks it at the end.
Last time I looked, this is required because truncate wants to
unmap 'even_cows', so we must hold the pagecache page locked
while instantiating the mapping on the cow page.

>- In do_no_page(), the unlock_page() which _is_ there doesn't test
>  VM_CAN_INVALIDATE before deciding to unlock the page.
>

It does a conditional lock if !VM_CAN_INVALIDATE based on a
suggestion from Hugh. I don't disagree with that, but it can
go away in the next patch as we won't be calling into
->page_mkwrite (if that callback can be implemented with ->fault).

--

Send instant messages to your online friends http://au.messenger.yahoo.com