[patch 0/6] fault vs truncate/invalidate race fix

!MAILaRCHIVE_VOTE_RePLACE
Previous message: [thread] [date] [author]
Next message: [thread] [date] [author]
To: Linux Memory Management <linux-mm@...>, Andrew Morton <akpm@...>
Cc: Linux Kernel <linux-kernel@...>, Nick Piggin <npiggin@...>, Benjamin Herrenschmidt <benh@...>
Date: Wednesday, February 21, 2007 - 12:49 am

The following set of patches are based on current git.

These fix the fault vs invalidate and fault vs truncate_range race for
filemap_nopage mappings, plus those and fault vs truncate race for nonlinear
mappings.

These patches fix silent data corruption that we've had several people hitting
in SUSE kernels. Our kernels have similar patches to lock the page over page
fault, and no problem.

I've also got rid of the horrible populate API, and integrated nonlinear pages
properly with the page fault path.

Downside is that this adds one more vector through which the buffered write
deadlock can occur. However this is just a very tiny one (pte being unmapped
for reclaim), compared to all the other ways that deadlock can occur (unmap,
reclaim, truncate, invalidate). I doubt it will be noticable. At any rate, it
is better than data corruption.

I hope these can get merged (at least into -mm) soon.

Thanks,
Nick

--
SuSE Labs

-
Previous message: [thread] [date] [author]
Next message: [thread] [date] [author]

Messages in current thread:
[patch 0/6] fault vs truncate/invalidate race fix, Nick Piggin, (Wed Feb 21, 12:49 am)
Re: [patch 0/6] fault vs truncate/invalidate race fix, Dave Airlie, (Tue Feb 27, 12:36 am)
Re: [patch 0/6] fault vs truncate/invalidate race fix, Andrew Morton, (Tue Feb 27, 1:32 am)
Re: [patch 0/6] fault vs truncate/invalidate race fix, Nick Piggin, (Tue Feb 27, 4:50 am)
Re: [patch 0/6] fault vs truncate/invalidate race fix, Dave Airlie, (Tue Feb 27, 2:26 am)
Re: [patch 0/6] fault vs truncate/invalidate race fix, Benjamin Herrenschmidt, (Tue Feb 27, 2:54 am)
Re: [patch 0/6] fault vs truncate/invalidate race fix, Dave Airlie, (Sun Mar 18, 7:13 pm)
Re: [patch 4/6] mm: merge populate and nopage into fault (fi..., Benjamin Herrenschmidt, (Wed Mar 7, 6:05 am)
Re: [patch 4/6] mm: merge populate and nopage into fault (fi..., Benjamin Herrenschmidt, (Wed Mar 7, 6:46 am)
[rfc][patch 7/6] mm: merge page_mkwrite, Nick Piggin, (Wed Mar 7, 6:30 am)
[RFC][PATCH] mm: fix page_mkclean() vs non-linear vmas, Peter Zijlstra, (Wed Mar 7, 12:58 pm)
[patch 6/6] mm: remove legacy cruft, Nick Piggin, (Wed Feb 21, 12:50 am)
[patch 5/6] mm: merge nopfn into fault, Nick Piggin, (Wed Feb 21, 12:50 am)
Re: [patch 5/6] mm: merge nopfn into fault, Nick Piggin, (Wed Feb 21, 1:13 am)
[patch 2/6] mm: simplify filemap_nopage, Nick Piggin, (Wed Feb 21, 12:49 am)