Re: [PATCH 3.16 327/346] mm: workingset: fix crash in shadow node shrinker caused by replace_page_cache_page()

[Ben Hutchings]

On Mon, 2016-11-14 at 10:42 -0500, Johannes Weiner wrote:
> On Mon, Nov 14, 2016 at 12:14:20AM +0000, Ben Hutchings wrote:
> > 3.16.39-rc1 review patch.ÂÂIf anyone has any objections, please let me know.
> > 
> > ------------------
> > 
> > From: Johannes Weiner <hannes@xxxxxxxxxxx>
> > 
> > commit 22f2ac51b6d643666f4db093f13144f773ff3f3a upstream.
[...]
> The added sanity checks in this patch can crash kernels built with
> CONFIG_DEBUG_VM. While I doubt many people run 3.16 with that enabled,
> please also consider taking the following two changes. The first one
> changes the new VM_BUG_ONs to mere warnings, and the second patch
> addresses the underlying issue that triggered them in the first place.
> 
> 21f54ddae449 Using BUG_ON() as an assert() is _never_ acceptable
> d3798ae8c6f3 mm: filemap: don't plant shadow entries without radix tree node
> 
> Attaching the latter below, since it's drastically different than the
> upstream change, but a lot simpler because it predates the DAX stuff.
[...]

Thanks, I've added these two.

Ben.

-- 
Ben Hutchings
Lowery's Law:
ÂÂÂÂÂÂÂÂÂÂÂÂÂIf it jams, force it. If it breaks, it needed replacing
anyway.

Attachment: signature.asc
Description: This is a digitally signed message part