Re: Heads up Linux 2.6.38-rc4 compile problems.

From: Linus Torvalds <torvalds@linux-foundation.org>
To: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Alex Riesen <raa.lkml@gmail.com>,
	David Miller <davem@davemloft.net>,
	Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
	Andrew Morton <akpm@linux-foundation.org>
Subject: Re: Heads up Linux 2.6.38-rc4 compile problems.
Date: Mon, 14 Feb 2011 07:26:13 -0800	[thread overview]
Message-ID: <AANLkTim2i1NTxLFszh7+DPWmpp35H8YEfFR+o-AWoO9D@mail.gmail.com> (raw)
In-Reply-To: <m1fwrrtc0m.fsf@fess.ebiederm.org>

On Sun, Feb 13, 2011 at 9:34 PM, Eric W. Biederman
<ebiederm@xmission.com> wrote:
>
> And for completeness.  When I was rebooting v2.6.38-rc4 to start running
> 795abaf1e4e188c4171e3cd3dbb11a9fcacaf505  I hit this.

.. but this was while still running the older kernel, right?

> <2>kernel BUG at mm/filemap.c:125!

I suspect this is "normal" after page table corruption. Any page that
was mapped but overwritten by the corruption would never get unmapped
(since it can't be found in the page tables), and then you trigger the

  BUG_ON(page_mapped(page));

in __remove_from_page_cache() at umount time. Your register state
shows that %eax is 0, and that's the count that we tested
("page->_mapcount" is -1 when there are no mappings, so you have one
lost mapping reference to that page).

So that oops isn't all that interesting, I'm afraid.

                             Linus