From: Linus Torvalds <torvalds@osdl.org>
To: Rafal Skoczylas <nils@secprog.org>
Cc: Kernel Mailing List <linux-kernel@vger.kernel.org>,
William Lee Irwin III <wli@holomorphy.com>
Subject: Re: 2.6.test11 bug
Date: Sun, 7 Dec 2003 21:17:35 -0800 (PST) [thread overview]
Message-ID: <Pine.LNX.4.58.0312072100250.13236@home.osdl.org> (raw)
In-Reply-To: <20031208034631.GA14081@secprog.org>
On Mon, 8 Dec 2003, Rafal Skoczylas wrote:
>
> --- But sometimes it get things like this:
>
> Unable to handle kernel paging request at virtual address 5a85fb5c
> printing eip:
> c011e6c4
> *pde = 00000000
> Oops: 0002 [#1]
> CPU: 0
> EIP: 0060:[remove_wait_queue+36/112] Not tainted
> EFLAGS: 00010002
> EIP is at remove_wait_queue+0x24/0x70
> eax: defb4000 ebx: da85fb58 ecx: 5a85fb58 edx: db0468b0
> esi: db0468bc edi: 00000292 ebp: defb5fa0 esp: defb5f58
> Trace:
> [poll_freewait+36/80] poll_freewait+0x24/0x50
> [sys_poll+581/656] sys_poll+0x245/0x290
> [__pollwait+0/208] __pollwait+0x0/0xd0
> [syscall_call+7/11] syscall_call+0x7/0xb
Ahh.. Btw, this looks like a single-bit error.
In particular, this is "list_del()" in remove_wait_queue(), and it's
list_del() on a list that usually only has a single entry (ie a wait
queue). In particular, it's the "prev->next = next" part, where "prev" is
5a85fb58 and next is da85fb58.
Now, with a single entry, prev and next _should_ be the same (they both
point to the head). And they are - except for a single bit error. The
difference is the high bit of the word. Interesting.
You get an oops because that 5a85fb58 _should_ be da85fb58. Notice?
> If there is any important information missing, feel free to ask.
It could be bad memory. We even know the address that is bad: it's
(%esi+4), ie bit 31 of the word at physical address 0x1b0468f0.
However, if you don't see random SIGSEGV's while compiling etc issues, it
doesnt' sound like flaky RAM.
I'm wondering if there is some bit operation out there somewhere with a
wild pointer?
Rafal - how consistent is the second form of the oops? Have you seen that
trace more than once? Might it actually be "poll()" itself that has some
bad behaviour?
Linus
next prev parent reply other threads:[~2003-12-08 5:17 UTC|newest]
Thread overview: 11+ messages / expand[flat|nested] mbox.gz Atom feed top
2003-12-08 3:46 2.6.test11 bug Rafal Skoczylas
2003-12-08 4:17 ` William Lee Irwin III
2003-12-08 5:17 ` Linus Torvalds [this message]
2003-12-08 9:02 ` Xavier Bestel
2003-12-08 16:27 ` Rafal Skoczylas
[not found] ` <20031208161742.GB9087@secprog.org>
[not found] ` <Pine.LNX.4.58.0312080848560.13236@home.osdl.org>
2003-12-08 17:12 ` 2.6.test11 bug Linus Torvalds
[not found] ` <20031209194827.GA22265@secprog.org>
[not found] ` <Pine.LNX.4.58.0312091221440.21456@home.osdl.org>
2003-12-09 22:31 ` Rafal Skoczylas
2003-12-09 23:26 ` Linus Torvalds
[not found] <10kzo-7mZ-11@gated-at.bofh.it>
[not found] ` <10m88-2wd-1@gated-at.bofh.it>
[not found] ` <10xdJ-28r-23@gated-at.bofh.it>
[not found] ` <10xdJ-28r-25@gated-at.bofh.it>
[not found] ` <10xdJ-28r-21@gated-at.bofh.it>
2003-12-08 22:28 ` Rafal Skoczylas
-- strict thread matches above, loose matches on Subject: below --
2003-12-08 2:24 Gordon Cormack
2003-12-08 2:37 ` Linus Torvalds
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=Pine.LNX.4.58.0312072100250.13236@home.osdl.org \
--to=torvalds@osdl.org \
--cc=linux-kernel@vger.kernel.org \
--cc=nils@secprog.org \
--cc=wli@holomorphy.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).