linux-kernel - Re: Kernel Oops on enabling CONFIG_LOCK

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20110722073417.GB2622@htj.dyndns.org>
Date:	Fri, 22 Jul 2011 09:34:17 +0200
From:	Tejun Heo <tj@...nel.org>
To:	Shreshtha <shreshthakumar@...il.com>
Cc:	linux-kernel@...r.kernel.org,
	Andrew Morton <akpm@...ux-foundation.org>
Subject: Re: Kernel Oops on enabling CONFIG_LOCK_STAT

Hello,

On Fri, Jul 22, 2011 at 08:10:54AM +0200, Tejun Heo wrote:
> Hrmm... so it's pcpu_populate_chunk() failure path.  Can you please
> attach full kernel log?  Attaching full kernel log and including a bit
> of hardware details is generally a good idea when reporting a bug.
> 
> It's most likely there's a bug in the code which rolls back from
> partial allocation after encountering alloc failure in the middle.
> I'll take a deeper look there and report what I find.

The code looks correct and seems to behave correct under induced
error.  Can you please apply the following patch and see whether the
problem goes away?

diff --git a/mm/percpu-vm.c b/mm/percpu-vm.c
index ea53496..53eae44 100644
--- a/mm/percpu-vm.c
+++ b/mm/percpu-vm.c
@@ -347,6 +347,7 @@ clear:
 	return 0;
 
 err_unmap:
+	pcpu_post_map_flush(chunk, page_start, unmap_end);
 	pcpu_pre_unmap_flush(chunk, page_start, unmap_end);
 	pcpu_for_each_unpop_region(chunk, rs, re, page_start, unmap_end)
 		pcpu_unmap_pages(chunk, pages, populated, rs, re);

If not, can you please apply the attached debug patch, trigger the
problem and post the log?

Thank you.

-- 
tejun

View attachment "pcpu-debug.patch" of type "text/plain" (3262 bytes)