linux-kernel - REISER4: tracking down opps in reiser4_do_readpage

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>] [day] [month] [year] [list]

Date:	Fri, 06 Apr 2007 10:53:02 +0400
From:	Ignatich <ignatich@...il.com>
To:	reiserfs-list@...esys.com
CC:	linux-kernel@...r.kernel.org
Subject: REISER4: tracking down opps in reiser4_do_readpage_extent

Hi,

I've been trying to narrow down  the possible sources of oppses in 
reiser4_do_readpage_extent (vs-1426, nikita-2688) patch by patch. 
Reverting reiser4-use-generic-file-read.patch & friends didn't have any 
effect, downgrading kernel from 2.6.21-rc5 to 2.6.20 too. Eventually it 
all came down to 0046-reiser4-drop-unused-semaphores.patch. I don't have 
a reliable way to trigger the bug, but one of my vmware sandboxes with 
reiser4 root always oppses when I do emerge binutils right after boot. 
With this patch reverted it no longer happens. **

You can find original patch at:
http://laurent.riffard.free.fr/reiser4/reiser4-for-2.6.21-rc1/broken-out/

I attached patch that adds mutex_write back to reiser4_inode, but as 
real mutex now instead of semaphore. Cryptcompress rw semaphore is not 
added back. Updated reiser4-fix-write_extent.patch with proper 
identation is also attached.

Perhaps this is not a proper fix and only hides real culprit, but at 
least I do not see oppses now. I do not yet know if this also solves 
problem with zeroed out files.

    Max

View attachment "reiser4-add-write-mutex.patch" of type "text/plain" (4766 bytes)

View attachment "reiser4-fix-write_extent.patch" of type "text/plain" (3746 bytes)