linux-kernel - CIFS 'cache=strict' breaks 'pv' utility

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:	Wed, 20 Feb 2013 11:35:13 -0500
From:	Jody - Tritech Computer Solutions <jody@...ritech.com>
To:	linux-kernel@...r.kernel.org, andrew.wood@...rch.com
Subject: CIFS 'cache=strict' breaks 'pv' utility

Greetings,

After the switch from 'cache=loose' to 'cache=strict' as the default 
CIFS cache mount option, I am unable to use the 'pv' utility in place of 
'cat' to read files beyond a certain size over CIFS. Mounting with 
'cache=loose' works around the problem, but does not change the 
breakage. If I mount with 'cache=strict' on 3.6.11 (the last stable 
kernel that works out-of-the-box) the breakage occurs again.

I hit this problem on 32-bit and 64-bit kernels, pv-1.2.0 and the latest 
pv-1.4.6, and the failure is tied to 'cache=strict' on all kernels from 
3.6.11 through 3.8.

My testing with a command such as 'pv /mnt/cifs/file.tar.xz | wc -c' 
indicates that 'pv' outputs the data up to a certain power-of-two point 
and then fails with "pv: (stdin): read failed: Input/output error". I 
have seen data stop transferring at 524288, 1048576, and 2097152 bytes 
when I tried feeding it various file sizes.

I am able to use 'pv' with files on CIFS mounted as 'cache=strict' if I 
reduce the buffer size on the command line (the -B xxx option) to 131072 
or lower, while setting a buffer of 262144 or higher always fails.

I am sending this to the LKML as well as the 'pv' maintainer, since I do 
not know the scope of the issue. If this exposes an unknown bug in the 
kernel CIFS driver, I don't want it to fly under the radar. I am 
available for testing and reporting as needed.

I don't subscribe to LKML, so please CC me on any responses.

--

Jody Lee Bruchon
Tritech Computer Solutions
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/