linux-kernel - Re: [ 133/180] cciss: fix incorrect scsi status reporting

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: <20121004232737.GA16406@1wt.eu>
Date:	Fri, 5 Oct 2012 01:27:37 +0200
From:	Willy Tarreau <w@....eu>
To:	Ben Hutchings <ben@...adent.org.uk>
Cc:	linux-kernel@...r.kernel.org, stable@...r.kernel.org,
	"Stephen M. Cameron" <scameron@...rdog.cce.hp.com>,
	Jens Axboe <axboe@...nel.dk>,
	Andrew Morton <akpm@...ux-foundation.org>,
	Linus Torvalds <torvalds@...ux-foundation.org>,
	Greg Kroah-Hartman <gregkh@...uxfoundation.org>
Subject: Re: [ 133/180] cciss: fix incorrect scsi status reporting

On Thu, Oct 04, 2012 at 11:49:50PM +0100, Ben Hutchings wrote:
> On Tue, Oct 02, 2012 at 12:54:10AM +0200, Willy Tarreau wrote:
> > 2.6.32-longterm review patch.  If anyone has any objections, please let me know.
> > 
> > ------------------
> > 
> > From: Stephen M. Cameron <scameron@...rdog.cce.hp.com>
> > 
> > commit b0cf0b118c90477d1a6811f2cd2307f6a5578362 upstream.
> > 
> > Delete code which sets SCSI status incorrectly as it's already been set
> > correctly above this incorrect code.  The bug was introduced in 2009 by
> > commit b0e15f6db111 ("cciss: fix typo that causes scsi status to be
> > lost.")
> 
> That commit was in 2.6.33 and it changed the '<' to '<<'.  It hasn't
> been backported to 2.6.32.y.

But apparently the status was already incorrect before the first patch
which tried to fix it first. I based myself on the comment from this
patch which says "it's already been set correctly above this incorrect code".

Above we find this :
        cmd->result = (DID_OK << 16);           /* host byte */
        cmd->result |= (COMMAND_COMPLETE << 8); /* msg byte */
        /* cmd->result |= (GOOD < 1); */                /* status byte */

        cmd->result |= (ei->ScsiStatus);

If such a status is valid, then I conclude that both the following forms
from the two previous versions are incorrect :

-       cmd->result |= (ei->ScsiStatus < 1);
+       cmd->result |= (ei->ScsiStatus << 1);

Hence I preferred to backport the fix and have the same code as in mainline
and newer versions which nobody has yet complained about.

> > -					cmd->result |= (ei->ScsiStatus < 1);
> [...]
> 
> Unless ei->ScsiStatus can be negative (it is declared as int, but
> I don't think it's actually meant to be negative), this statement
> is a no-op.

Hmmm I disagree here, the code above does exactly the same thing as :

        cmd->result |= !ei->ScsiStatus;

Which looks kind of strange to me after doing the exact opposite above,
since the result is that the lowest bit of cmd->result will always be
forced to 1 whatever ScsiStatus between 0 and 1. This might be what the
original patch author meant with "fix typo that causes scsi status to
be lost".

So I'd rather keep this fix.

Regards,
Willy

--
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/