hpaio backend writes to fd 0, breaks saned/net

[Mark Purcell]

Forwarded from: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=495005

From: Julien BLACHE <email address hidden>
To: <email address hidden>
Cc: <email address hidden>, <email address hidden>, <email address hidden>
Subject: Re: Bug#495005: xsane segfaults
Date: Wed, 13 Aug 2008 21:36:51 +0200

reassign 495005 hplip 2.8.6-2
retitle 495005 hpaio backend writes to fd 0, breaks saned/net
severity 495005 important
thanks

Julien Danjou <email address hidden> wrote:

Hi hplip folks,

>> That kind of segfault in the net protocol stack is usually due to a
>> standard violation by the remote backend.

> Acually, I managed to get it work using saned -a, so I can't get any
> debug value which might help, since inetd mode does not support debug.
>
> Using saned from inetd still make the remote xsane/xscanimage to
> ask for authentication and then segfault.

This is a clear indication that the hpaio backend is writing to fd 0.

It's a bug in the code, either a fd declared static that's not
initialized to -1, or an fd explicitely initialized to 0, or, more
likely, the fd member of a struct that gets memset() after allocation
and is not properly initialized to -1 afterwards.

In any case, this breaks saned when run through inetd which is
guaranteed to render hpaio unusable over the network with the net
backend as whatever is written to fd 0 will confuse the hell out of
the net protocol stack.

JB.

[David Suffield (david-suffield)]

Show me the code :)

Hpaio has no io that writes to stdout. If you set SANE_DEBUG_DLL the sanei_init_debug() will write to stderr.

-dave

[Aaron Albright (albrigha-deactivatedaccount)]

Please update with the information Dave is requesting.

Thanks!

Aaron

[Aaron Albright (albrigha-deactivatedaccount)]

No update from user, closing.

Aaron

[Mark Purcell (msp)]

On Saturday 28 March 2009 08:18:21 Aaron Albright wrote:
> No update from user, closing.
>
> Aaron
>
> ** Changed in: hplip
> Assignee: (unassigned) => Aaron Albright (kalosaurusrex)
> Status: Incomplete => Invalid
> --
> hpaio backend writes to fd 0, breaks saned/net
> https://bugs.launchpad.net/bugs/283699

Aaron,

This is premature.

Whilst I can't show you the code, this still appears to be an issue for the
upstream user.

 http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=495005

Mark

[Mark Purcell (msp)]

From: Julien BLACHE <email address hidden>
To: Mark Purcell <email address hidden>
Cc: Bug 283699 <email address hidden>, <email address hidden>, <email address hidden>, <email address hidden>
Subject: Re: [Bug 283699] Re: hpaio backend writes to fd 0, breaks saned/net
Date: Fri, 03 Apr 2009 14:35:31 +0200

Mark Purcell <email address hidden> wrote:

Hi,

> Whilst I can't show you the code, this still appears to be an issue for the
> upstream user.

The problem has been tracked down to libdbus complaining
unconditionally on stderr when it's not able to establish a connection
to the DBus daemon.

Consequently, I've added a workaround in saned CVS and sane-backends
1.0.19-26.

I have to had that this has been uncovered by a user digging deep
enough to reveal the problem. Enough said.

JB.

--
 Julien BLACHE <email address hidden> | Debian, because code matters more
 Debian & GNU/Linux Developer | <http://www.debian.org>
 Public key available on <http://www.jblache.org> - KeyID: F5D6 5169
 GPG Fingerprint : 935A 79F1 C8B3 3521 FD62 7CC7 CD61 4FD7 F5D6 5169