Re: [RFA] (ping) GDB crash when using command lines due to memory corruption

[Andrew Cagney <ac131313@redhat.com>]

[-- Attachment #1: Type: text/plain, Size: 30 bytes --]

Again.

Jim, Michael?

Andrew

[-- Attachment #2: mailbox-message://ac131313@movemail/fsf/gdb/patches#30103427 --]
[-- Type: message/rfc822, Size: 4283 bytes --]

From: Andrew Cagney <ac131313@redhat.com>
To: Michael Snyder <msnyder@redhat.com>, Jim Blandy <jimb@redhat.com>
Cc: Joel Brobecker <brobecker@gnat.com>, gdb-patches@sources.redhat.com
Subject: Re: [RFA] (ping) GDB crash when using command lines due to memory corruption
Date: Tue, 11 Feb 2003 17:55:49 +0100
Message-ID: <3E492B15.50601@redhat.com>

Jim? Michael?

Per the bug database, three variants on this patch have been submitted 
yet nothing got resolved.

Andrew

>> There is a patch that was submitted on Oct 30th 2002 but hasn't been
>> reviewed despite the fact that it addresses a GDB crash. It also
>> expands break.exp to test for this.
> 
> 
> Second ping...
> 
> 
>> On Wed, Dec 11, 2002 at 01:00:58PM -0500, Klee Dienes wrote:
> 
>> > A safer change for 5.3 might be the patch I submitted on October 30th.
>> > 
>> > 	http://sources.redhat.com/ml/gdb-patches/2002-10/msg00586.html
>> > 
>> > Rather than deal with sharing the command-line structure, I just 
>> > updated bpstat_copy to match its comment, and do a deep copy of the 
>> > command lines as well as the value.  I don't really have a strong 
>> > opinion about copying the command lines vs. managing them the way Joel 
>> > proposes, although my patch does have the argument of simplicity going 
>> > for it.  On the other hand, if/when we go to a more sophistiated 
>> > command-line evaluator, we'll probably want the command body to be some 
>> > opaque and externally managed structure anyway.
>> > 
>> > Whichever patch we end up taking, though, we should be sure to update 
>> > the comment in bpstat_copy and add my proposed change to the test suite.
> 
> 
> Thanks,
> -- Joel