[PATCH] Readline bug causes GDB crash on 64-bit

[PATCH] Readline bug causes GDB crash on 64-bit

[Ulrich Weigand]

Hello,

GDB on s390x (64-bit) is crashing when you press ctrl-V.
This is because xmalloc is used without prototype in scope
in _rl_callback_data_alloc (readline/callback.c), and thus
the returned pointer is improperly truncated/extended.

Fixed by including "xmalloc.h".  Tested on s390x-ibm-linux.

What's the policy for readline bugs in GDB?  It is OK to check
this patch into GDB, or does it have to go upstream first?
(Where is readline upstream?)

Bye,
Ulrich


ChangeLog:

	* callback.c: Include "xmalloc.h".
	* Makefile.in: Add dependency.


Index: readline/Makefile.in
===================================================================
RCS file: /cvs/src/src/readline/Makefile.in,v
retrieving revision 1.7
diff -c -p -r1.7 Makefile.in
*** readline/Makefile.in	5 May 2006 18:26:12 -0000	1.7
--- readline/Makefile.in	20 Oct 2006 20:48:49 -0000
*************** util.o: rlprivate.h
*** 446,451 ****
--- 446,452 ----
  vi_mode.o: rlprivate.h
  
  bind.o: xmalloc.h
+ callback.o: xmalloc.h
  complete.o: xmalloc.h
  display.o: xmalloc.h
  funmap.o: xmalloc.h
Index: readline/callback.c
===================================================================
RCS file: /cvs/src/src/readline/callback.c,v
retrieving revision 1.6
diff -c -p -r1.6 callback.c
*** readline/callback.c	5 May 2006 18:26:12 -0000	1.6
--- readline/callback.c	20 Oct 2006 20:48:49 -0000
***************
*** 43,48 ****
--- 43,49 ----
  #include "rldefs.h"
  #include "readline.h"
  #include "rlprivate.h"
+ #include "xmalloc.h"
  
  /* Private data for callback registration functions.  See comments in
     rl_callback_read_char for more details. */
-- 
  Dr. Ulrich Weigand
  Linux on zSeries Development
  Ulrich.Weigand@de.ibm.com

Re: [PATCH] Readline bug causes GDB crash on 64-bit

[Andreas Schwab]

"Ulrich Weigand" <uweigand@de.ibm.com> writes:

> What's the policy for readline bugs in GDB?  It is OK to check
> this patch into GDB, or does it have to go upstream first?

Upstream does not have this bug any more.

> (Where is readline upstream?)

bash

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux Products GmbH, Maxfeldstraße 5, 90409 Nürnberg, Germany
PGP key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: [PATCH] Readline bug causes GDB crash on 64-bit

[Daniel Jacobowitz]

On Fri, Oct 20, 2006 at 11:10:43PM +0200, Ulrich Weigand wrote:
> Hello,
> 
> GDB on s390x (64-bit) is crashing when you press ctrl-V.
> This is because xmalloc is used without prototype in scope
> in _rl_callback_data_alloc (readline/callback.c), and thus
> the returned pointer is improperly truncated/extended.
> 
> Fixed by including "xmalloc.h".  Tested on s390x-ibm-linux.
> 
> What's the policy for readline bugs in GDB?  It is OK to check
> this patch into GDB, or does it have to go upstream first?
> (Where is readline upstream?)

The right thing to do is to send the patch to Chet, and if he accepts
it for the next readline release, commit it to our copy.  I believe
bash-maintainers@gnu.org (CC'd) is the right place.

> ChangeLog:
> 
> 	* callback.c: Include "xmalloc.h".
> 	* Makefile.in: Add dependency.
> 
> 
> Index: readline/Makefile.in
> ===================================================================
> RCS file: /cvs/src/src/readline/Makefile.in,v
> retrieving revision 1.7
> diff -c -p -r1.7 Makefile.in
> *** readline/Makefile.in	5 May 2006 18:26:12 -0000	1.7
> --- readline/Makefile.in	20 Oct 2006 20:48:49 -0000
> *************** util.o: rlprivate.h
> *** 446,451 ****
> --- 446,452 ----
>   vi_mode.o: rlprivate.h
>   
>   bind.o: xmalloc.h
> + callback.o: xmalloc.h
>   complete.o: xmalloc.h
>   display.o: xmalloc.h
>   funmap.o: xmalloc.h
> Index: readline/callback.c
> ===================================================================
> RCS file: /cvs/src/src/readline/callback.c,v
> retrieving revision 1.6
> diff -c -p -r1.6 callback.c
> *** readline/callback.c	5 May 2006 18:26:12 -0000	1.6
> --- readline/callback.c	20 Oct 2006 20:48:49 -0000
> ***************
> *** 43,48 ****
> --- 43,49 ----
>   #include "rldefs.h"
>   #include "readline.h"
>   #include "rlprivate.h"
> + #include "xmalloc.h"
>   
>   /* Private data for callback registration functions.  See comments in
>      rl_callback_read_char for more details. */
> -- 
>   Dr. Ulrich Weigand
>   Linux on zSeries Development
>   Ulrich.Weigand@de.ibm.com
> 

-- 
Daniel Jacobowitz
CodeSourcery

Re: [PATCH] Readline bug causes GDB crash on 64-bit

[Daniel Jacobowitz]

Oops, didn't see Andreas's reply before I sent mine...

On Fri, Oct 20, 2006 at 11:30:15PM +0200, Andreas Schwab wrote:
> "Ulrich Weigand" <uweigand@de.ibm.com> writes:
> 
> > What's the policy for readline bugs in GDB?  It is OK to check
> > this patch into GDB, or does it have to go upstream first?
> 
> Upstream does not have this bug any more.

I guess we ought to import readline 5.2 sometime.

-- 
Daniel Jacobowitz
CodeSourcery

Re: [PATCH] Readline bug causes GDB crash on 64-bit

[Chet Ramey]

Daniel Jacobowitz wrote:
> On Fri, Oct 20, 2006 at 11:10:43PM +0200, Ulrich Weigand wrote:
>> Hello,
>>
>> GDB on s390x (64-bit) is crashing when you press ctrl-V.
>> This is because xmalloc is used without prototype in scope
>> in _rl_callback_data_alloc (readline/callback.c), and thus
>> the returned pointer is improperly truncated/extended.
>>
>> Fixed by including "xmalloc.h".  Tested on s390x-ibm-linux.
>>
>> What's the policy for readline bugs in GDB?  It is OK to check
>> this patch into GDB, or does it have to go upstream first?
>> (Where is readline upstream?)
> 
> The right thing to do is to send the patch to Chet, and if he accepts
> it for the next readline release, commit it to our copy.  I believe
> bash-maintainers@gnu.org (CC'd) is the right place.

Either that address or bug-readline@gnu.org will work.  But in this
case, someone beat you to it, and the fix is in the recently-released
readline-5.2.

Chet

-- 
``The lyf so short, the craft so long to lerne.'' - Chaucer
		       Live Strong.  No day but today.
Chet Ramey, ITS, CWRU    chet@case.edu    http://cnswww.cns.cwru.edu/~chet/

Re: [PATCH] Readline bug causes GDB crash on 64-bit

[Daniel Jacobowitz]

On Fri, Oct 20, 2006 at 11:10:43PM +0200, Ulrich Weigand wrote:
> Hello,
> 
> GDB on s390x (64-bit) is crashing when you press ctrl-V.
> This is because xmalloc is used without prototype in scope
> in _rl_callback_data_alloc (readline/callback.c), and thus
> the returned pointer is improperly truncated/extended.
> 
> Fixed by including "xmalloc.h".  Tested on s390x-ibm-linux.
> 
> What's the policy for readline bugs in GDB?  It is OK to check
> this patch into GDB, or does it have to go upstream first?
> (Where is readline upstream?)

I don't want to mess with a readline import just now.  So, unless you
want to try it, this patch is fine for now.

-- 
Daniel Jacobowitz
CodeSourcery

Re: [PATCH] Readline bug causes GDB crash on 64-bit

[Ulrich Weigand]

Daniel Jacobowitz wrote:

> I don't want to mess with a readline import just now.  So, unless you
> want to try it, this patch is fine for now.

I've committed the patch now, thanks.

Bye,
Ulrich

-- 
  Dr. Ulrich Weigand
  Linux on zSeries Development
  Ulrich.Weigand@de.ibm.com