[RFA] solib-som.c fix for hpux

[RFA] solib-som.c fix for hpux

[Randolph Chung]

Dan J's change to solib handling
(http://sources.redhat.com/ml/gdb-patches/2006-01/msg00198.html) exposed
a probably unfounded assumption in the SOM solib code that assumes that
the solib_create_inferior_hook will always be called before current_sos.
As a result, gdb cvs on HPUX 32-bit would always error out when trying
to run any executable with the message "Debugging dynamic executables
loaded via the hpux8 dld.sl is not supported.". The most straightforward
fix is to remove the error and handle the case where there is not yet a
link map; that is what I have in the patch below. OTOH this does remove
the error message if anybody does try to run gdb on hpux8 (!!!)

Comments? ok to check in?

randolph

2006-03-24  Randolph Chung  <tausq@debian.org>

	* solib-som.c (link_map_start): Don't error out if there is
	not yet a link map.

Index: solib-som.c
===================================================================
RCS file: /cvs/src/src/gdb/solib-som.c,v
retrieving revision 1.8
diff -u -p -r1.8 solib-som.c
--- solib-som.c 24 Feb 2006 23:52:04 -0000      1.8
+++ solib-som.c 24 Mar 2006 05:18:35 -0000
@@ -520,7 +523,7 @@ link_map_start (void)
   read_memory (addr, buf, 4);
   addr = extract_unsigned_integer (buf, 4);
   if (addr == 0)
-    error (_("Debugging dynamic executables loaded via the hpux8 dld.sl
is not supported."));
+    return 0;

   read_memory (addr, buf, 4);
   return extract_unsigned_integer (buf, 4);

-- 
Randolph Chung
Debian GNU/Linux Developer, hppa/ia64 ports
http://www.tausq.org/

Re: [RFA] solib-som.c fix for hpux

[Daniel Jacobowitz]

On Fri, Mar 24, 2006 at 01:25:37PM +0800, Randolph Chung wrote:
> Dan J's change to solib handling
> (http://sources.redhat.com/ml/gdb-patches/2006-01/msg00198.html) exposed
> a probably unfounded assumption in the SOM solib code that assumes that
> the solib_create_inferior_hook will always be called before current_sos.
> As a result, gdb cvs on HPUX 32-bit would always error out when trying
> to run any executable with the message "Debugging dynamic executables
> loaded via the hpux8 dld.sl is not supported.". The most straightforward
> fix is to remove the error and handle the case where there is not yet a
> link map; that is what I have in the patch below. OTOH this does remove
> the error message if anybody does try to run gdb on hpux8 (!!!)

Grr!  I tried to catch all of these.  Some assume that add is called
first... others assume that create is called first...

> 2006-03-24  Randolph Chung  <tausq@debian.org>
> 
> 	* solib-som.c (link_map_start): Don't error out if there is
> 	not yet a link map.

It seems OK to me, but I don't want to approve SOM changes; maybe get
Kevin's opinion, since this is a shared library topic?

-- 
Daniel Jacobowitz
CodeSourcery

Re: [RFA] solib-som.c fix for hpux

[Randolph Chung]

> Grr!  I tried to catch all of these.  Some assume that add is called
> first... others assume that create is called first...

I found out that hpux 64-bit is also broken by this change, though it's 
a bit more complicated. I'll look at this more next week.

randolph

Re: [RFA] solib-som.c fix for hpux

[Kevin Buettner]

On Fri, 24 Mar 2006 15:23:15 -0500
Daniel Jacobowitz <drow@false.org> wrote:

> On Fri, Mar 24, 2006 at 01:25:37PM +0800, Randolph Chung wrote:
> > Dan J's change to solib handling
> > (http://sources.redhat.com/ml/gdb-patches/2006-01/msg00198.html) exposed
> > a probably unfounded assumption in the SOM solib code that assumes that
> > the solib_create_inferior_hook will always be called before current_sos.
> > As a result, gdb cvs on HPUX 32-bit would always error out when trying
> > to run any executable with the message "Debugging dynamic executables
> > loaded via the hpux8 dld.sl is not supported.". The most straightforward
> > fix is to remove the error and handle the case where there is not yet a
> > link map; that is what I have in the patch below. OTOH this does remove
> > the error message if anybody does try to run gdb on hpux8 (!!!)
> 
> Grr!  I tried to catch all of these.  Some assume that add is called
> first... others assume that create is called first...
> 
> > 2006-03-24  Randolph Chung  <tausq@debian.org>
> > 
> > 	* solib-som.c (link_map_start): Don't error out if there is
> > 	not yet a link map.
> 
> It seems OK to me, but I don't want to approve SOM changes; maybe get
> Kevin's opinion, since this is a shared library topic?

It looks okay to me too.

Kevin

Re: [RFA] solib-som.c fix for hpux

[Randolph Chung]

>>> 2006-03-24  Randolph Chung  <tausq@debian.org>
>>>
>>> 	* solib-som.c (link_map_start): Don't error out if there is
>>> 	not yet a link map.
>> It seems OK to me, but I don't want to approve SOM changes; maybe get
>> Kevin's opinion, since this is a shared library topic?
> 
> It looks okay to me too.

Thanks. This has been committed.

randolph

Re: [RFA] solib-som.c fix for hpux

[Mark Kettenis]

> Date: Fri, 24 Mar 2006 13:25:37 +0800
> From: Randolph Chung <randolph@tausq.org>
> 
> Dan J's change to solib handling
> (http://sources.redhat.com/ml/gdb-patches/2006-01/msg00198.html) exposed
> a probably unfounded assumption in the SOM solib code that assumes that
> the solib_create_inferior_hook will always be called before current_sos.
> As a result, gdb cvs on HPUX 32-bit would always error out when trying
> to run any executable with the message "Debugging dynamic executables
> loaded via the hpux8 dld.sl is not supported.". The most straightforward
> fix is to remove the error and handle the case where there is not yet a
> link map; that is what I have in the patch below. OTOH this does remove
> the error message if anybody does try to run gdb on hpux8 (!!!)
> 
> Comments? ok to check in?

In any case, that error message is wrong; we really don't support
HP-UX 8 anymore.  So go ahead and slap this in.

> 2006-03-24  Randolph Chung  <tausq@debian.org>
> 
> 	* solib-som.c (link_map_start): Don't error out if there is
> 	not yet a link map.