[RFA] get go32 target to compile again.

[RFA] get go32 target to compile again.

[Pierre Muller]

[-- Attachment #1: Type: text/plain, Size: 661 bytes --]



   The go32 target does not compile anymore.
Eli told me that this is a main trunck only problem
and that he was mainly interested in the 5.1 branch
for the moment.

  Thus I tried to fix the compilation error myself.
As I succeeded, I send here enclosed the patch.

   This probably make the regno_mapping.size field 
not useful anymore, I suppose that we should trust regcache_collect
function that it will use the correct size for each register.


   2001-12-04  Pierre Muller  <muller@ics.u-strasbg.fr>

	* go32-nat.c (go32_fetch_registers): Remove call to register_buffer
	(which is now a static function in regcache.c)and use regcache_collect instead.


[-- Attachment #2: go32-nat.diffs --]
[-- Type: text/plain, Size: 714 bytes --]

Index: go32-nat.c
===================================================================
RCS file: /cvs/src/src/gdb/go32-nat.c,v
retrieving revision 1.25
diff -u -p -r1.25 go32-nat.c
--- go32-nat.c	2001/11/25 11:29:31	1.25
+++ go32-nat.c	2001/12/04 12:31:39
@@ -492,12 +492,8 @@ go32_fetch_registers (int regno)
 static void
 store_register (int regno)
 {
-  void *rp;
-  void *v = (void *) register_buffer (regno);
-
   if (regno < FP0_REGNUM)
-    memcpy ((char *) &a_tss + regno_mapping[regno].tss_ofs,
-	    v, regno_mapping[regno].size);
+    regcache_collect (regno, (void *) &a_tss + regno_mapping[regno].tss_ofs);
   else if (regno <= LAST_FPU_CTRL_REGNUM)
     i387_fill_fsave ((char *)&npx, regno);
   else

[-- Attachment #3: Type: text/plain, Size: 176 bytes --]



Pierre Muller
Institut Charles Sadron
6,rue Boussingault
F 67083 STRASBOURG CEDEX (France)
mailto:muller@ics.u-strasbg.fr
Phone : (33)-3-88-41-40-07  Fax : (33)-3-88-41-40-99

Re: [RFA] get go32 target to compile again.

[Andrew Cagney]

>  The go32 target does not compile anymore.
> Eli told me that this is a main trunck only problem
> and that he was mainly interested in the 5.1 branch
> for the moment.
> 
>   Thus I tried to fix the compilation error myself.
> As I succeeded, I send here enclosed the patch.
> 
>    This probably make the regno_mapping.size field 
> not useful anymore, I suppose that we should trust regcache_collect
> function that it will use the correct size for each register.

Oops.  Eli's approval, but thanks!  I noticed this last night but hey, 
you've beaten me to it.

regcache_collect() always uses REGISTER_RAW_SIZE() so if it is of 
concern an assertion would be a good idea.

	Andrew

Re: [RFA] get go32 target to compile again.

[Eli Zaretskii]

> Date: Tue, 04 Dec 2001 11:40:28 -0500
> From: Andrew Cagney <ac131313@cygnus.com>
> > 
> >   Thus I tried to fix the compilation error myself.
> > As I succeeded, I send here enclosed the patch.
> > 
> >    This probably make the regno_mapping.size field 
> > not useful anymore, I suppose that we should trust regcache_collect
> > function that it will use the correct size for each register.
> 
> Oops.  Eli's approval, but thanks!

I didn't approve the original patch, and I don't think I needed to.

As for this one, go ahead and commit it.  I don't have time to look
into this now, but if it works for Pierre and you, it's good enough
for me.

Thanks.

Re: [RFA] get go32 target to compile again.

[Pierre Muller]

At 18:43 04/12/2001 , vous avez écrit:
> > Date: Tue, 04 Dec 2001 11:40:28 -0500
> > From: Andrew Cagney <ac131313@cygnus.com>
> > > 
> > >   Thus I tried to fix the compilation error myself.
> > > As I succeeded, I send here enclosed the patch.
> > > 
> > >    This probably make the regno_mapping.size field 
> > > not useful anymore, I suppose that we should trust regcache_collect
> > > function that it will use the correct size for each register.
> > 
> > Oops.  Eli's approval, but thanks!
>
>I didn't approve the original patch, and I don't think I needed to.

But you are the go32 target maintainer, 
so you should always approve or refuse 
patches that concerns go32.

>As for this one, go ahead and commit it.  I don't have time to look
>into this now, but if it works for Pierre and you, it's good enough
>for me.

  Andrew, did you really test my patch?
Will you commit it?


Pierre Muller
Institut Charles Sadron
6,rue Boussingault
F 67083 STRASBOURG CEDEX (France)
mailto:muller@ics.u-strasbg.fr
Phone : (33)-3-88-41-40-07  Fax : (33)-3-88-41-40-99

Re: [RFA] get go32 target to compile again.

[Andrew Cagney]

>>I didn't approve the original patch, and I don't think I needed to.
> 
> 
> But you are the go32 target maintainer, 
> so you should always approve or refuse 
> patches that concerns go32.
> 
> 
>>As for this one, go ahead and commit it.  I don't have time to look
>>into this now, but if it works for Pierre and you, it's good enough
>>for me.
> 
> 
> Andrew, did you really test my patch?
> Will you commit it?

I can't test it.  I don't have a go32 environment set up.  It looks 
right - I was about to post a patch with a plee for someone to test it. 
  I would suggest assuming it is correct.

	Andrew

Re: [RFA] get go32 target to compile again.

[Eli Zaretskii]

> Date: Wed, 05 Dec 2001 15:49:52 +0100
> From: Pierre Muller <muller@cerbere.u-strasbg.fr>
> > > 
> > > Oops.  Eli's approval, but thanks!
> >
> >I didn't approve the original patch, and I don't think I needed to.
> 
> But you are the go32 target maintainer, 
> so you should always approve or refuse 
> patches that concerns go32.

With the exception of the obvious ones.  Andrew checked in this
change as an obvious.  And since I knew that Andrew builds GDB after
each change for several targets, including DJGPP, I didn't see any
reason to demand that I approve the change.

Your change is even more obvious than Andrew's, since it fixes a
problem that prevents compilation.

Re: [RFA] get go32 target to compile again.

[Pierre Muller]

At 18:46 05/12/2001 , Eli Zaretskii a écrit:
> > Date: Wed, 05 Dec 2001 15:49:52 +0100
> > From: Pierre Muller <muller@cerbere.u-strasbg.fr>
> > > > 
> > > > Oops.  Eli's approval, but thanks!
> > >
> > >I didn't approve the original patch, and I don't think I needed to.
> > 
> > But you are the go32 target maintainer, 
> > so you should always approve or refuse 
> > patches that concerns go32.
>
>With the exception of the obvious ones.  Andrew checked in this
>change as an obvious.  And since I knew that Andrew builds GDB after
>each change for several targets, including DJGPP, I didn't see any
>reason to demand that I approve the change.

OK, I committed this under the obvious fix rule 
without explicit approval. 

>Your change is even more obvious than Andrew's, since it fixes a
>problem that prevents compilation.



Pierre Muller
Institut Charles Sadron
6,rue Boussingault
F 67083 STRASBOURG CEDEX (France)
mailto:muller@ics.u-strasbg.fr
Phone : (33)-3-88-41-40-07  Fax : (33)-3-88-41-40-99