[RFA] Add comment describing the target_gdbarch global

[RFA] Add comment describing the target_gdbarch global

[Joel Brobecker]

[-- Attachment #1: Type: text/plain, Size: 530 bytes --]

Hello,

I remember a few months ago when I stumbled across a piece of code
using target_gdbarch, and I couldn't remember what this global was,
and what the different was with gdbarch.  I eventually found the
original discussion again

(http://sourceware.org/ml/gdb-patches/2008-08/msg00358.html)

but I think a comment would be nice.  I'm finally writing one today.
Does this look accurate?

2009-05-06  Joel Brobecker  <brobecker@adacore.com>

        * gdbarch.h (target_gdbarch): Add comment documenting this global.

-- 
Joel

[-- Attachment #2: comment.diff --]
[-- Type: text/x-diff, Size: 851 bytes --]

diff --git a/gdb/gdbarch.h b/gdb/gdbarch.h
index 5e508d7..31c935f 100644
--- a/gdb/gdbarch.h
+++ b/gdb/gdbarch.h
@@ -54,6 +54,17 @@ struct displaced_step_closure;
 struct core_regset_section;
 
 extern struct gdbarch *current_gdbarch;
+
+/* The architecture associated with the connection to the target.
+ 
+   The architecture vector provides some information that is really
+   a property of the target: The layout of certain packets, for instance;
+   or the solib_ops vector.  Etc.  To differentiate architecture accesses
+   to per-target properties to per-thread/pre-frame/per-objfile properties,
+   accesses to per-target properties should be made through target_gdbarch.
+
+   Eventually, when support for multiple targets is implemented in
+   GDB, this global should be made target-specific.  */
 extern struct gdbarch *target_gdbarch;
 
 

Re: [RFA] Add comment describing the target_gdbarch global

[Ulrich Weigand]

Joel Brobecker wrote:

> I remember a few months ago when I stumbled across a piece of code
> using target_gdbarch, and I couldn't remember what this global was,
> and what the different was with gdbarch.  I eventually found the
> original discussion again
> 
> (http://sourceware.org/ml/gdb-patches/2008-08/msg00358.html)
> 
> but I think a comment would be nice.  I'm finally writing one today.
> Does this look accurate?
> 
> 2009-05-06  Joel Brobecker  <brobecker@adacore.com>
> 
>         * gdbarch.h (target_gdbarch): Add comment documenting this global.

This looks fine to me.

Thanks,
Ulrich

-- 
  Dr. Ulrich Weigand
  GNU Toolchain for Linux on System z and Cell BE
  Ulrich.Weigand@de.ibm.com

Re: [RFA] Add comment describing the target_gdbarch global

[Joel Brobecker]

[-- Attachment #1: Type: text/plain, Size: 362 bytes --]

> > 2009-05-06  Joel Brobecker  <brobecker@adacore.com>
> > 
> >         * gdbarch.h (target_gdbarch): Add comment documenting this global.
> 
> This looks fine to me.

Excellent! Thanks for double-checking. I committed the patch after
having fixed a couple of typos. Patch attached for reference, but
basically a "pre" -> "per", and a "to" -> "from".

-- 
Joel

[-- Attachment #2: comment.diff --]
[-- Type: text/x-diff, Size: 853 bytes --]

diff --git a/gdb/gdbarch.h b/gdb/gdbarch.h
index 5e508d7..31c935f 100644
--- a/gdb/gdbarch.h
+++ b/gdb/gdbarch.h
@@ -54,6 +54,17 @@ struct displaced_step_closure;
 struct core_regset_section;
 
 extern struct gdbarch *current_gdbarch;
+
+/* The architecture associated with the connection to the target.
+ 
+   The architecture vector provides some information that is really
+   a property of the target: The layout of certain packets, for instance;
+   or the solib_ops vector.  Etc.  To differentiate architecture accesses
+   to per-target properties from per-thread/per-frame/per-objfile properties,
+   accesses to per-target properties should be made through target_gdbarch.
+
+   Eventually, when support for multiple targets is implemented in
+   GDB, this global should be made target-specific.  */
 extern struct gdbarch *target_gdbarch;
 
 

Re: [RFA] Add comment describing the target_gdbarch global

[Ulrich Weigand]

Joel Brobecker wrote:

> > > 2009-05-06  Joel Brobecker  <brobecker@adacore.com>
> > > 
> > >         * gdbarch.h (target_gdbarch): Add comment documenting this global.
> > 
> > This looks fine to me.
> 
> Excellent! Thanks for double-checking. I committed the patch after
> having fixed a couple of typos. Patch attached for reference, but
> basically a "pre" -> "per", and a "to" -> "from".

Oops, I didn't notice that you're changing gdbarch.h directly.
This is a generated file; you should be changing gdbarch.sh
instead and regnerate the header file ...

Bye,
Ulrich

-- 
  Dr. Ulrich Weigand
  GNU Toolchain for Linux on System z and Cell BE
  Ulrich.Weigand@de.ibm.com

Re: [RFA] Add comment describing the target_gdbarch global

[Joel Brobecker]

[-- Attachment #1: Type: text/plain, Size: 943 bytes --]

> > 2009-05-06  Joel Brobecker  <brobecker@adacore.com>
> > 
> >         * gdbarch.h (target_gdbarch): Add comment documenting this global.
> 
> Oops, I didn't notice that you're changing gdbarch.h directly.
> This is a generated file; you should be changing gdbarch.sh
> instead and regnerate the header file ...

(head shaking) This is really a rookie mistake. I can't believe I wasn't
paying more attention. Sorry about that!!!

Fixed thusly:

2009-05-14  Joel Brobecker  <brobecker@adacore.com>

        * gdbarch.sh: Add comment documenting the target_gdbarch global.
        gdbarch.h does NOT need to be regenerated as the corresponding
        change has already been made in a previous commit.

As explained in the ChangeLog, I reran gdbarch.sh, and it generated
the exact same files.

Checked in.

-- 
Joel

PS: I'll review your Ada patches tomorrow morning (it's getting
    too late for me, now). But thanks much for sending them.

[-- Attachment #2: gdbarch.diff --]
[-- Type: text/x-diff, Size: 1017 bytes --]

Index: gdbarch.sh
===================================================================
RCS file: /cvs/src/src/gdb/gdbarch.sh,v
retrieving revision 1.487
diff -u -p -r1.487 gdbarch.sh
--- gdbarch.sh	30 Apr 2009 02:52:16 -0000	1.487
+++ gdbarch.sh	14 May 2009 19:13:43 -0000
@@ -850,6 +850,17 @@ struct displaced_step_closure;
 struct core_regset_section;
 
 extern struct gdbarch *current_gdbarch;
+
+/* The architecture associated with the connection to the target.
+ 
+   The architecture vector provides some information that is really
+   a property of the target: The layout of certain packets, for instance;
+   or the solib_ops vector.  Etc.  To differentiate architecture accesses
+   to per-target properties from per-thread/per-frame/per-objfile properties,
+   accesses to per-target properties should be made through target_gdbarch.
+
+   Eventually, when support for multiple targets is implemented in
+   GDB, this global should be made target-specific.  */
 extern struct gdbarch *target_gdbarch;
 EOF