Re: [PATCH 3/N] some minor fixes in sim, gold, gdb

Re: [PATCH 3/N] some minor fixes in sim, gold, gdb

[Ralf Wildenhues]

* Ralf Wildenhues wrote on Sat, Aug 15, 2009 at 01:33:03PM CEST:
> > - some minor fixes in sim, gold, gdb (src only)

Of these changes, this one still needs approval:

> - sim/common/aclocal.m4: SIM_CHECK_MEMBER is alias for AC_CHECK_MEMBER now.
>   This also needs to happen at the same time as the Autoconf update, to
>   avoid exposing whatever bugs 2.59's AC_CHECK_MEMBER had.
>   Is this ok or would you rather I replace every use of SIM_CHECK_MEMBER
>   with AC_CHECK_MEMBER?

Thanks,
Ralf

> sim/common/ChangeLog:
> 2009-08-15  Ralf Wildenhues  <Ralf.Wildenhues@gmx.de>
> 
> 	* aclocal.m4 (SIM_CHECK_MEMBER): Replace definition with
> 	definition of AC_CHECK_MEMBER.

> diff --git a/sim/common/aclocal.m4 b/sim/common/aclocal.m4
> index e8fd147..25ed460 100644
> --- a/sim/common/aclocal.m4
> +++ b/sim/common/aclocal.m4
> @@ -942,28 +942,7 @@ dnl
>  dnl  ---------------------------------------------------------
>  dnl  AGGREGATE.MEMBER is for instance `struct passwd.pw_gecos', shell
>  dnl  variables are not a valid argument.
> -AC_DEFUN([SIM_CHECK_MEMBER],
> -dnl Extract the aggregate name, and the member name
> -[AC_CACHE_CHECK([for $1], [ac_]patsubst([$1], [[\. ]], [_]),
> -[ac_]patsubst([$1], [[\. ]], [_])[=no;]
> -AC_TRY_COMPILE([$4],[
> -dnl AGGREGATE ac_aggr;
> -static ]patsubst([$1], [\..*])[ ac_aggr;
> -dnl ac_aggr.MEMBER;
> -if (ac_aggr.]patsubst([$1], [^[^.]*\.])[)
> -return 0;],[ac_]patsubst([$1], [[\. ]], [_])[=yes;],
> -AC_TRY_COMPILE([$4],[
> -dnl AGGREGATE ac_aggr;
> -static ]patsubst([$1], [\..*])[ ac_aggr;
> -dnl ac_aggr.MEMBER;
> -if (sizeof ac_aggr.]patsubst([$1], [^[^.]*\.])[)
> -return 0;],
> -[ac_]patsubst([$1], [[\. ]], [_])[=yes;],
> -[ac_]patsubst([$1], [[\. ]], [_])[=no;]))
> -[if test [$]ac_]patsubst([$1], [[\. ]], [_])[ = yes; then :; [$2]
> -else :; [$3]
> -fi])
> -])dnl SIM_CHECK_MEMBER
> +AC_DEFUN([SIM_CHECK_MEMBER], defn([AC_CHECK_MEMBER]))
>  dnl
>  dnl Translated from a FC2 autoconf-2.59-3 installation.
>  dnl  SIM_CHECK_MEMBERS([AGGREGATE.MEMBER, ...])

Re: [PATCH 3/N] some minor fixes in sim, gold, gdb

[Tom Tromey]

>>>>> "Ralf" == Ralf Wildenhues <Ralf.Wildenhues@gmx.de> writes:

Ralf> Of these changes, this one still needs approval:

>> - sim/common/aclocal.m4: SIM_CHECK_MEMBER is alias for AC_CHECK_MEMBER now.
>> This also needs to happen at the same time as the Autoconf update, to
>> avoid exposing whatever bugs 2.59's AC_CHECK_MEMBER had.
>> Is this ok or would you rather I replace every use of SIM_CHECK_MEMBER
>> with AC_CHECK_MEMBER?

This is ok.

I think it would be preferable to just do the replacement.
However, I realize your patch is pretty huge already.
So, if you would prefer to just change the definition of
SIM_CHECK_MEMBER, that is fine by me.

Tom

Re: [PATCH 3/N] some minor fixes in sim, gold, gdb

[Ralf Wildenhues]

* Tom Tromey wrote on Wed, Aug 19, 2009 at 08:35:53PM CEST:
> >>>>> "Ralf" == Ralf Wildenhues writes:
> 
> Ralf> Of these changes, this one still needs approval:
> 
> >> - sim/common/aclocal.m4: SIM_CHECK_MEMBER is alias for AC_CHECK_MEMBER now.
> >> This also needs to happen at the same time as the Autoconf update, to
> >> avoid exposing whatever bugs 2.59's AC_CHECK_MEMBER had.
> >> Is this ok or would you rather I replace every use of SIM_CHECK_MEMBER
> >> with AC_CHECK_MEMBER?
> 
> This is ok.

Thanks, got all reviews now.

> I think it would be preferable to just do the replacement.
> However, I realize your patch is pretty huge already.
> So, if you would prefer to just change the definition of
> SIM_CHECK_MEMBER, that is fine by me.

Yes, I think I'd prefer to do this kind of cleanup later.

Cheers,
Ralf

Re: [PATCH 3/N] some minor fixes in sim, gold, gdb

[Frank Ch. Eigler]

Ralf Wildenhues <Ralf.Wildenhues@gmx.de> writes:

> Of these changes, this one still needs approval:
>
>> - sim/common/aclocal.m4: SIM_CHECK_MEMBER is alias for AC_CHECK_MEMBER now.
>>   This also needs to happen at the same time as the Autoconf update, to
>>   avoid exposing whatever bugs 2.59's AC_CHECK_MEMBER had.
>>   Is this ok or would you rather I replace every use of SIM_CHECK_MEMBER
>>   with AC_CHECK_MEMBER?

According to the FIXME note in that file, please excise SIM_CHECK_*
altogether if it has been mooted by a modern autoconf.

- FChE

Re: [PATCH 3/N] some minor fixes in sim, gold, gdb

[Ralf Wildenhues]

* Frank Ch. Eigler wrote on Wed, Aug 19, 2009 at 09:34:37PM CEST:
> Ralf Wildenhues <Ralf.Wildenhues@gmx.de> writes:
> > Of these changes, this one still needs approval:
> >
> >> - sim/common/aclocal.m4: SIM_CHECK_MEMBER is alias for AC_CHECK_MEMBER now.
> >>   This also needs to happen at the same time as the Autoconf update, to
> >>   avoid exposing whatever bugs 2.59's AC_CHECK_MEMBER had.
> >>   Is this ok or would you rather I replace every use of SIM_CHECK_MEMBER
> >>   with AC_CHECK_MEMBER?
> 
> According to the FIXME note in that file, please excise SIM_CHECK_*
> altogether if it has been mooted by a modern autoconf.

I will work on a patch to clean this up after the move.

Thanks,
Ralf

Re: Do not point to sources.redhat.com for autotools tarballs

[Ralf Wildenhues]

[ trimming Cc:, moving from gdb@ to -patches ]

Hello Eli,

* Eli Zaretskii wrote on Sun, Aug 16, 2009 at 07:04:32PM CEST:
> > 	* gdbint.texinfo (Releasing GDB): Point to
> > 	README-maintainer-mode file for required autoconf version.
> 
> Fine with me (although I'd prefer you fixed the confusing ``system
> installed version'' thingy, if you can).

Not sure I understand what you mean.  Like this?

Thanks,
Ralf

Fix confusing sentence about autoconf in gdbint.texinfo.

gdb/doc/ChangeLog:
2009-08-25  Ralf Wildenhues  <Ralf.Wildenhues@gmx.de>

	* gdbint.texinfo (Releasing GDB): Fix confusing sentence
	about autoconf.

diff --git a/gdb/doc/gdbint.texinfo b/gdb/doc/gdbint.texinfo
index 6d0a226..e706caa 100644
--- a/gdb/doc/gdbint.texinfo
+++ b/gdb/doc/gdbint.texinfo
@@ -7079,8 +7079,8 @@ Notes:
 @item
 Check the @code{autoconf} version carefully.  You want to be using the
 version documented in the toplevel @file{README-maintainer-mode} file.
-It is very unlikely that a system installed version of @code{autoconf}
-(e.g., @file{/usr/bin/autoconf}) is correct.
+It is very unlikely that the version of @code{autoconf} installed in
+system directories (e.g., @file{/usr/bin/autoconf}) is correct.
 @end itemize
 
 @subsubheading Check out the relevant modules:

fixup SIM_CHECK_* (was: [PATCH 3/N] some minor fixes in sim, gold, gdb)

[Ralf Wildenhues]

* Frank Ch. Eigler wrote on Wed, Aug 19, 2009 at 09:34:37PM CEST:
> According to the FIXME note in that file, please excise SIM_CHECK_*
> altogether if it has been mooted by a modern autoconf.

This was still left TODO.  Of course, I managed to actually commit a
patch, that would not expand AC_CHECK_MEMBERS correctly, as that uses
_$0_BODY internally, and $0 is still SIM_CHECK_MEMBER (Paolo, it's ugly
to have to remember this, so we maybe shouldn't use m4_defn so much):

 ac_fn_c_check_member ()
 {
   _SIM_CHECK_MEMBER_BODY
 }

So I propose to fix this in two steps, first fixing up my blunder, for
documentation purposes, then removing SIM_CHECK_MEMBER*.

I've (hopefully) managed to test this a bit better now, at least the
generated sim/common/cconfig.h file looks sane to me now.

OK to commit both?

Thanks, and sorry for the blunder,
Ralf

Fix SIM_CHECK_MEMBER definition with Autoconf 2.64.

sim/common/ChangeLog:
2009-08-25  Ralf Wildenhues  <Ralf.Wildenhues@gmx.de>

	* aclocal.m4 (_SIM_CHECK_MEMBER_BODY): New define.
	* configure: Regenerate.

diff --git a/sim/common/aclocal.m4 b/sim/common/aclocal.m4
index 24b1aa2..80d3b4b 100644
--- a/sim/common/aclocal.m4
+++ b/sim/common/aclocal.m4
@@ -946,6 +946,7 @@ dnl  ---------------------------------------------------------
 dnl  AGGREGATE.MEMBER is for instance `struct passwd.pw_gecos', shell
 dnl  variables are not a valid argument.
 AC_DEFUN([SIM_CHECK_MEMBER], defn([AC_CHECK_MEMBER]))
+AC_DEFUN([_SIM_CHECK_MEMBER_BODY], defn([_AC_CHECK_MEMBER_BODY]))
 dnl
 dnl Translated from a FC2 autoconf-2.59-3 installation.
 dnl  SIM_CHECK_MEMBERS([AGGREGATE.MEMBER, ...])


Remove SIM_CHECK_MEMBER*.

sim/common/ChangeLog:
2009-08-25  Ralf Wildenhues  <Ralf.Wildenhues@gmx.de>

	* aclocal.m4 (SIM_CHECK_MEMBER, SIM_CHECK_MEMBERS)
        (SIM_CHECK_MEMBERS_1, _SIM_CHECK_MEMBER_BODY): Remove.
	* configure.ac: Replace SIM_CHECK_MEMBERS call with equivalent
	AC_CHECK_MEMBERS one.
	* configure: Regenerate.

diff --git a/sim/common/aclocal.m4 b/sim/common/aclocal.m4
index 80d3b4b..7beaebd 100644
--- a/sim/common/aclocal.m4
+++ b/sim/common/aclocal.m4
@@ -933,51 +933,3 @@ AC_SUBST(CGEN_MAINT)
 AC_SUBST(cgendir)
 AC_SUBST(cgen)
 ])
-dnl FIXME: When upgrading to modern autoconf, remove
-dnl        SIM_CHECK_MEMBER and SIM_CHECK_MEMBERS et al and use
-dnl        AC_CHECK_MEMBERS from autoconf.
-dnl
-dnl Translated from a FC2 autoconf-2.59-3 installation.
-dnl  AC_CHECK_MEMBER(AGGREGATE.MEMBER,
-dnl                  [ACTION-IF-FOUND], [ACTION-IF-NOT-FOUND],
-dnl                  [INCLUDES])
-dnl
-dnl  ---------------------------------------------------------
-dnl  AGGREGATE.MEMBER is for instance `struct passwd.pw_gecos', shell
-dnl  variables are not a valid argument.
-AC_DEFUN([SIM_CHECK_MEMBER], defn([AC_CHECK_MEMBER]))
-AC_DEFUN([_SIM_CHECK_MEMBER_BODY], defn([_AC_CHECK_MEMBER_BODY]))
-dnl
-dnl Translated from a FC2 autoconf-2.59-3 installation.
-dnl  SIM_CHECK_MEMBERS([AGGREGATE.MEMBER, ...])
-dnl except we just work with a limited set of fixed includes.
-dnl
-AC_DEFUN([SIM_CHECK_MEMBERS_1],
-[ifelse($#, 1,
-[SIM_CHECK_MEMBER([$1],
-AC_DEFINE_UNQUOTED([HAVE_]translit([$1], [a-z .], [A-Z__]), 1,
-[Define to 1 if ]patsubst([$1],
-[^[^.]*\.])[ is a member of ]patsubst([$1], [\..*])[. ]),,
-[#ifdef HAVE_SYS_TYPES_H
-#include <sys/types.h>
-#endif
-#ifdef HAVE_SYS_STAT_H
-#include <sys/stat.h>
-#endif])],
-[SIM_CHECK_MEMBER([$1],
-AC_DEFINE_UNQUOTED([HAVE_]translit([$1], [a-z .], [A-Z__]), 1,
-[Define to 1 if ]patsubst([$1],
-[^[^.]*\.])[ is a member of ]patsubst([$1], [\..*])[. ]),,
-[#ifdef HAVE_SYS_TYPES_H
-#include <sys/types.h>
-#endif
-#ifdef HAVE_SYS_STAT_H
-#include <sys/stat.h>
-#endif])
-SIM_CHECK_MEMBERS_1(builtin(shift,$@))])])dnl SIM_CHECK_MEMBERS
-dnl
-AC_DEFUN([SIM_CHECK_MEMBERS],
-[ifelse($#, 1, [SIM_CHECK_MEMBERS_1($1)],
-[errprint(__file__:__line__:
-[This SIM_CHECK_MEMBERS only supports one argument,]
-[the list of struct tests])])])dnl SIM_CHECK_MEMBERS
diff --git a/sim/common/configure.ac b/sim/common/configure.ac
index 5516d9e..5f5845a 100644
--- a/sim/common/configure.ac
+++ b/sim/common/configure.ac
@@ -38,10 +38,16 @@ AC_SUBST(TARGET_SUBDIR)
 # These aren't all needed yet, but will be eventually.
 AC_CHECK_HEADERS(stdlib.h string.h strings.h time.h sys/times.h sys/stat.h sys/mman.h)
 AC_CHECK_FUNCS(mmap munmap lstat truncate ftruncate)
-SIM_CHECK_MEMBERS([[struct stat.st_dev], [struct stat.st_ino],
+AC_CHECK_MEMBERS([[struct stat.st_dev], [struct stat.st_ino],
 [struct stat.st_mode], [struct stat.st_nlink], [struct stat.st_uid],
 [struct stat.st_gid], [struct stat.st_rdev], [struct stat.st_size],
 [struct stat.st_blksize], [struct stat.st_blocks], [struct stat.st_atime],
-[struct stat.st_mtime], [struct stat.st_ctime]])
+[struct stat.st_mtime], [struct stat.st_ctime]], [], [],
+[[#ifdef HAVE_SYS_TYPES_H
+#include <sys/types.h>
+#endif
+#ifdef HAVE_SYS_STAT_H
+#include <sys/stat.h>
+#endif]])
 AC_OUTPUT(Makefile,
 [case x$CONFIG_HEADERS in xcconfig.h:config.in) echo > stamp-h ;; esac])

Re: Do not point to sources.redhat.com for autotools tarballs

[Eli Zaretskii]

> Date: Tue, 25 Aug 2009 21:13:57 +0200
> From: Ralf Wildenhues <Ralf.Wildenhues@gmx.de>
> Cc: gerald@pfeifer.com, gdb-patches@sourceware.org
> 
> * Eli Zaretskii wrote on Sun, Aug 16, 2009 at 07:04:32PM CEST:
> > > 	* gdbint.texinfo (Releasing GDB): Point to
> > > 	README-maintainer-mode file for required autoconf version.
> > 
> > Fine with me (although I'd prefer you fixed the confusing ``system
> > installed version'' thingy, if you can).
> 
> Not sure I understand what you mean.  Like this?

Yes, thanks.

Re: fixup SIM_CHECK_*

[Paolo Bonzini]

On 08/25/2009 09:55 PM, Ralf Wildenhues wrote:
> * Frank Ch. Eigler wrote on Wed, Aug 19, 2009 at 09:34:37PM CEST:
>> According to the FIXME note in that file, please excise SIM_CHECK_*
>> altogether if it has been mooted by a modern autoconf.
>
> This was still left TODO.  Of course, I managed to actually commit a
> patch, that would not expand AC_CHECK_MEMBERS correctly, as that uses
> _$0_BODY internally, and $0 is still SIM_CHECK_MEMBER (Paolo, it's ugly
> to have to remember this, so we maybe shouldn't use m4_defn so much):

Hey, Eric (Blake) is the big optimizer in Autoconf-land, not me. :-)  I 
agree with you, though of course I don't like this kind of relying on 
Autoconf internals.

The patch seems fine.

Paolo

Re: fixup SIM_CHECK_*

[Ralf Wildenhues]

Hi Paolo,

* Paolo Bonzini wrote on Wed, Aug 26, 2009 at 09:21:27AM CEST:
> On 08/25/2009 09:55 PM, Ralf Wildenhues wrote:
> >This was still left TODO.  Of course, I managed to actually commit a
> >patch, that would not expand AC_CHECK_MEMBERS correctly, as that uses
> >_$0_BODY internally, and $0 is still SIM_CHECK_MEMBER (Paolo, it's ugly
> >to have to remember this, so we maybe shouldn't use m4_defn so much):
> 
> Hey, Eric (Blake) is the big optimizer in Autoconf-land, not me. :-)
> I agree with you, though of course I don't like this kind of relying
> on Autoconf internals.

Hmm, I guess I didn't explain myself well.  What I meant was this: if we
override Autoconf macros, say in config/override.m4, then we should try
to use something other than m4_defn.  For example AU_ALIAS should work
better.  This wasn't meant as a bug report against Autoconf, more as a
hint how to avoid such override issues in GCC.

> The patch seems fine.

Well, since I meant it to be two patches, and exactly to remind of the
wrong defn use, I propose that I do eat my own dogfood and would like to
commit these two patches instead.  The end result is the same, as are
the generated files in between.

Thanks,
Ralf

Fix SIM_CHECK_MEMBER definition with Autoconf 2.64.

sim/common/ChangeLog:
2009-08-26  Ralf Wildenhues  <Ralf.Wildenhues@gmx.de>

	* aclocal.m4 (SIM_CHECK_MEMBER): Use AU_ALIAS to define, not defn.
	* configure: Regenerate.

diff --git a/sim/common/aclocal.m4 b/sim/common/aclocal.m4
index 24b1aa2..6820164 100644
--- a/sim/common/aclocal.m4
+++ b/sim/common/aclocal.m4
@@ -945,7 +945,7 @@ dnl
 dnl  ---------------------------------------------------------
 dnl  AGGREGATE.MEMBER is for instance `struct passwd.pw_gecos', shell
 dnl  variables are not a valid argument.
-AC_DEFUN([SIM_CHECK_MEMBER], defn([AC_CHECK_MEMBER]))
+AU_ALIAS([SIM_CHECK_MEMBER], [AC_CHECK_MEMBER])
 dnl
 dnl Translated from a FC2 autoconf-2.59-3 installation.
 dnl  SIM_CHECK_MEMBERS([AGGREGATE.MEMBER, ...])


Remove SIM_CHECK_MEMBER*.

sim/common/ChangeLog:
2009-08-26  Ralf Wildenhues  <Ralf.Wildenhues@gmx.de>

	* aclocal.m4 (SIM_CHECK_MEMBER, SIM_CHECK_MEMBERS)
	(SIM_CHECK_MEMBERS_1): Remove.
	* configure.ac: Replace SIM_CHECK_MEMBERS call with equivalent
	AC_CHECK_MEMBERS one.
	* configure: Regenerate.

diff --git a/sim/common/aclocal.m4 b/sim/common/aclocal.m4
index 6820164..7beaebd 100644
--- a/sim/common/aclocal.m4
+++ b/sim/common/aclocal.m4
@@ -933,50 +933,3 @@ AC_SUBST(CGEN_MAINT)
 AC_SUBST(cgendir)
 AC_SUBST(cgen)
 ])
-dnl FIXME: When upgrading to modern autoconf, remove
-dnl        SIM_CHECK_MEMBER and SIM_CHECK_MEMBERS et al and use
-dnl        AC_CHECK_MEMBERS from autoconf.
-dnl
-dnl Translated from a FC2 autoconf-2.59-3 installation.
-dnl  AC_CHECK_MEMBER(AGGREGATE.MEMBER,
-dnl                  [ACTION-IF-FOUND], [ACTION-IF-NOT-FOUND],
-dnl                  [INCLUDES])
-dnl
-dnl  ---------------------------------------------------------
-dnl  AGGREGATE.MEMBER is for instance `struct passwd.pw_gecos', shell
-dnl  variables are not a valid argument.
-AU_ALIAS([SIM_CHECK_MEMBER], [AC_CHECK_MEMBER])
-dnl
-dnl Translated from a FC2 autoconf-2.59-3 installation.
-dnl  SIM_CHECK_MEMBERS([AGGREGATE.MEMBER, ...])
-dnl except we just work with a limited set of fixed includes.
-dnl
-AC_DEFUN([SIM_CHECK_MEMBERS_1],
-[ifelse($#, 1,
-[SIM_CHECK_MEMBER([$1],
-AC_DEFINE_UNQUOTED([HAVE_]translit([$1], [a-z .], [A-Z__]), 1,
-[Define to 1 if ]patsubst([$1],
-[^[^.]*\.])[ is a member of ]patsubst([$1], [\..*])[. ]),,
-[#ifdef HAVE_SYS_TYPES_H
-#include <sys/types.h>
-#endif
-#ifdef HAVE_SYS_STAT_H
-#include <sys/stat.h>
-#endif])],
-[SIM_CHECK_MEMBER([$1],
-AC_DEFINE_UNQUOTED([HAVE_]translit([$1], [a-z .], [A-Z__]), 1,
-[Define to 1 if ]patsubst([$1],
-[^[^.]*\.])[ is a member of ]patsubst([$1], [\..*])[. ]),,
-[#ifdef HAVE_SYS_TYPES_H
-#include <sys/types.h>
-#endif
-#ifdef HAVE_SYS_STAT_H
-#include <sys/stat.h>
-#endif])
-SIM_CHECK_MEMBERS_1(builtin(shift,$@))])])dnl SIM_CHECK_MEMBERS
-dnl
-AC_DEFUN([SIM_CHECK_MEMBERS],
-[ifelse($#, 1, [SIM_CHECK_MEMBERS_1($1)],
-[errprint(__file__:__line__:
-[This SIM_CHECK_MEMBERS only supports one argument,]
-[the list of struct tests])])])dnl SIM_CHECK_MEMBERS
diff --git a/sim/common/configure.ac b/sim/common/configure.ac
index 5516d9e..5f5845a 100644
--- a/sim/common/configure.ac
+++ b/sim/common/configure.ac
@@ -38,10 +38,16 @@ AC_SUBST(TARGET_SUBDIR)
 # These aren't all needed yet, but will be eventually.
 AC_CHECK_HEADERS(stdlib.h string.h strings.h time.h sys/times.h sys/stat.h sys/mman.h)
 AC_CHECK_FUNCS(mmap munmap lstat truncate ftruncate)
-SIM_CHECK_MEMBERS([[struct stat.st_dev], [struct stat.st_ino],
+AC_CHECK_MEMBERS([[struct stat.st_dev], [struct stat.st_ino],
 [struct stat.st_mode], [struct stat.st_nlink], [struct stat.st_uid],
 [struct stat.st_gid], [struct stat.st_rdev], [struct stat.st_size],
 [struct stat.st_blksize], [struct stat.st_blocks], [struct stat.st_atime],
-[struct stat.st_mtime], [struct stat.st_ctime]])
+[struct stat.st_mtime], [struct stat.st_ctime]], [], [],
+[[#ifdef HAVE_SYS_TYPES_H
+#include <sys/types.h>
+#endif
+#ifdef HAVE_SYS_STAT_H
+#include <sys/stat.h>
+#endif]])
 AC_OUTPUT(Makefile,
 [case x$CONFIG_HEADERS in xcconfig.h:config.in) echo > stamp-h ;; esac])