About 329338 - Declaration conflict for basename

About 329338 - Declaration conflict for basename

[Vladimir Mezentsev]

Hi team,

I am evaluating 29338 - Declaration conflict for basename 
<https://sourceware.org/bugzilla/show_bug.cgi?id=29338>
and I see a regression in the efbecf1e953 commit.

% git log --pretty=format:"%h - %an, %ar --> %s" --graph --name-status*
...
  efbecf1e953 - Nick Clifton, 6 days ago --> Update version number|
| M     bfd/configure
| M     bfd/po/bfd.pot
| M     bfd/version.m4
| M     binutils/configure
| M     binutils/po/binutils.pot
| M     gas/configure
| M     gas/po/gas.pot
| M     gold/po/gold.pot
| M     gprof/configure
| M     gprof/po/gprof.pot
| M     gprofng/common/config.h.in       <---------------- This update 
is wrong.
| M     gprofng/configure
| M     gprofng/libcollector/configure
| M     ld/configure
| M     ld/po/ld.pot
| M     opcodes/configure
| M     opcodes/po/opcodes.pot
| M     zlib/Makefile.in
| M     zlib/aclocal.m4
| M     zlib/configure


But the real problem is in our configure.ac scripts.

% grep -n AC_CONFIG_HEADERS gprofng/configure.ac
220:AC_CONFIG_HEADERS([config.h:common/config.h.in])
   This config.h is used in all our sources.

% grep -n AC_CONFIG_HEADERS gprofng/libcollector/configure.ac
65:AC_CONFIG_HEADERS([lib-config.h:../common/config.h.in])

  autoreconf updates ../common/config.h.in, which is not what we wanted.
  lib-config.h is not used. In build time, we use config.h from the ../ 
directory.
  We use AC_CONFIG_HEADERS() in gprofng/libcollector/configure.ac to 
pass -DHAVE_CONFIG_H for compilation.



I can make the following fix:
% git diff gprofng/libcollector/configure.ac
diff --git a/gprofng/libcollector/configure.ac 
b/gprofng/libcollector/configure.ac
index d3c6bd5588d..2e2185a11a3 100644
--- a/gprofng/libcollector/configure.ac
+++ b/gprofng/libcollector/configure.ac
@@ -62,6 +62,6 @@ ACX_PROG_CC_WARNING_OPTS([-Wno-nonnull-compare], 
[GPROFNG_NO_NONNULL_COMPARE_CFL
  AC_SUBST(GPROFNG_NO_NONNULL_COMPARE_CFLAGS)

  AC_CONFIG_FILES([Makefile])
-AC_CONFIG_HEADERS([lib-config.h:../common/config.h.in])
+AC_CONFIG_HEADERS([lib-config.h])
  AC_OUTPUT

% ( cd gprofng && autoreconf )


This will fix build.
But gprofng/libcollector/lib-config.h.in will be created.
% git status
On branch vmezents/2_39
Changes not staged for commit:
   (use "git add <file>..." to update what will be committed)
   (use "git restore <file>..." to discard changes in working directory)
     modified:   gprofng/common/config.h.in
     modified:   gprofng/libcollector/Makefile.in
     modified:   gprofng/libcollector/configure
     modified:   gprofng/libcollector/configure.ac

Untracked files:
   (use "git add <file>..." to include in what will be committed)
* gprofng/libcollector/lib-config.h.in*


Maybe there is another way to fix this.
We need a fake AC_CONFIG_HEADERS that will pass -DHAVE_CONFIG_H for 
compilation but will not generate lib-config.h and lib-config.h.in.


Thank you,
-Vladimir

Re: About 329338 - Declaration conflict for basename

[Alan Modra]

On Wed, Jul 13, 2022 at 11:41:49PM -0700, Vladimir Mezentsev via Binutils wrote:
> Hi team,
> 
> I am evaluating 29338 - Declaration conflict for basename
> <https://sourceware.org/bugzilla/show_bug.cgi?id=29338>
> and I see a regression in the efbecf1e953 commit.
> 
> % git log --pretty=format:"%h - %an, %ar --> %s" --graph --name-status*
> ...
>  efbecf1e953 - Nick Clifton, 6 days ago --> Update version number|
> | M     bfd/configure
> | M     bfd/po/bfd.pot
> | M     bfd/version.m4
> | M     binutils/configure
> | M     binutils/po/binutils.pot
> | M     gas/configure
> | M     gas/po/gas.pot
> | M     gold/po/gold.pot
> | M     gprof/configure
> | M     gprof/po/gprof.pot
> | M     gprofng/common/config.h.in       <---------------- This update is
> wrong.

Did commit b95830931f fix the problem?

-- 
Alan Modra
Australia Development Lab, IBM

Re: About 329338 - Declaration conflict for basename

[Vladimir Mezentsev]

On 7/14/22 00:35, Alan Modra wrote:
> On Wed, Jul 13, 2022 at 11:41:49PM -0700, Vladimir Mezentsev via Binutils wrote:
>> Hi team,
>>
>> I am evaluating 29338 - Declaration conflict for basename
>> <https://sourceware.org/bugzilla/show_bug.cgi?id=29338>
>> and I see a regression in the efbecf1e953 commit.
>>
>> % git log --pretty=format:"%h - %an, %ar --> %s" --graph --name-status*
>> ...
>>   efbecf1e953 - Nick Clifton, 6 days ago --> Update version number|
>> | M     bfd/configure
>> | M     bfd/po/bfd.pot
>> | M     bfd/version.m4
>> | M     binutils/configure
>> | M     binutils/po/binutils.pot
>> | M     gas/configure
>> | M     gas/po/gas.pot
>> | M     gold/po/gold.pot
>> | M     gprof/configure
>> | M     gprof/po/gprof.pot
>> | M     gprofng/common/config.h.in       <---------------- This update is
>> wrong.
> Did commit b95830931f fix the problem?

I don't know. This commit is in the master branch.
I  klnow the master branch has no problems now.
I know If I run:
    cd gprofng;
   autoreconf
this fiixes problem.

Somehow Nick Clifton updated gprofng/common/config.h.in differently.
Maybe `cd gprofng/libcollector && autoreconf`


In any case,  we need to fix gprofng/libcollector/configure.ac.
For the  binutils-2_39-branch branch, we can rerun `cd gprofng && 
autoreconf`. This fixes build.

-Vladimir




>