Build problem on ia64

Build problem on ia64

[Zdenek Dvorak]

Hello,

when configure is run for libstdc++-v3 during build, I get the following error:

configure:2417: checking for C compiler default output file name
configure:2420:  /abuild/gcc-test/gcc/objdir/gcc/xgcc -B/abuild/gcc-test/gcc/objdir/gcc/ -B/usr/local/ia64-unknown-linux-gnu/bin/ -B/usr/local/ia64-unknown-linux-gnu/lib/ -isystem /usr/local/ia64-unknown-linux-gnu/include -isystem /usr/local/ia64-unknown-linux-gnu/sys-include -O2 -g -O2 -O2 -g -O2  conftest.c  >&5
configure:2423: $? = 0
configure:2469: result: a.out
configure:2474: checking whether the C compiler works
configure:2480: ./a.out
./a.out: error while loading shared libraries: libunwind.so.7: cannot open shared object file: No such file or directory
configure:2483: $? = 127
configure:2492: error: cannot run C compiled programs.
If you meant to cross compile, use `--host'.
See `config.log' for more details.

libunwind.so.7 is built in gcc directory, but linker apparently is not
told in where to find it (with LD_LIBRARY_PATH set the things work
correctly).

Where is the right place to fix this (I have very little understanding
of how gcc's build system works :-( )?

Zdenek

Re: Build problem on ia64

[Paolo Carlini]

Zdenek Dvorak wrote:

>Hello,
>
>when configure is run for libstdc++-v3 during build, I get the following error:
>  
>
Isn't this basically the long standing

    http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17369

???

Indeed, I cannot bootstrap on ia64 either...

Paolo.

Re: Build problem on ia64

[H. J. Lu]

On Wed, Sep 22, 2004 at 12:00:14PM +0200, Zdenek Dvorak wrote:
> Hello,
> 
> when configure is run for libstdc++-v3 during build, I get the following error:
> 
> configure:2417: checking for C compiler default output file name
> configure:2420:  /abuild/gcc-test/gcc/objdir/gcc/xgcc -B/abuild/gcc-test/gcc/objdir/gcc/ -B/usr/local/ia64-unknown-linux-gnu/bin/ -B/usr/local/ia64-unknown-linux-gnu/lib/ -isystem /usr/local/ia64-unknown-linux-gnu/include -isystem /usr/local/ia64-unknown-linux-gnu/sys-include -O2 -g -O2 -O2 -g -O2  conftest.c  >&5
> configure:2423: $? = 0
> configure:2469: result: a.out
> configure:2474: checking whether the C compiler works
> configure:2480: ./a.out
> ./a.out: error while loading shared libraries: libunwind.so.7: cannot open shared object file: No such file or directory
> configure:2483: $? = 127
> configure:2492: error: cannot run C compiled programs.
> If you meant to cross compile, use `--host'.
> See `config.log' for more details.
> 
> libunwind.so.7 is built in gcc directory, but linker apparently is not
> told in where to find it (with LD_LIBRARY_PATH set the things work
> correctly).
> 
> Where is the right place to fix this (I have very little understanding
> of how gcc's build system works :-( )?

See

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17464


H.J.

Re: Build problem on ia64 (reverting patch on 3.4 branch)

[Geert Bosch]

On Sep 22, 2004, at 11:51, H. J. Lu wrote:
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17464

Apparently, a similar failure occurs with Ada (make gnatlib_and_tools)
on the GCC 3.4 branch. In my opinion a patch that:
   1) requires updating the system linker, and
   2) has such a high potential for bootstrap/build problems
should not go on a stable branch like the gcc-3_4-branch. If we have
so many problems building our own tools, likely people will run in
related issues on their own projects to. This is not a safe patch!
As far as I understand the new ABI, there should still be backward
compatibility, so there would be no pressing need for gcc-3_4 to
implement this new ABI.

Because of the above concerns I'd like to request that this patch
be reverted on the 3.4 branch.

   -Geert

Re: Build problem on ia64 (reverting patch on 3.4 branch)

[H. J. Lu]

On Thu, Sep 23, 2004 at 02:12:29PM -0400, Geert Bosch wrote:
> 
> On Sep 22, 2004, at 11:51, H. J. Lu wrote:
> >http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17464
> 
> Apparently, a similar failure occurs with Ada (make gnatlib_and_tools)
> on the GCC 3.4 branch. In my opinion a patch that:
>   1) requires updating the system linker, and
>   2) has such a high potential for bootstrap/build problems

Please try

http://gcc.gnu.org/ml/gcc-patches/2004-09/msg02413.html

> should not go on a stable branch like the gcc-3_4-branch. If we have
> so many problems building our own tools, likely people will run in
> related issues on their own projects to. This is not a safe patch!
> As far as I understand the new ABI, there should still be backward
> compatibility, so there would be no pressing need for gcc-3_4 to
> implement this new ABI.
> 
> Because of the above concerns I'd like to request that this patch
> be reverted on the 3.4 branch.

The problem is nothing new. If you make the system libgcc_s.so.1
unavailable to the newly built gcc, many things will fail. See

http://gcc.gnu.org/ml/gcc/2004-09/msg00209.html


H.J.

Re: Build problem on ia64 (reverting patch on 3.4 branch)

[Geert Bosch]

On Sep 23, 2004, at 14:42, H. J. Lu wrote:

> On Thu, Sep 23, 2004 at 02:12:29PM -0400, Geert Bosch wrote:
>>
>> On Sep 22, 2004, at 11:51, H. J. Lu wrote:
>>> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17464
>>
>> Apparently, a similar failure occurs with Ada (make gnatlib_and_tools)
>> on the GCC 3.4 branch. In my opinion a patch that:
>>   1) requires updating the system linker, and
>>   2) has such a high potential for bootstrap/build problems
>
> Please try
>
> http://gcc.gnu.org/ml/gcc-patches/2004-09/msg02413.html
>

It fails bootstrap complaining about a missing separator in 
gcc/Makefile.
It's this line:

   # Set up library path if needed.
   @set_gcc_lib_path@

Any idea what might be going wrong?

Re: Build problem on ia64 (reverting patch on 3.4 branch)

[H. J. Lu]

On Thu, Sep 23, 2004 at 03:38:43PM -0400, Geert Bosch wrote:
> 
> On Sep 23, 2004, at 14:42, H. J. Lu wrote:
> 
> >On Thu, Sep 23, 2004 at 02:12:29PM -0400, Geert Bosch wrote:
> >>
> >>On Sep 22, 2004, at 11:51, H. J. Lu wrote:
> >>>http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17464
> >>
> >>Apparently, a similar failure occurs with Ada (make gnatlib_and_tools)
> >>on the GCC 3.4 branch. In my opinion a patch that:
> >>  1) requires updating the system linker, and
> >>  2) has such a high potential for bootstrap/build problems
> >
> >Please try
> >
> >http://gcc.gnu.org/ml/gcc-patches/2004-09/msg02413.html
> >
> 
> It fails bootstrap complaining about a missing separator in 
> gcc/Makefile.
> It's this line:
> 
>   # Set up library path if needed.
>   @set_gcc_lib_path@
> 
> Any idea what might be going wrong?

You need to regenerate configure and gcc/configure by hand.


H.J.

Re: Build problem on ia64 (reverting patch on 3.4 branch)

[H. J. Lu]

On Thu, Sep 23, 2004 at 02:00:14PM -0700, H. J. Lu wrote:
> On Thu, Sep 23, 2004 at 03:38:43PM -0400, Geert Bosch wrote:
> > 
> > On Sep 23, 2004, at 14:42, H. J. Lu wrote:
> > 
> > >On Thu, Sep 23, 2004 at 02:12:29PM -0400, Geert Bosch wrote:
> > >>
> > >>On Sep 22, 2004, at 11:51, H. J. Lu wrote:
> > >>>http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17464
> > >>
> > >>Apparently, a similar failure occurs with Ada (make gnatlib_and_tools)
> > >>on the GCC 3.4 branch. In my opinion a patch that:
> > >>  1) requires updating the system linker, and
> > >>  2) has such a high potential for bootstrap/build problems
> > >
> > >Please try
> > >
> > >http://gcc.gnu.org/ml/gcc-patches/2004-09/msg02413.html
> > >
> > 
> > It fails bootstrap complaining about a missing separator in 
> > gcc/Makefile.
> > It's this line:
> > 
> >   # Set up library path if needed.
> >   @set_gcc_lib_path@
> > 
> > Any idea what might be going wrong?
> 
> You need to regenerate configure and gcc/configure by hand.
> 
>

You also need to regenerate top level Makefile.in.


H.J.

Re: Build problem on ia64 (reverting patch on 3.4 branch)

[Geert Bosch]

On Sep 23, 2004, at 17:39, H. J. Lu wrote:
> You also need to regenerate top level Makefile.in.

Doesn't seem to work: the variable gets replaced with nothing:
> lang.stageprofile:  ada.stageprofile
> lang.stagefeedback:  ada.stagefeedback
>
> # Set up library path if needed.
>
>
> # per-language makefile fragments
> ifneq ($(LANG_MAKEFRAGS),)
> include $(LANG_MAKEFRAGS)
> endif

Do you know where this might have gone wrong?

Re: Build problem on ia64 (reverting patch on 3.4 branch)

[H. J. Lu]

On Fri, Sep 24, 2004 at 03:06:51PM -0400, Geert Bosch wrote:
> 
> On Sep 23, 2004, at 17:39, H. J. Lu wrote:
> >You also need to regenerate top level Makefile.in.
> 
> Doesn't seem to work: the variable gets replaced with nothing:
> >lang.stageprofile:  ada.stageprofile
> >lang.stagefeedback:  ada.stagefeedback
> >
> ># Set up library path if needed.
> >
> >
> ># per-language makefile fragments
> >ifneq ($(LANG_MAKEFRAGS),)
> >include $(LANG_MAKEFRAGS)
> >endif
> 
> Do you know where this might have gone wrong?

Please try today's CVS. My patch is in now. If it still fails for
you, I need to see the output of

# grep SET_GCC_LIB_PATH_CMD "top level Makefile"


H.J.