./configure problems on Fedora Core 8

[Carlos Picazzo]

I've asked on the Fedora Forums and the postgres IRC channel but haven't
received any feedback.  I'm guessing there's just a library I'm missing
and I've tried this in 8.1.x, 8.2.6 and now 8.3.0 with the same result
so at least that's consistent.  But when I do a './configure
--enable-thread-safety', it eventually bombs out saying:

   ...
   checking for sgmlspl... sgmlspl
   checking thread safety of required library functions... no
   configure: error: thread test program failed
   This platform is not thread-safe. Check the file 'config.log'for the
exact reason.

Upon checking the config.log, it says:

   | #define MEMSET_LOOP_LIMIT 1024
   | /* end confdefs.h. */
   | #include "./src/test/thread/thread_test.c"
   configure:22450: result: no
   configure:22459: error: thread test program failed
   This platform is not thread-safe. Check the file 'config.log'for the
exact reason.

Google searches show some posts where there have been instances where
thread_test.c doesn't actually exist but this isn't the case here.  This
completes successfully on an old and obsolete FC4 system so I think it's
just that I'm missing a particular safe thread library or something.
Can anyone point me in the right direction?

Thanks!

Carlos

---------------------------(end of broadcast)---------------------------
TIP 4: Have you searched our list archives?

               http://archives.postgresql.org

[Tom Lane-2]

Carlos Picazzo <ojaiguy@...> writes:
> Upon checking the config.log, it says:

>    | #define MEMSET_LOOP_LIMIT 1024
>    | /* end confdefs.h. */
>    | #include "./src/test/thread/thread_test.c"
>    configure:22450: result: no
>    configure:22459: error: thread test program failed
>    This platform is not thread-safe. Check the file 'config.log'for the
> exact reason.

The relevant error message is probably in the lines before where it
prints the failing test program.

                        regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 6: explain analyze is your friend

[Carlos Picazzo]

On 06.02.2008 10:26 Uhr, Tom Lane wrote:
>
> The relevant error message is probably in the lines before where it
> prints the failing test program.

Thanks for the reply, Tom.

I've attached the both the output of './configure
----enable-thread-safety' and the config.log from 8.3.0.  Naturally, the
config.log was a bit big so I've gzipped it.  Hopefully it will contain
the important information.

Thanks again!

Carlos




---------------------------(end of broadcast)---------------------------
TIP 2: Don't 'kill -9' the postmaster

[Tom Lane-2]

Carlos Picazzo <ojaiguy@...> writes:
> I've attached the both the output of './configure
> ----enable-thread-safety' and the config.log from 8.3.0.  Naturally, the
> config.log was a bit big so I've gzipped it.  Hopefully it will contain
> the important information.

I think the problem is that every single one of the test compilations
produces this error message:

/usr/bin/ld: error in /usr/lib/gcc/i686-redhat-linux/4.2.2/crtend.o(.eh_frame); no .eh_frame_hdr table will be created.

Apparently it is just a warning because the compiler returns status
zero anyway, but nonetheless it will confuse our configure script ---
in particular, I think what's biting you is that the test to see which
threads flags to use can't tell that this isn't a complaint about a
bogus thread switch, so it omits to use -pthread which is necessary.
There are some other tests that might be going wrong too.

I see that the gcc on my stock Fedora 8 machine is 4.1.2, so I suppose
that this 4.2.2 you are using came from someplace other than Red Hat.
I fear it's a bit broken :-(

                        regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 2: Don't 'kill -9' the postmaster