Re: [pcre-dev] hang running configure on redhat AS2.1

Top Page
Delete this message
Author: Todd Rinaldo
Date:  
To: pcre-dev
Subject: Re: [pcre-dev] hang running configure on redhat AS2.1

On Oct 29, 2009, at 4:16 AM, Philip Hazel wrote:

> On Wed, 28 Oct 2009, Todd Rinaldo wrote:
>
>> I'm running configure for PCRE 8.00 on Redhat AS2.1 and hanging
>> during
>> the second occurrence of "checking for alias support in the
>> linker". I
>> was hoping you could point me towards a means to fix this or some way
>> to get more information as to why it's hanging here.
>
> Oh gosh, that is somewhere in the magic that is generated by autoconf,
> about which I know nothing, and I rather suspect that most folks on
> this
> list know nothing about it either. Prove me wrong, anybody?
>
> Under Gentoo Linux, I get two identical lines:
>
> checking for alias support in the linker... no
> checking for alias support in the linker... no
>
> and then it moves on. I've had a quick look at ./configure, and it
> seems
> that this check is concerned with compiling the C++ wrapper. First, I
> suggest you try ./configure --disable-cpp to see if you can compile
> the
> main C library without the C++ wrapper. If you really need the C++
> wrapper, all I can suggest is that you hack about in ./configure, look
> for that message, and cut out the section where it appears.
>
> Further checking shows that the test originates in ./configure.ac. Is
> that in the distribution? Perhaps not. You can get it from the svn
> repository if you want to play with things at that level:
>


Thanks for the pointers. I've narrowed this problem down to an hang in
the linker call in that section:

g++ -o conftest -O2 -Wl,-
i__ZN7pcrecpp6no_argE:__ZN7pcrecpp2RE6no_argE conftest.cpp >&5

The hang is actually in the underlying ld command being called. Redhat
AS2.1 uses GCC 2.96 I cannot re-produce this hang on other gcc 2.96
systems. A google search for mention of linker hangs was fruitless for
me. Changing the optimizer to 0 (-O0) had no effect.

In my case, I'm compiling PCRE statically for exim 4.70, so it appears
that the --disable-cpp option will suffice for me. If you want to
introduce an alternative fix, I would be happy to test it for you. The
only option I see would be to introduce a skip on this test if this
shell command comes back non-zero

cat /etc/redhat-release |grep -c Pensacola

Thanks,
Todd Rinaldo