autoconf
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Gdb compilation argz.z No Such file or directory


From: Keith Marshall
Subject: Re: Gdb compilation argz.z No Such file or directory
Date: Fri, 2 May 2008 21:50:34 +0100
User-agent: KMail/1.9.1

Hi Mudeem,

This looks suspiciously like a cross-compiler installation problem.

On Friday 02 May 2008 20:32, Mudeem Siddiqui wrote:
> configure:5589: checking argz.h usability

Here, we see configure correctly invoking the cross-compiler, to check 
that argz.h can be compiled:

> configure:5601: mipsel-uclibc-gcc -c
> -I/home/mudeem/QV140X/LINUX/KERNEL/linux-2.4.25/include  conftest.c

And the outcome says "no, there's no such header in my search path":

> >&5 conftest.c:64:18: argz.h: No such file or directory
> configure:5607: $? = 1
>
> configure: failed program was:
> [...elided...]
> configure:5630: result: no
> configure:5634: checking argz.h presence

Yet here, we see an (unrelated?) preprocessor being invoked, to check 
for presence of the same header:

> configure:5644: /lib/cpp  conftest.c

And the outcome says "yes, this unrelated preprocessor can see that 
header, in *its* search path":

> configure:5650: $? = 0
> configure:5670: result: yes

but that doesn't help you, because it belongs to a different compiler 
tool set.

FWIW, here's what I see with my (correctly installed) i586-mingw32 
cross-compiler, for the minimal test case:

  $ cat configure.ac
  AC_INIT
  AC_PROG_CC
  AC_CHECK_HEADERS([nonsuch.h])

  $ ./configure --build=linux --host=i586-mingw32
  ...
  configure:3240: checking nonsuch.h usability
  configure:3257: i586-mingw32-gcc -c -g -O2  conftest.c >&5
  conftest.c:51:21: nonsuch.h: No such file or directory
  configure:3263: $? = 1
  ...
  configure:3277: result: no
  configure:3281: checking nonsuch.h presence
  configure:3296: i586-mingw32-gcc -E  conftest.c
  conftest.c:18:21: nonsuch.h: No such file or directory
  configure:3302: $? = 1

Note how, in my case, the cross-compiler is invoked a second time, to 
check for presence of the header, and it confirms its nonexistence.

It seems like you need to find out why a preprocessor, which is 
apparently unrelated to your cross-compiler, is being invoked for the 
second phase of the test.  The relevant section of config.log starts:

> configure:2959: checking how to run the C preprocessor
> configure:2994: mipsel-uclibc-gcc -E  conftest.c

and that fails because you are apparently missing appropriate versions 
of limits.h and assert.h, which are expected to be present for any 
correctly installed compiler suite; when it does so fail, configure 
then falls back to using /lib/cpp, which isn't what you want.

Regards,
Keith.




reply via email to

[Prev in Thread] Current Thread [Next in Thread]