On 04/10/2011 9:05 AM, Kåre Edvardsen wrote:
On Tue, Oct 4, 2011 at 12:13 PM, Marco Atzeri wrote:
(snip)
Kare,
in addition to the TOFU advise,
running configure of grib_api-1.9.9_libtool.tar.gz
on my XP-SP3 gave no problem.
"Configuration completed."
The ANSI C test is a basic one
configure:6808: checking for ANSI C header files
configure:6828: gcc -c -g -O2 conftest.c>&5
configure:6828: $? = 0
configure:6901: gcc -o conftest.exe -g -O2 conftest.c>&5
configure:6901: $? = 0
configure:6901: ./conftest.exe
configure:6901: $? = 0
configure:6912: result: yes
I reproduced the hang on my Cygwin CYGWIN_NT-6.0 1.7.9(0.237/5/3)
2011-03-29 10:10 i686 Cygwin
and gcc version 4.5.3
This appears to be the critical spot (line 6820-6828)
if (eval "$ac_cpp conftest.$ac_ext") 2>&5 |
$EGREP "memchr">/dev/null 2>&1; then
:
else
ac_cv_header_stdc=no
fi
It tries to run the preprocessor and pipe its output into egrep.
Unfortunately, the EGREP variable appears to be empty, so the end
result is similar to:
gcc -E conftest.c | memchr
Trying the above from the bash prompt (in both mintty and classic
Cygwin prompt) results in an (unkillable from Cygwin) cc1.exe which is
perpetually waiting for output:
$ ps
PID PPID PGID WINPID TTY UID STIME COMMAND
4484 1 4484 4752 con 539771 Oct 3 /usr/bin/XWin
7848 1 7848 7848 ? 539771 14:03:11 /usr/bin/mintty
6468 7848 6468 7460 1 539771 14:03:11 /usr/bin/bash
6692 1 6692 6692 ? 539771 10:12:37 /usr/bin/mintty
4600 6692 4600 7156 3 539771 10:12:37 /usr/bin/ssh
6272 6468 6272 6356 1 539771 13:08:43 /usr/bin/gcc-4
O 7492 6272 6272 6856 1 539771 13:08:43
/usr/lib/gcc/i686-pc-cygwin/4.5.3/cc1
7404 6468 7404 524 1 539771 13:08:46 /usr/bin/ps
So, the workaround is to replace $EGREP with egrep in configure; then
configure runs successfully.
This seems to be some peculiarity of Cygwin's current GCC 4.5.3;
neither GCC 3.4.4 in Cygwin 1.7 nor GCC 4.3.4 in Cygwin 1.7 nor GCC
4.3.2 in Cygwin 1.5 exhibit this hang.
Csaba
Csaba& Marco.
Marco: Your try on the XP-SP3 gave no problem. My try on Win 7 Pro
hangs, but as Csaba points out: This seems to be some peculiarity of
Cygwin's current GCC 4.5.3
Csaba: Replacing $EGREP with egrep in configure did make the difference!
Thanx to both of you for excellent help.
It wouldn't surprise me at all if the real problem were an earlier fork
failure that caused $EGREP to come up blank. Why gcc would then hang on
the resulting broken pipeline is a different question, tho...
Ryan
--
Problem reports: http://cygwin.com/problems.html
FAQ: http://cygwin.com/faq/
Documentation: http://cygwin.com/docs.html
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
