summaryrefslogtreecommitdiff
path: root/m4
diff options
context:
space:
mode:
authorMike Frysinger <vapier@gentoo.org>2010-09-12 03:39:55 -0400
committerDmitry V. Levin <ldv@altlinux.org>2010-09-15 15:36:00 +0000
commit6d12afdac423ccdc9182022987130a3346a6405f (patch)
tree659a7223dab8e751e3b98f03f62b631d41165507 /m4
parent260e571343b12221ad848a9c972fc5518d66ef2c (diff)
downloadstrace-6d12afdac423ccdc9182022987130a3346a6405f.tar.gz
Fix long long little endian detection when cross-compiling
The long long endian detection code does an AC_TRY_RUN() and since that doesn't work when cross-compiling, it sets a fallback value. However, rather than do any sort of default endian detection, the code simply sets it to "no". This probably breaks most little endian systems out there when cross-compiling for them. It certainly breaks Blackfin systems. So use the common endian detection code provided by autoconf and key off of that when cross-compiling. * configure.ac: Call AC_C_BIGENDIAN. * m4/long_long.m4 (AC_LITTLE_ENDIAN_LONG_LONG): Set cross-compiling logic based on ac_cv_c_bigendian. Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Diffstat (limited to 'm4')
-rw-r--r--m4/long_long.m48
1 files changed, 6 insertions, 2 deletions
diff --git a/m4/long_long.m4 b/m4/long_long.m4
index 98ca3707e..5c5c80aec 100644
--- a/m4/long_long.m4
+++ b/m4/long_long.m4
@@ -57,8 +57,12 @@ int main () {
return 0;
return 1;
}
-]])],[ac_cv_have_little_endian_long_long=yes],[ac_cv_have_little_endian_long_long=no],[# Should try to guess here
-ac_cv_have_little_endian_long_long=no
+]])],[ac_cv_have_little_endian_long_long=yes],[ac_cv_have_little_endian_long_long=no],[
+if test "x$ac_cv_c_bigendian" = "xyes"; then
+ ac_cv_have_little_endian_long_long=no
+else
+ ac_cv_have_little_endian_long_long=yes
+fi
])])
AC_MSG_RESULT($ac_cv_have_little_endian_long_long)
if test "$ac_cv_have_little_endian_long_long" = yes