On 12 June 2015 at 04:23, Robert Yang <[email protected]> wrote:
> + for var in tocheck.split():
> + if var.startswith('ac_cv_'):
> + varname = var.split('=')[0]
> + cfgscript=d.expand("${S}/configure")
> + if os.path.exists(cfgscript):
> + statement = "grep -q -F %s %s > /dev/null" %
> (varname, cfgscript)
>
So your assumption that ac_cv_* variables are referenced in full in
configure is bad. This is what strace has for the giant AC_CHECK_HEADERS
block:
for ac_header in asm/cachectl.h asm/sysmips.h bluetooth/bluetooth.h elf.h
inttypes.h ioctls.h linux/bsg.h linux/falloc.h linux/filter.h
linux/hiddev.h linux/mmtimer.h linux/perf_event.h linux/seccomp.h
linux/securebits.h linux/utsname.h mqueue.h netinet/sctp.h poll.h scsi/sg.h
stropts.h sys/conf.h sys/epoll.h sys/fanotify.h sys/filio.h sys/ioctl.h
sys/poll.h sys/reg.h sys/vfs.h sys/xattr.h
do :
as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
ac_fn_c_check_header_mongrel "$LINENO" "$ac_header" "$as_ac_Header"
"$ac_includes_default"
The variable name is constructed, so this test can't work.
Ross
--
_______________________________________________
Openembedded-core mailing list
[email protected]
http://lists.openembedded.org/mailman/listinfo/openembedded-core
