On Fri, Oct 10, 2014 at 05:33:00PM +0200, Hector Oron wrote: > Package: gcc-4.9 > Version: 4.9.1-16 > Severity: important > > Hello, > > Found a FTBFS while trying to build unzip package in Debian/sid on armhf host. > > gcc -c -g -O2 -fstack-protector-strong -Wall -D_FORTIFY_SOURCE=2 -I. -DACORN_FTYPE_NFS -DWILD_STOP_AT_DIR -DLARGE_FILE_SUPPORT -DUNICODE_SUPPORT -DUNICODE_WCHAR -DUTF8_MAYBE_NATIVE -DNO_LCHMOD -DDATE_FORMAT=DF_YMD -DUSE_BZIP2 -DIZ_HAVE_UXUIDGID -DNOMEMCPY -DNO_WORKING_ISPRINT inflate.c > gcc -c -g -O2 -fstack-protector-strong -Wall -D_FORTIFY_SOURCE=2 -I. -DACORN_FTYPE_NFS -DWILD_STOP_AT_DIR -DLARGE_FILE_SUPPORT -DUNICODE_SUPPORT -DUNICODE_WCHAR -DUTF8_MAYBE_NATIVE -DNO_LCHMOD -DDATE_FORMAT=DF_YMD -DUSE_BZIP2 -DIZ_HAVE_UXUIDGID -DNOMEMCPY -DNO_WORKING_ISPRINT list.c > gcc -c -g -O2 -fstack-protector-strong -Wall -D_FORTIFY_SOURCE=2 -I. -DACORN_FTYPE_NFS -DWILD_STOP_AT_DIR -DLARGE_FILE_SUPPORT -DUNICODE_SUPPORT -DUNICODE_WCHAR -DUTF8_MAYBE_NATIVE -DNO_LCHMOD -DDATE_FORMAT=DF_YMD -DUSE_BZIP2 -DIZ_HAVE_UXUIDGID -DNOMEMCPY -DNO_WORKING_ISPRINT match.c > gcc -c -g -O2 -fstack-protector-strong -Wall -D_FORTIFY_SOURCE=2 -I. -DACORN_FTYPE_NFS -DWILD_STOP_AT_DIR -DLARGE_FILE_SUPPORT -DUNICODE_SUPPORT -DUNICODE_WCHAR -DUTF8_MAYBE_NATIVE -DNO_LCHMOD -DDATE_FORMAT=DF_YMD -DUSE_BZIP2 -DIZ_HAVE_UXUIDGID -DNOMEMCPY -DNO_WORKING_ISPRINT process.c > process.c: In function 'wide_to_local_string': > process.c:2453:7: warning: variable 'state_dependent' set but not used [-Wunused-but-set-variable] > int state_dependent; > ^ > /tmp/cc0hC5mg.s: Assembler messages: > /tmp/cc0hC5mg.s:898: Error: offset out of range > /tmp/cc0hC5mg.s:899: Error: offset out of range > make[1]: *** [process.o] Error 1 > unix/Makefile:234: recipe for target 'process.o' failed > make[1]: Leaving directory '/«PKGBUILDDIR»' > make: *** [build] Error 2 > debian/rules:34: recipe for target 'build' failed > dpkg-buildpackage: error: debian/rules build gave error exit status 2 > > > After trying different toolchain versions, for binutils and gcc-4.9, it builds with -O1, but fails with -O2. So, apparently the bug is on the GCC optimizer. > > I was about to file a bug upstream for that one, but I was unable to reproduce with gcc-snapshot, therefore I file it here for now. Re-tested with gcc-4.9 (4.9.2-1) same thing happens. Using gcc-snapshot (20141017-1), I am unable to reproduce. Regards, -- Hector Oron
Attachment:
signature.asc
Description: Digital signature