Build process dies during compile with make: *** [out/Release/nacl_bootstrap_raw] Error 252 Reproducible: Always Steps to Reproduce: 1.set CLFAGS="-march=atom -O2 -pipe" in /etc/make.conf or through export 2.emerge chromium (with gcc 4.5.3 or higher) 3.Build fails Actual Results: Build process dies during compile with make: *** [out/Release/nacl_bootstrap_raw] Error 252 Expected Results: Compile without error It appears that with GCC 4.5.3 and march=atom set in CFLAGS we need to set -Ddisable_nacl=1 during src_configure and not install native client files (similar to arch=arm currently is in the ebuild. I was able to copy the ebuild into a local repo, make the changes, and complete the compile. I am not sure how to parse the CFLAGS in the ebuild to perform this only when $(tc-arch)=amd64 AND march=atom (since this action is not required on other amd64 processors).
Created attachment 302905 [details] emerge --info =www-client/chromium-17.0.963.56
Created attachment 302907 [details] emerge -pqv =www-client/chromium-17.0.963.56
Created attachment 302909 [details] build.log.tar.gz
Attaching patched ebuild that allowed compiling to complete. The patch may break build on systems other than gcc-4.5.3 with march=atom. Need to find a way to parse the CFLAGS to determine if march=atom and then set changes as needed.
Created attachment 302911 [details] patched ebuild
Comment on attachment 302911 [details] patched ebuild --- chromium-17.0.963.56.ebuild 2012-02-17 12:14:45.000000000 +0100 +++ - 2012-02-23 00:49:33.729445623 +0100 @@ -320,7 +320,7 @@ local myarch="$(tc-arch)" if [[ $myarch = amd64 ]] ; then - myconf+=" -Dtarget_arch=x64" + myconf+=" -Dtarget_arch=x64 -Ddisable_nacl=1" elif [[ $myarch = x86 ]] ; then myconf+=" -Dtarget_arch=ia32" elif [[ $myarch = arm ]] ; then @@ -405,11 +405,11 @@ # Install Native Client files on platforms that support it. insinto "${CHROMIUM_HOME}" case "$(tc-arch)" in - amd64) - doexe out/Release/nacl_helper{,_bootstrap} || die - doins out/Release/nacl_irt_x86_64.nexe || die - doins out/Release/libppGoogleNaClPluginChrome.so || die - ;; + # amd64) + #doexe out/Release/nacl_helper{,_bootstrap} || die + #doins out/Release/nacl_irt_x86_64.nexe || die + #doins out/Release/libppGoogleNaClPluginChrome.so || die + #;; x86) doexe out/Release/nacl_helper{,_bootstrap} || die doins out/Release/nacl_irt_x86_32.nexe || die
Comment on attachment 302909 [details] build.log.tar.gz Er, a single file in a tar archive?
1. We're _not_ going to disable NaCl. 2. Please take a look at bug #394683, are you possibly hitting the same issue? 3. Are you sure -march=atom is valid for your CPU? 4. Are you running this or anything else in a chroot?
(In reply to comment #8) Paweł Hajdan, Jr., thanks for looking into this. > 1. We're _not_ going to disable NaCl. > 2. Please take a look at bug #394683, are you possibly hitting the same issue? After reviewing this bug, yes it does appear to be the same issue and I'll use the suggested workaround. Sorry I missed that when I was searching through the bugs yesterday. > 3. Are you sure -march=atom is valid for your CPU? > 4. Are you running this or anything else in a chroot? This was being compiled in a chroot environment before being transferred to the target machine. Given this, please close this bug as invalid.
(In reply to comment #9) > Given this, please close this bug as invalid. Closing. *** This bug has been marked as a duplicate of bug 394683 ***