Summary: | grub-2.00* x86_64-efi built with gcc-4.8 fails to load kernel | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Justin Lecher (RETIRED) <jlec> |
Component: | [OLD] Core system | Assignee: | Mike Gilbert <floppym> |
Status: | RESOLVED FIXED | ||
Severity: | major | CC: | base-system, bkohler, jlp.bugs, pums974, tdalman, toolchain |
Priority: | Highest | ||
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Bug Depends on: | |||
Bug Blocks: | 461954 | ||
Attachments: | build.log grub with gcc-4.8 |
Description
Justin Lecher (RETIRED)
2013-05-01 12:47:25 UTC
The i386-pc platform works fine for me when built with gcc-4.8. I guess there could be an issue with the x86_64-efi platform. Can you give the live ebuild (grub-9999) a shot? If that works, we can look for a patch. Also, be aware that I added a USE flag (multislot) to the live ebuild which toggles the grub -> grub2 rename. I will also see if I can reproduce this in qemu. I can reproduce the problem with grub-2.00-r3 under qemu with OVMF. The latest grub trunk works fine. Bisected to this commit, which is the first in a refactoring series to remove nested functions from the codebase. http://bzr.savannah.gnu.org/lh/grub/trunk/grub/revision/4648 I'm not comfortable backporting this myself; it would probably be better to add a snapshot. Justin: Please confirm that the grub trunk works on actual EFI hardware; I'm just testing with an emulator. I couldn't reboot until now. Will Do it after breakfast. I tried revno 4940 which works correctly. Seems to be an important bug Can the fix be backported ? Is grub-9999 safe ? (In reply to Alexandre from comment #7) > Seems to be an important bug + > Can the fix be backported ? Its definitely needed. I will set priority higher. > Is grub-9999 safe ? No guarantee for "safe" but I rebuilded it once after this report using 9999 and it worked. Sorry for the delay guys. I have added a new snapshot, which is currently hard-masked. I will unmask it as soon as you can confirm that it works without any major problems. +*grub-2.00_p4946 (28 Jul 2013) + + 28 Jul 2013; Mike Gilbert <floppym@gentoo.org> +grub-2.00_p4946.ebuild, + grub-9999.ebuild: + Add new snapshot for bug 468136. As a bonus, we don't need autogen anymore. Yay! + 28 Jul 2013; Mike Gilbert <floppym@gentoo.org> grub-2.00_p4946.ebuild, + grub-9999.ebuild: + We don't need autogen with a pristine snapshot. I upgraded 2 machines to gcc-4.8.1 this morning and got a chance to test this. Here are my results: (machine A) gcc-4.8.1 + grub-2.00-r4 + x86_64-efi = fail, hangs on "Loading kernel x.y.z" gcc-4.8.1 + grub-2.00_p4946 + x86_64-efi = success (machine B) <did not attempt to reproduce failure again, but..> gcc-4.8.1 + grub-2.00_p4946 + x86_64-efi = success +*grub-2.00_p5086 (04 Aug 2013) + + 04 Aug 2013; Mike Gilbert <floppym@gentoo.org> +grub-2.00_p5086.ebuild, + grub-9999.ebuild: + New snapshot, resolves bug 468136. Append -fuse-ld=bfd to LDFLAGS for gcc-4.8+ + as a workaround for bug 439082. |