grub-efi: add .bbappend for visionfive2 #553
Conversation
|
I'm waiting to get confirmation from the original patch creator to keep their sign-off, so this shouldn't be merged quite yet, but it'd be good to get general feedback. |
recipes-bsp/grub/grub-efi/.0263-Use-medany-instead-of-large-model-for-RISCV.patch.swp
Outdated
Show resolved
Hide resolved
Builds with newer gcc versions fail for the visionfive2 MACHINE target with the following error: |grub-mkimage: error: relocation 0x2b is not implemented yet. Add a patch to use medany instead of the large model for the target. This could apply to all builds, but only visionfive2 has thus far been tested. See: https://savannah.gnu.org/bugs/?65909 Signed-off-by: Trevor Gamblin <[email protected]>
threexc
force-pushed
the
tgamblin/grub-efi-patch
branch
from
81e45a6 to
5f9c5d6
Compare
| Date: Fri, 3 May 2024 13:18:37 -0400 | ||
| Subject: [PATCH] Use medany instead of large model for RISCV | ||
|
|
||
| Upstream-Status: Submitted [https://savannah.gnu.org/bugs/?65909] |
There was a problem hiding this comment.
why not submit it to oe-core directly ?, the patch is already submitted upstream
There was a problem hiding this comment.
I could. I'll probably have to change the logic to match against TARGET_ARCH or something similar. I'd like to give the original patch creator 24-48h before doing that though.
There was a problem hiding this comment.
On the flipside, I haven't run into this with qemuriscv64 in oe-core. Is this something that should be there instead of in meta-riscv (at least right now)?
There was a problem hiding this comment.
hmmm that is interesting. If its specific to a real h/w then it might make sense to keep it in meta-riscv, although the change looks generic so if it does not hurt qemu, I think it would still make. sense to have it in core. I would ask to analyse it a bit more from regression POV
There was a problem hiding this comment.
It's possible it could be triggered from oe-core without meta-riscv, but I've been doing some builds that way for experimenting/debugging with ptests and haven't seen it. The VisionFive 2 is the only board I've been able to test with as of now (I have an OrangePi RV2 coming soon and an unresponsive BeagleV-Ahead, but no other usable hardware except maybe a Nezha yet).
|
I'm going to leave this one open for a while, as I don't think we can do much with it until #560 is resolved upstream. |
3 participants
Builds with newer gcc versions fail for the visionfive2 MACHINE target with the following error:
|grub-mkimage: error: relocation 0x2b is not implemented yet.
Add a patch to use medany instead of the large model for the target. This could apply to all builds, but only visionfive2 has thus far been tested.
See: https://savannah.gnu.org/bugs/?65909