Backport main-next non-clangable#1003
Conversation
Signed-off-by: Jose Quaresma <[email protected]>
Signed-off-by: Jose Quaresma <[email protected]>
|
should unblock foundriesio/lmp-manifest#277 |
| OBJCOPY:pn-linux-lmp-ti-staging:toolchain-clang = "${HOST_PREFIX}objcopy" | ||
| OBJCOPY:pn-linux-lmp-toradex-imx:toolchain-clang = "${HOST_PREFIX}objcopy" | ||
| OBJCOPY:pn-linux-lmp-xlnx:toolchain-clang = "${HOST_PREFIX}objcopy" | ||
| OBJCOPY:pn-linux-tegra:toolchain-clang = "${HOST_PREFIX}objcopy" |
There was a problem hiding this comment.
Do you know why this started to happen?
There was a problem hiding this comment.
No, don't know where it is been introduced but can find it locally.
On main next I think see something about this on clang layer but on this stable update this makes less sense.
Anyway it fixes the build foundriesio/lmp-manifest#284 that almost done.
There was a problem hiding this comment.
Sure, I'm ok merging this, I would just like to have a better understanding so we know when to revert later on.
There was a problem hiding this comment.
Right, I don't even remember what happened just to change from one branch to another.
I will find something about this clang issues to update the commit message.
There was a problem hiding this comment.
This seems that was added in meta-clang as well:
+# Subprocess output:mips-yoe-linux-llvm-objcopy: error: Link field value 22 in section .rel.dyn is not a symbol table
+# also seen on riscv64 and x86-64
+OBJCOPY:pn-linux-yocto:toolchain-clang = "${HOST_PREFIX}objcopy"
There was a problem hiding this comment.
@quaresmajose with these commits we have an error like:
Command '['aarch64-lmp-linux-llvm-strip', '--remove-section=.comment', '--remove-section=.note', '.../boot/vmlinux-5.15.81-lmp-standard']' returned non-zero exit status 1.
Subprocess output:aarch64-lmp-linux-llvm-strip: error: Link field value 30 in section .rela.dyn is not a symbol table
I fixed it with the commit MrCry0@2b982d3
Testing.
But it's better to find the root cause of all these issues.
|
|
So yeah, please add the objcopy and strip changes for all kernel recipes we use (since kernel is indeed not yet using clang during build-time). |
|
We have now most of the layers updated with foundriesio/lmp-manifest#287 so my next task will be update oe-core layer and find where was introduced this issues that mainly affect the clang toolchain. |
@quaresmajose Ricardo found that meta-clang has a similar fix for the Linux kernel, so our fix may be acceptable: |
|
@MrCry0 we have already meta-clang updated with latest kirkstone version but this clang issues comes when we bump oe-core so the root cause is in oe-core and not in meta-clang imo |
|
This are been integrated in foundriesio/lmp-manifest#287 |
No description provided.