Enable booting MIPS and PowerPC kernels#73
Conversation
baa7336 to
b472f9f
Compare
|
I do have RISC-V working already (#72), which I can combine into this pull if it is desirable. Alternatively, we merge this and enable RISC-V in a separate pull. |
nickdesaulniers
left a comment
There was a problem hiding this comment.
Let's keep the riscv changes distinct for now.
| - {<< : *ppc32, << : *mainline, << : *llvm, boot: true, llvm_version: *llvm_tot} | ||
| - {<< : *ppc64, << : *mainline, << : *ppc64_llvm, boot: true, llvm_version: *llvm_tot} | ||
| - {<< : *ppc64le, << : *mainline, << : *llvm, boot: true, llvm_version: *llvm_tot} | ||
| - {<< : *riscv, << : *mainline, << : *riscv_llvm_full, boot: true, llvm_version: *llvm_tot} |
There was a problem hiding this comment.
will this depend on the riscv qemu changes?
There was a problem hiding this comment.
Yes. See my other comment for why it might be advantageous to just enable booting RISC-V with this pull.
| - &ppc64le {config: powernv_defconfig, kernel_image: zImage.epapr, << : *powerpc64le-triple, << : *kernel_modules} | ||
| # https://github.com/ClangBuiltLinux/linux/issues/1143 | ||
| - &riscv {config: [defconfig, CONFIG_EFI=n], << : *riscv-triple, << : *kernel_modules} | ||
| - &riscv {config: [defconfig, CONFIG_EFI=n], kernel_image: Image, << : *riscv-triple, << : *kernel_modules} |
There was a problem hiding this comment.
No. The modules are a separate target and we always build those.
| - {<< : *ppc64le, << : *next, << : *llvm, boot: true, llvm_version: *llvm_latest} | ||
| - {<< : *riscv, << : *next, << : *riscv_llvm_full, boot: true, llvm_version: *llvm_latest} | ||
| # riscv: Boot disabled (https://github.com/ClangBuiltLinux/continuous-integration2/issues/72) | ||
| - {<< : *riscv, << : *next, << : *riscv_llvm_full, boot: false, llvm_version: *llvm_latest} |
There was a problem hiding this comment.
maybe fold this commit into the one that enabled it? (Should be straightforward via interactive rebase)
There was a problem hiding this comment.
The reason I did a separate commit was so that it was easy to revert when we wanted to re-enable RISC-V. Now that qemu-binaries and boot-utils have had their pulls merged, I could just update check_logs.py to have --use-cbl-qemu like I suggested in #72 and avoid disabling the builds altogether.
There was a problem hiding this comment.
sure, so one more commit on top then?
There was a problem hiding this comment.
Yes, basically :P
Please see https://github.com/nathanchance/continuous-integration2/commits/image-targets-with-riscv for what it would look like. If you think that is okay, I will push it under this branch so that the pull updates.
| "x86_64": ["qemu-system-x86"], | ||
| "s390": [], | ||
| "riscv": ["qemu-system-riscv64"], | ||
| "riscv": ["opensbi", "qemu-system-riscv64"], |
There was a problem hiding this comment.
this should wait until we enable riscv, separately.
|
I found it much easier to review multiple smaller commits in github's UI; thank for you for breaking up commits into smaller concerns. |
Good! I will continue to do this in the future. |
malta_kvm_guest_defconfig is a little endian configuration. If we want it to be big endian, we need to pass CONFIG_CPU_BIG_ENDIAN, which we can then use to set the right CBL architecture. Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Link: https://gitlab.com/Linaro/tuxsuite/-/merge_requests/95 Signed-off-by: Nathan Chancellor <nathan@kernel.org>
…name Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
We are not using Ubuntu's QEMU right now so there is not much point to installing it. Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
i386 is still broken with clang-12 and newer. Certain powerpc builds are still broken with tuxmake and older versions of clang. Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Certain powerpc builds are still broken with tuxmake and older versions of clang. Signed-off-by: Nathan Chancellor <nathan@kernel.org>
i386 is still broken with clang-12 and newer. Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
b472f9f to
df1c60a
Compare
|
Sigh that is not what I meant to do... |
nickdesaulniers
left a comment
There was a problem hiding this comment.
nice, this gets us back to parity with all arches now, right? 🧇 🍟
|
Yes, all of the architectures should boot if they can build :) |
|
Thanks for the review! |
I will rebase this on #71 after it is merged but I wanted to get ahead of this for review.
TuxSuite has merged support for
kernel_imageso we can add that togenerator.ymlfor architectures that need it.Closes: #1
Closes: #27