-
Notifications
You must be signed in to change notification settings - Fork 18
Drop ext4 images and use initramfs for all targets #221
Conversation
I pulled this down locally and started a set of builds before I had to rush off to work. I’ll post the results and formally review it tonight. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I was hesitant to remove these, as locally I was using these images to run QEMU tests quickly locally, but I should just update my workflow to use these. Thanks for the cleanup and simplification. 🐇 🏇
Full results:
Looks like powerpc64le is the only regression but it appears to be a QEMU one because I am using v4.2.0-rc4 and v4.1.1 appears fine:
I'll bisect the breaking commit and report it upstream. I assume that our version of QEMU in the Docker image should be fine but I do not have the time to verify right now. |
There is currently an issue with booting a ppc64le kernel with QEMU 4.2.0-rc4. A bisect points to the QEMU commit linked below. Updating to the latest stable skiboot 6.5.1 fixes the issue and does not break QEMU 3.1.0 that we currently use for our Docker image. Link: ClangBuiltLinux#221 (comment) Link: https://git.qemu.org/?p=qemu.git;a=commit;h=f30c843ced5055fde71d28d10beb15af97fdfe39 Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
There is currently an issue with booting a ppc64le kernel with QEMU 4.2.0-rc4 with the skiboot we package in this repo (done in ClangBuiltLinux#138), see the link in ClangBuiltLinux#221 below. A bisect points to the QEMU commit linked below. Updating to the latest stable skiboot 6.5.1 fixes the issue and does not break QEMU 3.1.0 that we currently use for our Docker image. [skip ci] Link: ClangBuiltLinux#221 (comment) Link: https://git.qemu.org/?p=qemu.git;a=commit;h=f30c843ced5055fde71d28d10beb15af97fdfe39 Presubmit: https://travis-ci.com/nathanchance/continuous-integration/builds/139635572 Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
I think this is good to land, now? Or are we still blocked on something? |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Sorry, forgot to update it here. That regression was not in this pull request, it has been handled in #224. This should be good now.
As discussed in #184 (comment), ext4 images can be dropped and slimmer initramfs images should be used instead.
Most ext4 images were unused. Only arm32_v7, arm64, mipsel and some x86_64 configs used them. They were modified to use initramfs instead.
Travis for basic set of configurations shows no regressions, but I suggest testing the full set that's normally run by cron.