Closed martell closed 5 years ago
Nice! I wanted to update buildroot, but was to lazy to build all images with it. RISC-V support is a good reason for an update. Hopefully, there are no regressions.
This script is not run as part of our CI. The rebuild.sh
script is run locally, which I can do later if you want. To test for regressions, use the build-all.sh
script locally if you don't want to wait for Travis.
I am going to rebuild and test the images locally now, I'll either report back later tonight or first thing tomorrow morning.
interesting, seems there's also some kind of 2019.05.01 builds, too. Wonder how they do their versioning...
Tentatively, the ARMv5 and ARMv6 builds regressed with this update: https://gist.github.com/088f6057f3bc032acdfad530ef7990eb
I am doing a full before/after right now though.
Before rebuild:
Running 'ARCH=arm32_v5 ./driver.sh'... Successful in 0m 57s
Running 'ARCH=arm32_v6 ./driver.sh'... Successful in 1m 6s
Running 'ARCH=arm32_v7 LD=ld.lld ./driver.sh'... Successful in 0m 55s
Running 'ARCH=arm64 LD=ld.lld ./driver.sh'... Successful in 1m 18s
Running 'ARCH=ppc32 ./driver.sh'... Successful in 0m 37s
Running 'ARCH=ppc64le ./driver.sh'... Successful in 1m 2s
Running 'ARCH=x86_64 LD=ld.lld ./driver.sh'... Successful in 0m 46s
Running 'ARCH=arm32_v5 REPO=linux-next ./driver.sh'... Successful in 0m 55s
Running 'ARCH=arm32_v6 REPO=linux-next ./driver.sh'... Successful in 0m 55s
Running 'ARCH=arm32_v7 LD=ld.lld REPO=linux-next ./driver.sh'... Successful in 0m 55s
Running 'ARCH=arm64 LD=ld.lld REPO=linux-next ./driver.sh'... Successful in 1m 20s
Running 'ARCH=ppc32 REPO=linux-next ./driver.sh'... Failed in 2m 20s
Running 'ARCH=ppc64le REPO=linux-next ./driver.sh'... Successful in 1m 1s
Running 'ARCH=x86_64 LD=ld.lld REPO=linux-next ./driver.sh'... Successful in 0m 46s
Running 'ARCH=arm32_v7 REPO=4.19 ./driver.sh'... Successful in 1m 26s
Running 'ARCH=arm64 REPO=4.19 ./driver.sh'... Successful in 1m 33s
Running 'ARCH=ppc64le REPO=4.19 ./driver.sh'... Successful in 1m 0s
Running 'ARCH=x86_64 LD=ld.lld REPO=4.19 ./driver.sh'... Successful in 0m 45s
Running 'ARCH=arm32_v7 REPO=4.14 ./driver.sh'... Successful in 1m 19s
Running 'ARCH=arm64 REPO=4.14 ./driver.sh'... Successful in 1m 11s
Running 'ARCH=ppc64le REPO=4.14 ./driver.sh'... Successful in 1m 46s
Running 'ARCH=x86_64 LD=ld.lld REPO=4.14 ./driver.sh'... Successful in 1m 47s
Running 'ARCH=arm32_v7 REPO=4.9 ./driver.sh'... Successful in 2m 5s
Running 'ARCH=arm64 REPO=4.9 ./driver.sh'... Successful in 1m 46s
Running 'ARCH=x86_64 LD=ld.lld REPO=4.9 ./driver.sh'... Successful in 1m 24s
Running 'ARCH=arm64 REPO=4.4 ./driver.sh'... Successful in 1m 46s
Running 'ARCH=x86_64 LD=ld.lld REPO=4.4 ./driver.sh'... Successful in 1m 18s
Running 'ARCH=arm64 REPO=common-4.9 ./driver.sh'... Successful in 2m 8s
Running 'ARCH=arm64 REPO=common-4.14 ./driver.sh'... Successful in 3m 2s
Running 'ARCH=arm64 REPO=common-4.19 ./driver.sh'... Successful in 3m 11s
Running 'ARCH=x86_64 LD=ld.lld REPO=common-4.9 ./driver.sh'... Successful in 1m 59s
Running 'ARCH=x86_64 LD=ld.lld REPO=common-4.14 ./driver.sh'... Successful in 2m 50s
Running 'ARCH=x86_64 LD=ld.lld REPO=common-4.19 ./driver.sh'... Successful in 3m 8s
After rebuild:
Running 'ARCH=arm32_v5 ./driver.sh'... Failed in 2m 55s
Running 'ARCH=arm32_v6 ./driver.sh'... Failed in 2m 29s
Running 'ARCH=arm32_v7 LD=ld.lld ./driver.sh'... Successful in 0m 56s
Running 'ARCH=arm64 LD=ld.lld ./driver.sh'... Successful in 1m 19s
Running 'ARCH=ppc32 ./driver.sh'... Successful in 0m 37s
Running 'ARCH=ppc64le ./driver.sh'... Successful in 1m 3s
Running 'ARCH=x86_64 LD=ld.lld ./driver.sh'... Successful in 0m 47s
Running 'ARCH=arm32_v5 REPO=linux-next ./driver.sh'... Failed in 2m 34s
Running 'ARCH=arm32_v6 REPO=linux-next ./driver.sh'... Failed in 2m 29s
Running 'ARCH=arm32_v7 LD=ld.lld REPO=linux-next ./driver.sh'... Successful in 0m 55s
Running 'ARCH=arm64 LD=ld.lld REPO=linux-next ./driver.sh'... Successful in 1m 20s
Running 'ARCH=ppc32 REPO=linux-next ./driver.sh'... Failed in 2m 20s
Running 'ARCH=ppc64le REPO=linux-next ./driver.sh'... Successful in 1m 1s
Running 'ARCH=x86_64 LD=ld.lld REPO=linux-next ./driver.sh'... Successful in 0m 46s
Running 'ARCH=arm32_v7 REPO=4.19 ./driver.sh'... Successful in 1m 24s
Running 'ARCH=arm64 REPO=4.19 ./driver.sh'... Successful in 1m 32s
Running 'ARCH=ppc64le REPO=4.19 ./driver.sh'... Successful in 0m 59s
Running 'ARCH=x86_64 LD=ld.lld REPO=4.19 ./driver.sh'... Successful in 0m 45s
Running 'ARCH=arm32_v7 REPO=4.14 ./driver.sh'... Successful in 1m 17s
Running 'ARCH=arm64 REPO=4.14 ./driver.sh'... Successful in 1m 12s
Running 'ARCH=ppc64le REPO=4.14 ./driver.sh'... Successful in 0m 57s
Running 'ARCH=x86_64 LD=ld.lld REPO=4.14 ./driver.sh'... Successful in 0m 48s
Running 'ARCH=arm32_v7 REPO=4.9 ./driver.sh'... Successful in 1m 6s
Running 'ARCH=arm64 REPO=4.9 ./driver.sh'... Successful in 0m 47s
Running 'ARCH=x86_64 LD=ld.lld REPO=4.9 ./driver.sh'... Successful in 0m 29s
Running 'ARCH=arm64 REPO=4.4 ./driver.sh'... Successful in 1m 6s
Running 'ARCH=x86_64 LD=ld.lld REPO=4.4 ./driver.sh'... Successful in 0m 28s
Running 'ARCH=arm64 REPO=common-4.9 ./driver.sh'... Successful in 1m 13s
Running 'ARCH=arm64 REPO=common-4.14 ./driver.sh'... Successful in 2m 2s
Running 'ARCH=arm64 REPO=common-4.19 ./driver.sh'... Successful in 2m 4s
Running 'ARCH=x86_64 LD=ld.lld REPO=common-4.9 ./driver.sh'... Successful in 0m 48s
Running 'ARCH=x86_64 LD=ld.lld REPO=common-4.14 ./driver.sh'... Successful in 1m 40s
Running 'ARCH=x86_64 LD=ld.lld REPO=common-4.19 ./driver.sh'... Successful in 1m 29s
Just the ARMv5 and ARMv6 builds regressed (I am bisecting the ppc32 failure on -next, it is new as of next-20190708
).
I'll test the latest Buildroot stable release as well as the current master LTS branch to see if this will be fixed then.
Sigh, this is a problem that has existed since #101 because of https://github.com/ClangBuiltLinux/continuous-integration/pull/101#issuecomment-447205229 :(
diff --git a/buildroot/arm.config b/buildroot/arm.config
index b1d2a22..be6ee29 100644
--- a/buildroot/arm.config
+++ b/buildroot/arm.config
@@ -5,7 +5,7 @@ BR2_BINUTILS_VERSION_2_31_X=y
BR2_GCC_VERSION_8_X=y
BR2_TARGET_GENERIC_ROOT_PASSWD="root"
BR2_TARGET_GENERIC_GETTY_PORT="ttyAMA0"
-BR2_ROOTFS_OVERLAY="../overlay-poweroff"
+BR2_ROOTFS_OVERLAY="../overlay-reboot"
BR2_PACKAGE_BUSYBOX_SHOW_OTHERS=y
BR2_PACKAGE_BASH=y
BR2_TARGET_ROOTFS_CPIO=y
This resolves it. I'll rebuild all of the images, do another set of tests, then push it so the rebuild can be merged as part of this.
https://github.com/nathanchance/continuous-integration/commits/update-buildroot
Everything works fine now.
@martell let me know if you want me to push my branch as a new pull or if you want to push it here.
Once that is done, I think this should be good to go if nobody objects.
@nathanchance thanks for the additional fixes. Just issue a new PR carrying @martell 's commit plus yours on top, and close this on in favor of that.
Done: #189
RISCV needs a newer buildroot #70 This is probably a good chance to put this change in it's own PR and ensure there are no ci regressions. I kept it in line continuing to use the LTS version. See how the tests go. :)