Message ID | 20240708173522.2218293-1-andrew.cooper3@citrix.com (mailing list archive) |
---|---|
Headers | show |
Series | CI: part 3 (improvments to PPC) | expand |
On 08.07.2024 19:35, Andrew Cooper wrote: > Here is a run with these improvements in place: > > https://gitlab.com/xen-project/people/andyhhp/xen/-/jobs/7277624801 > > Jan: I hope to backport this series to 4.18 so we can remove > qemu-system-ppc64/8.1.0-ppc64 entirely. This will simplify changes in the > future by making the testing more consistent. This being part 3 a relevant question would be in how far this is independent of the earlier two parts? > Andrew Cooper (4): > CI: Fix CONTAINER_UID0=1 scripts/containerize > CI: Introduce a bookworm-ppc64le container > CI: Use bookworm-ppc64le for both build and test > CI: Refresh bullseye-ppc64le The other question, seeing that the latter three patches being strictly PPC only, would concern only patch 1: It not having a Fixes: tag and things (supposedly) working okay right now makes me wonder whether that would need to be part of the backport set. Jan
On 09/07/2024 7:31 am, Jan Beulich wrote: > On 08.07.2024 19:35, Andrew Cooper wrote: >> Here is a run with these improvements in place: >> >> https://gitlab.com/xen-project/people/andyhhp/xen/-/jobs/7277624801 >> >> Jan: I hope to backport this series to 4.18 so we can remove >> qemu-system-ppc64/8.1.0-ppc64 entirely. This will simplify changes in the >> future by making the testing more consistent. > This being part 3 a relevant question would be in how far this is independent > of the earlier two parts? Patch 1 is only tangentially related to 2-4. It's a long-standing script bug that I've encountered for the second time now, and will compound everywhere (even older branches) by making the containers be non-root by default. We should never had had root-based containers. It's not necessary for our build or qemu smoke test containers, and it's a stupid risk given that patchew consumes *everything* from the mailing list which applies. In terms of a fixes tag, its 17fbe6504dfd from 6 years ago. It's a problem for anyone using the containers manually with the containerize[sic] script, not for Gitlab itself. ~Andrew