On Wed, 15 Oct 2025 at 03:37, Reinhard Tartler <[email protected]> wrote: > full buildlog here: > https://buildd.debian.org/status/fetch.php?pkg=docker.io&arch=mips64el&ver=27.5.1%2Bdfsg1-1&stamp=1760502779&raw=0 > > seems another buildkit on mips issue: > > # github.com/moby/buildkit/snapshot > src/github.com/moby/buildkit/snapshot/diffapply_linux.go:128:8: cannot use > stat.Dev (variable of type uint32) as uint64 value in struct literal > src/github.com/moby/buildkit/snapshot/diffapply_linux.go:297:12: cannot use > unix.Mkdev(0, 0) (value of type uint64) as uint32 value in struct literal > > Tianon, what's your take on this? What's the best way to fix this these days?
https://github.com/moby/buildkit/issues/5129 strikes again 😅😂😠On the upside, I've been running with that patch on my (single) mips64le system for ages now and haven't noticed a problem (this is the same system that all of the mips64le variants of https://hub.docker.com/u/library are built on, so if there's a mips64le + Docker user out there who *might* notice, I would hope they would've by now). I also apply that patch on all my personal builds of the Docker engine across all architectures I use/support, and it seems to be fine. That being said, I think mips64le + Docker usage is now low to zero (although I don't have explicit stats to back that up). https://www.debian.org/releases/trixie/release-notes/issues.en.html#mips-architectures-removed agrees though, and even explicitly recommends users of this hardware switch to different hardware, so this is effectively an unstable-only architecture now and I'd personally not lose much sleep if we decided to (request FTP masters) drop mips64le from this package instead of applying the patch, but the patch is pretty simple so this could reasonably go either way. ♥, - Tianon 4096R / B42F 6819 007F 00F8 8E36 4FD4 036A 9C25 BF35 7DD4
