After uploading a lot of packages a good number of issues occurred in autopkgtests which were not related to RISC-V. Most could be resolved by retries and migration/0 references.
Two GCC issues have been resolved by upstream.
WebKit has issues with included third party packages:
Skia is trying to force tail calls where they are not support. I suggested a patch for this to the Skia project.
The simde version used by WebKit has an issue with correctly setting up vector support for RISC-V but it can be switched off via SIMDE_RISCV_V_NO_NATIVE.
Evaluated and rejected the possibility of switching to strict confinement
Started debugging snapcraft linter warning around incorrect ELF interpreter. Standard patchelf invocation doesn’t behave as expected, and so I am finding an alternative way to set the interpreter in a platform-neutral way.
Other
Helped verify some arm64 cross-compilation workflows
Interacted with the security team about the rustc 1.83 Noble backport – I may need to open an SRU after all because it is currently unclear whether QEMU updates are actually blocking on rustc.
Worked on rustc 1.83 Jammy backport. It compiles on my PPA but does not pass the tests because it looks like I have not vendored LLVM properly. Currently waiting on the PPA to build my fixed version so I can run the autopkgtests
I’ve previously written a replacement for mk-origtargz which runs ~20x faster; this week I worked on a presentation about it that I’ll distribute internally on Friday.