| Summary: | Rust SDK Test ( RustCompileTest ) fails | ||
|---|---|---|---|
| Product: | [Build System, Metadata & Runtime] OE-Core | Reporter: | Randy MacLeod <randy.macleod> |
| Component: | devtools / tool chain | Assignee: | Sundeep KOKKONDA <sundeep.kokkonda> |
| Status: | RESOLVED DUPLICATE | QA Contact: | |
| Severity: | major | ||
| Priority: | Undecided | CC: | meta.mr.watcher, meta.watcher, richard.purdie |
| Version: | 4.1 | ||
| Target Milestone: | --- | ||
| Hardware: | x86 | ||
| OS: | Multiple | ||
| Whiteboard: | |||
| OS type for building Yocto: | --- | Type of Regression: | --- |
| Verified: | Documentation change: | No (bug/feature does not impact docs) | |
|
Description
Randy MacLeod
2022-08-09 13:33:56 UTC
Sundeep, this one is more urgent than the MIPS n32 bug. Note that this was with master-next which did have an LD_LIBRARY_PATH tweak in it (I'd posted patches on the mailing list). I suspect this tweak is causing some different issues but without the patch, /bin/sh didn't work for other reasons. This means that effectively we can't have something here with /bin/sh or anything from the host system, we need to only use things within the SDK. I think what we're going to need to do is stop ccld being called with LD_LIBRARY_PATH being set, or put a wrapper in place which filters it out. How we do that without using anything from the host system is going to be tricky. I tried an experiment on one of the failing builds and if we swap /bin/sh in the wrapper for a pointer to sh in the SDK (which is bash), it works correctly. That is probably the easiest fix for this given it is hard to change what rust is passing in the environment. |