| Summary: | Tracing: create separate recipe for perf | ||
|---|---|---|---|
| Product: | [Build System, Metadata & Runtime] Meta-yocto | Reporter: | Richard Purdie <richard.purdie> |
| Component: | meta-yocto | Assignee: | Bruce Ashfield <bruce.ashfield> |
| Status: | VERIFIED FIXED | QA Contact: | |
| Severity: | enhancement | ||
| Priority: | Low | CC: | dvhart, jessica.zhang, josh, laurentiu.serban, poky.bs.watcher, poky.watcher, radu.moisan, sgw, song.liu |
| Version: | unspecified | ||
| Target Milestone: | 1.3 M2 | ||
| Hardware: | x86 | ||
| OS: | Multiple | ||
| Whiteboard: | (P2)(Scheduled) | ||
| OS type for building Yocto: | --- | Type of Regression: | --- |
| Verified: | Documentation change: | --- | |
|
Description
Richard Purdie
2011-10-03 06:28:39 UTC
Perf is currently built as part of the kernel recipe. Perf really wants to use libnewt, libslang, embedded Python, embedded Perl, etc. That ends up in a ridiculous situation where the kernel depends on e.g. libnewt. It also apparently causes problems for x32 enablement. So, making perf a separate recipe seems to make sense. perf is also causing issues for non-gplv3 builds since I added binutils to the DEPENDS, see bug #1880. I think the best resolution to bug 1880 is to move perf to a separate recipe in the 1.2 time frame. Thoughts? It really just comes down to time. If someone else is available with the cycles to come up with a patch, then we're all for getting it done in time for 1.2. This is nearly done. Patches are prep'd and will be submitted soon. yocto commit: 6b9c19140cd3c75c9862b22febb17af59930d865 Verified in master revision b0510e547ebcac5548b36df66ad907f7911d33d5 |