Bug 531

Summary: 'rm_work' is run much later than expected, when INHERIT'ed
Product: [Build System, Metadata & Runtime] OE-Core Reporter: Tom Rini <tom.rini>
Component: coreAssignee: Dongxiao Xu <dongxiao.xu>
Status: CLOSED FIXED QA Contact:
Severity: normal    
Priority: Medium CC: ke.yu, kevin.tian, meta.mr.watcher, meta.watcher, richard.purdie, sgw
Version: 0.9   
Target Milestone: 1.0 M1   
Hardware: x86   
OS: Multiple   
Whiteboard:
OS type for building Yocto: --- Type of Regression: ---
Verified: Documentation change: No (bug/feature does not impact docs)

Description Tom Rini 2010-11-10 13:52:35 UTC 
When I've used the rm_work class in OpenEmbedded, it will run the rm_work task "in the middle" of the build.  Which is to say, I can count on rm_work cleaning up more room for me to build on while I'm building and not run out of disk space in limited space environments.  With poky 0.9 it seems that rm_work isn't started for anything until all other tasks are run for all other recipes.  This means that I only reclaim space at the end of the build and have run out of disk in the middle of a build.
Comment 1 Richard Purdie 2010-11-11 02:47:38 UTC 
The problem appears to be that OE uses the completion scheduler in rm_work.bbclass and Poky doesn't have this. We should add those lines of code (I think I wrote the OE version!).
Comment 3 Tom Rini 2014-07-22 15:44:39 UTC 
Close this out...