Bug 2395

Summary: Bitbake spins
Product: [Build System, Metadata & Runtime] BitBake Reporter: Rich Pixley <rich.pixley>
Component: bitbakeAssignee: Richard Purdie <richard.purdie>
Status: RESOLVED FIXED QA Contact:
Severity: normal    
Priority: Medium CC: meta.layer.watcher, meta.watcher, poky.bs.watcher, poky.watcher
Version: unspecified   
Target Milestone: 1.3   
Hardware: x86   
OS: Multiple   
Whiteboard:
OS type for building Yocto: --- Type of Regression: ---
Verified: Documentation change: ---

Description Rich Pixley 2012-05-02 20:54:23 UTC 
Sometimes, but consistently repeatable when it occurs, attempting to build core-image-minimal appears to hang.  "Top" shows a single processor pinned at 98 - 99% utilization running "python".

The spin appears to occur on "Executing SetScene tasks".  When it works, this usually runs and completes in seconds for me.  When it hangs, it hangs for at least an hour, (I haven't let it run any longer than that.)

I'm not sure where else to look or how to go about debugging this.

Here's the relevant tail of a build log...

NOTE: Executing SetScene Tasks
DEBUG: Stamp for underlying task 12(virtual:native:/home/rich/projects/webos/openembedded-core/meta/recipes-devtools/opkg/opkg_svn.bb, do_populate_sysroot) is current, so skipping setscene variant
DEBUG: Stamp for underlying task 16(virtual:native:/home/rich/projects/webos/openembedded-core/meta/recipes-devtools/opkg-utils/opkg-utils_git.bb, do_populate_sysroot) is current, so skipping setscene variant
DEBUG: Stamp for underlying task 20(virtual:native:/home/rich/projects/webos/openembedded-core/meta/recipes-devtools/makedevs/makedevs_1.0.0.bb, do_populate_sysroot) is current, so skipping setscene variant
DEBUG: Stamp for underlying task 24(/home/rich/projects/webos/openembedded-core/meta/recipes-core/eglibc/ldconfig-native_2.12.1.bb, do_populate_sysroot) is current, so skipping setscene variant
DEBUG: Stamp for underlying task 32(virtual:native:/home/rich/projects/webos/openembedded-core/meta/recipes-devtools/genext2fs/genext2fs_1.4.1.bb, do_populate_sysroot) is current, so skipping setscene variant
DEBUG: Stamp for underlying task 36(virtual:native:/home/rich/projects/webos/openembedded-core/meta/recipes-devtools/e2fsprogs/e2fsprogs_1.42.1.bb, do_populate_sysroot) is current, so skipping setscene variant
DEBUG: Stamp for underlying task 40(virtual:native:/home/rich/projects/webos/openembedded-core/meta/recipes-devtools/qemu/qemu_0.15.1.bb, do_populate_sysroot) is current, so skipping setscene variant
DEBUG: Stamp for underlying task 44(/home/rich/projects/webos/openembedded-core/meta/recipes-devtools/qemu/qemu-helper-native_1.0.bb, do_populate_sysroot) is current, so skipping setscene variant
Comment 1 Saul Wold 2012-05-03 19:43:01 UTC 
Can you send more infomration about which OS you are using, also have you tried to use strace?  Have you added any layers or recipes?  What target/image are you building?

If you try strace, please use the following: 

strace -s 8192 -f -o bb.strace bitbake ...

Then ^C it when it hangs and attach the bb.strace file
Comment 2 Rich Pixley 2012-05-03 20:25:21 UTC 
Ubuntu-12.04 amd64 server.

I have a git repository that includes a configure script, makefiles, and git submodules pointing to the git repositories for bitbake, openembedded-core, and meta-webos.

My first build in a working directory works.  My second one hangs with the spinning problem.

Image seems to be irrelevant, but I've seen the problem on all targets I've tried, including core-image-minimal, on each of qemux86, qemuarm, qemumips, and qemuppc.  (I haven't tried any other targets.)

I'm looking at strace now, but I haven't yet identified any loop.

How much strace would you like?  :).
Comment 3 Saul Wold 2012-05-03 22:47:20 UTC 
Maybe I will try to reproduce it here knowing which layers you are using, so there was no obvious blocker in the strace file, oh well it was a thought.
Comment 4 Rich Pixley 2012-05-03 22:54:48 UTC 
No obvious blocker.  It's spinning, not blocked.  And the presumed loop is over 1e6 strace lines long.

What I do see are tons of access and stat's and periodic mkdir's on stamp files and stamp file directories.  Thousands of quilt-native checks, for instance.  And all of the do_fetch_setscene checks are "no such file or directory".  I dunno if those are clues, but that's what I see.
Comment 5 Saul Wold 2012-05-04 00:44:40 UTC 
Is our git repo and the webos layer that you have available?  I could not find a quick pointer to it. You could add it to: http://www.openembedded.org/wiki/LayerIndex
Comment 6 Rich Pixley 2012-05-04 00:46:49 UTC 
Sadly, no, not yet.  The goal is for it to become available, but that may take some time.