| Summary: | replace unionfs with aufs in live image | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Product: | [Build System, Metadata & Runtime] OE-Core | Reporter: | Chen Qi <Qi.Chen> | ||||||
| Component: | core | Assignee: | Chen Qi <Qi.Chen> | ||||||
| Status: | RESOLVED FIXED | QA Contact: | |||||||
| Severity: | enhancement | ||||||||
| Priority: | Medium | CC: | bluelightning, bruce.ashfield, jessica.zhang, liezhi.yang, mark.hatle, meta.mr.watcher, meta.watcher, richard.purdie, sgw, yi.zhao | ||||||
| Version: | 1.4 | ||||||||
| Target Milestone: | 1.5 M2 | ||||||||
| Hardware: | x86 | ||||||||
| OS: | Multiple | ||||||||
| Whiteboard: | |||||||||
| OS type for building Yocto: | --- | Type of Regression: | --- | ||||||
| Verified: | Documentation change: | Don't know | |||||||
| Bug Depends on: | 1487 | ||||||||
| Bug Blocks: | |||||||||
| Attachments: |
|
||||||||
|
Description
Chen Qi
2013-06-21 10:38:39 UTC
I looked at this late in the 1.4 cycle. I'm no expert on these matters but it seemed to me based on what I've read that aufs is in no better position as regards making it into the mainline kernel than unionfs, which would make me question why we would switch to it from unionfs. It looked like overlayfs might be coming closer to being merged but AFAICT discussion on that seems to have died out. Add Bruce, since I think he had some input in the original discussion aufs is still a valid choice. I have a sample recipe for aufs-utils that I can contribute. It has been updated in the past month, I have it in both linux-yocto-3.8 and linux-yocto-dev. overlayfs has been "knocking at the door" for years now, and is also a valid option, but hasn't been updated since 3.4. unionfs is still dormant and hasn't been updated for even longer, consider it completely dead. So the choice is still pretty clear in the short term. Go with aufs, while we wait for the overlayfs revival. if/when you need aufs-utils, send me an email and I'll track down my recipe. (In reply to comment #4) > if/when you need aufs-utils, send me an email and I'll track down my recipe. Thanks. I added the following line to my local.conf and tested my patch with qemux86. KERNEL_FEATURES_append = " features/aufs/aufs-enable.scc" Everything worked out fine so far. I'll do more testing before I sent out the patch. Best Regards, Chen Qi Created attachment 1290 [details]
aufs-utils
in case you want it, I attached my aufs-utils recipe. It isn't perfect, but did build and work for x86 at the time. Created attachment 1292 [details]
aufs-util patch for cross compilation
Create attachment for the patch.
(In reply to comment #7) > in case you want it, I attached my aufs-utils recipe. It isn't perfect, but > did build and work for x86 at the time. Thanks. I just tried it out and made a patch to fix a cross compilation error. The problem here is that, ver, c2sh and c2tmac are run on host, so they need to be compiled first. I'm not familiar with aufs, so I'm not sure if I'm doing right. Please take a look at the patch and let me know if I made some mistakes. //Chen Qi On a quick scan, the native extensions look ok to me. I was only running on the target when I tested, so I didn't run into the need. I'm not 100% sure of the configure modification though, I thought you could do it via: do_configure_append_class-native() merged into poky/master: 29e55997caeb9c6367f9a340e153f829228f36ed http://git.openembedded.org/?p=openembedded-core.git&a=commit;h=ba5e437bc7335468a70ea293496f78e1a9d66287 http://git.openembedded.org/?p=openembedded-core.git&a=commit;h=3cd6fedd815688b2f3fd97a56feb5f8696ebeace FYI, I've turned the attachments into a patch adding an aufs-util recipe to meta-filesystems, tweaked it a little bit and build-tested it, the results are here: http://cgit.openembedded.org/meta-openembedded-contrib/log/?h=paule/aufs-util I haven't tested it at runtime though. |