| Summary: | unfs-qemu x86 failed to boot | ||||||
|---|---|---|---|---|---|---|---|
| Product: | [Build System, Metadata & Runtime] Meta-yocto | Reporter: | liping Ke <liping.ke> | ||||
| Component: | meta-yocto | Assignee: | Bruce Ashfield <bruce.ashfield> | ||||
| Status: | RESOLVED FIXED | QA Contact: | |||||
| Severity: | normal | ||||||
| Priority: | Medium | CC: | bruce.ashfield, edwin.zhai, jessica.zhang, lianhao.lu, poky.bs.watcher, poky.watcher, scott.a.garman, sgw | ||||
| Version: | unspecified | ||||||
| Target Milestone: | 1.0 M3 | ||||||
| Hardware: | x86 | ||||||
| OS: | Multiple | ||||||
| Whiteboard: | |||||||
| OS type for building Yocto: | --- | Type of Regression: | --- | ||||
| Verified: | Documentation change: | --- | |||||
| Attachments: |
|
||||||
|
Description
liping Ke
2010-12-28 21:36:02 UTC
Created attachment 86 [details]
screen shot of booting failure for x86 unfs qemu
Some update for the bug: I asked Edwin's help to identify the problem since from script side, the cmdline passed into kernel seems to be correct and portmap should be working fine. and also, non-unfs mode qemu bootup normally. Edwin doubt it might be the kernel problem such whether nfs driver is there or working properly. Since we meet the problem of serial, so hard to debug right now. So the bug might need further debugging efforts. Thanks & Regards, criping Confirmed that NFS drivers are there, so it should not be the kernel driver problem... I will continue some easy investigation, such as using some spare machine resource, revert to 4.0 release workable version, and then judge it's root fs problem and kernel problem, but it needs time to build. I will update the findings when got result. Scott, you can go ahead with the bug since you're much more familiar with it than I:) criping Today I build 4.0 release tree and use that old bzImage-qemux86.bin. For the rootfs side, I used the latest rootfs we extracted (top of master), it works and qemu could be started. So I can conclude the problem is caused by the latest bzImage-qemux86.bin, it should be kernel side problem. Thanks& Regards, criping Are we seeing this on all the qemu boards ? Hi, You mean different archs? I tried i586 and arm, both have the same issues. Thanks a lot! criping aha. So I most likely know what this is. Who can fill me in on the details of the usermode NFS implementation details ? Is it using alternate RPC ports ? There are a series of patches that I've been carrying for years that deal with this. They haven't been ported to 2.6.37 yet, so without them this definitely won't work. Hi Scott, Can you help answer Bruce's question regarding unfs? thx Bruce, We're using the same codebase for the userspace NFS server that WindRiver has been using. It does use alternate RPC ports, so we definitely need those patches in 2.6.37 and all future Yocto kernels for this to work. Given this, I'm reassigning the bug to you so you can close it when the patches have been ported. I know these patches far to well, they aren't always trivial to port and probably won't be here either. So is the command line for the test boot: poky-qemu qemux86 bzImageXXX.bin unfs_folder Or something else ? Bruce, Start by extracting the filesystem using the poky-extract-sdk script: poky-extract-sdk <poky-sdk-tarball> <extract-dir> (this ensures the tarball is extracted using pseudo) then you can boot the filesystem using: poky-qemu qemux86 bzImageXXX.bin unfs_folder |