| Summary: | [PPC] kernel panic when booting poky-image-sdk-qemuppc through UNFS | ||
|---|---|---|---|
| Product: | [Build System, Metadata & Runtime] Meta-yocto | Reporter: | Lianhao Lu <lianhao.lu> |
| Component: | meta-yocto | Assignee: | Liming Wang <liming.wang> |
| Status: | VERIFIED FIXED | QA Contact: | |
| Severity: | major | ||
| Priority: | High | CC: | bruce.ashfield, dvhart, jessica.zhang, julie.n.fleischer, kevin.tian, liang.li2, liming.wang, liping.ke, mark.hatle, poky.bs.watcher, poky.watcher, qing.he, richard.purdie, sgw, song.liu, wenhuax.fan |
| Version: | 0.9 | ||
| Target Milestone: | 1.2 | ||
| Hardware: | x86 | ||
| OS: | ppc | ||
| Whiteboard: | post 1.1, but need work around script for now. | ||
| OS type for building Yocto: | --- | Type of Regression: | --- |
| Verified: | Documentation change: | --- | |
| Bug Depends on: | 1204 | ||
| Bug Blocks: | |||
Lianhao, This is similar to the tap issue that I mentioned in our meeting, can you run ifconfig and make sure all tap device are gone and re-try? I ran into this for ARM this afternoon and when I do ifconfig, I didn't see tap0, tap1, but tap2 and tap3 are there. I used tunctl -d to remove taps, also delete ~/.poky-sdk I'm pretty sure when I launch the qemu, there is no such tap0, tap1, ... Besides, the qemumips & qemux86 is ok. I meet the same problem in ARM architectures... So please fix the bug both in PPC and ARM Lianhao and Liping, Can you clearly specify how you try bring up unfs qemu, from command line, what exactly is your command and options. Also, is it under poky-tree or installed SDK and I'm assuming you've done source the environment setup script before running poky-qemu, right? If under poky-tree, did you do bitbake meta-ide-support? To reproduce the ppc bug: 1. download the ppc kernel and image from autobuilder http://autobuilder.pokylinux.org/nightly/20101007-1/qemu/qemuppc/zImage-2.6.34+git0+71cc215e36b87302af8e619d87d6b76ada5ebae4_0+9ac0daee43dd19d8bea828cf79450c9748ae0daa-r6-qemuppc-20101007051748.bin http://autobuilder.pokylinux.org/nightly/20101007-1/qemu/qemuppc/poky-image-sdk-qemuppc-20101007065703.rootfs.tar.bz2 2. bitbake meta-ide-support 3. poky-extract-sdk poky-image-sdk-qemuppc-20101007065703.rootfs.tar.bz2 <rootfs dir> 4. poky-qemu qemuppc <kernel> <rootfs dir> (In reply to comment #5) > To reproduce the ppc bug: > > 1. download the ppc kernel and image from autobuilder > http://autobuilder.pokylinux.org/nightly/20101007-1/qemu/qemuppc/zImage-2.6.34+git0+71cc215e36b87302af8e619d87d6b76ada5ebae4_0+9ac0daee43dd19d8bea828cf79450c9748ae0daa-r6-qemuppc-20101007051748.bin > > http://autobuilder.pokylinux.org/nightly/20101007-1/qemu/qemuppc/poky-image-sdk-qemuppc-20101007065703.rootfs.tar.bz2 > > 2. bitbake meta-ide-support > > 3. poky-extract-sdk poky-image-sdk-qemuppc-20101007065703.rootfs.tar.bz2 > <rootfs dir> > 4. poky-qemu qemuppc <kernel> <rootfs dir> One step is missing before poky-qemu. User should "source" the environment file built by step 2 (bitbake meta-ide-support) For arm side, I just build the meta-toolchain-sdk by myself, and get the same result... Thanks& Regards, criping I have tested the kernel and unfs rootfs referenced in this bug above and it boots fine for me. My guess is there is still cruft left over from one of the previous poky-qemu implementations I shared with the SDK team. This should be a fail-safe way of starting from a clean slate: 1. Reboot your computer (this will ensure the tap devices are all gone) 2. rm -rf ~/.poky-sdk 3. rm -rf on your extracted nfs rootfs areas Then start from scratch extracting the rootfs using poky-extract-sdk and try booting to it again. Lianhao, one additional step could be necessary for you to start with a clean state. After the reboot, also be sure to delete all files in /tmp/qemu-tap-locks/ (In reply to comment #8) > I have tested the kernel and unfs rootfs referenced in this bug above and it > boots fine for me. My guess is there is still cruft left over from one of the > previous poky-qemu implementations I shared with the SDK team. > > This should be a fail-safe way of starting from a clean slate: > > 1. Reboot your computer (this will ensure the tap devices are all gone) > 2. rm -rf ~/.poky-sdk > 3. rm -rf on your extracted nfs rootfs areas > > Then start from scratch extracting the rootfs using poky-extract-sdk and try > booting to it again. Following the above steps (and that in comment #9), it seems the PPC is OK now. Liping, could you please try the above steps with ARM? Lianhao and Liping, There's still problem that discovered by Scott and me when played more with unfs qemu boot up. Lianhao, if you try to boot up an ARM qemu, they try ppc again, likely you'll run into nfs server not responding again, that's what happened to me, can you play more with shutdown qume and bring up again, etc and see whether it performs well? When I am trying x86 qemu, I continuously meet MOUNTD ERROR when I delete all tmp folder and delete all tapX. After reboot, I finally successfully boot. Not knowing what's the reason... Anyway, it works now. Thanks a lot! criping (In reply to comment #11) > Lianhao and Liping, > > There's still problem that discovered by Scott and me when played more with > unfs qemu boot up. Lianhao, if you try to boot up an ARM qemu, they try ppc > again, likely you'll run into nfs server not responding again, that's what > happened to me, can you play more with shutdown qume and bring up again, etc > and see whether it performs well? I don't have ARM image aside, so I played with PPC and x86. I tried bringing up both ppc & x86 and had them running at the same time, it seemed fine to me. There is NFS server not responding during the shutdown for every qemu architecture. It is caused by the connman(I reported a separate bug #364 several days ago). To workaround the problem without sabotaging tap environment, you may directly kill the qemu process before the bug#364 get fixed. Oops, now I run into the same problem as described in my the very first original description. Here is what I did: 1. after successfully playing with the ppc/x86 nfs booting several times without any problems. 2. have a single x86 qemu running for some time. 3. stop the x86 qemu. 4. rm -rf ~/.poky-sdk and rm -rf /tmp/qemu-tap-locks/ 5. boot the ppc qemu, and run into the problem: Looking up port of RPC 11111/2 on 192.168.7.1 Looking up port of RPC 21111/1 on 192.168.7.1 VFS: Mounted root (nfs filesystem) on device 0:12. Freeing unused kernel memory: 180k init nfs: server 192.168.7.1 not responding, still trying 6. while leaving the problematic ppc qemu hanging, boot another x86 qemu and it boots fine. It seems the PPC UNFS is not very stable. I tried the following steps for 3 times: 1. reboot machine 2. rm -rf ~/.poky-sdk/ 3. rm -rf /tmp/qemu-tap-locks/ 4. boot ppc qemu The first 2 times fails with the "nfs: server 192.168.7.1 not responding, still trying" and the 3 time it succeeds. I googled and found a patch at http://patchwork.ozlabs.org/patch/36846/. I'm not sure if it's related to our problem. I searched the qemu git and found that this patch is not committed. Down grade the bug to minor+low but keep it open till we don't see the problem anymore. I've played a little bit more and seems it behaves much better after rebooting the machine... I rebuilt all using the latest builds/milestone branch, and the problem is still there and easy to reproduce. 2 out of 3 booting would fail and the remaining 1 would succeed. Tcpdump showed that during failed the PPC UNFS booting, the unfs server suddenly stopped sending the udp packet back to the qemu target. Lianhao & Liping Can you try other arches, e.g. arm and mips to see whether similar symptom. If so, I'll bump the priority back to normal+medium, but given the timing, it probably will move to 1.0 M1... we rarely see the problem on arm/x86 now. I'll play more with mips to see if it exists. Lianhao's machine is x86_64. And seems meet more problems recently. And jessica, your machine is 64 or 32? Thanks& Regards, criping I tried the ppc on liping'(In reply to comment #20) > Lianhao's machine is x86_64. And seems meet more problems recently. And > jessica, your machine is 64 or 32? > It seems not related to x86_64 host. I tried PPC on liping's 32 bit host, still met the problem. (In reply to comment #19) > we rarely see the problem on arm/x86 now. I'll play more with mips to see if it > exists. Mips seems fine. The problem seems only reproducible on PPC. Now I met the "nfs: server 192.168.7.1 not responding, still trying" hang error after a successful boot, during the normal OS running. Seem nfs for ppc is really unstable and barely usable... I am going to see if I can apply the patch Lianhao found in comment #15 when I have a free moment. Unfortunately the patch referenced in Comment 15 is already in qemu 0.12.4 which we are using. I do not have any other ideas at this time. Tried on F14 against arm and x86, the problem seems apply to all unfs qemu, the error all are triggered by rpcbind: 192.168.7.1 not responding then it failed to do mount of rootfs... Scott hasn't been able to reproduce. Scott will to do a final verification just in case. If he cannot reproduce this, it will be closed. I've retested both the 0.9 release and poky master branch(latest commid id 92a9d59fccdfb6587fbd12f686899a8ace5e1d2f), using the image of poky-image-sato-sdk for qemuppc, the problem seems still existing. way to reproduce it: 1. bitbake meta-ide-support 2. open a new bash console 3. source <tmpdir>/environment-setup-ppc603e-poky-linux 4. poky-extract-sdk poky-image-sato-sdk-qemuppc.tar.bz2 rootfs 5. poky-qemu zImage-qemuppc.bin rootfs/ I might need to try out several times to reproduce this bug in 0.9 release, but the latest master branch seems always ending up in the "nfs: server 192.168.7.1 not responding, still trying" issue. I tested both on Liping's 32bit and my 64bit machine, both are Ubuntu10.04. Qing had some interesting findings about this problem when he debugged another zypper related bug. He changed the unfs to Ubuntu's kernel nfs server and enabled the NFS logging. He found that the hang during the boot happened when some gtk applications trying to access some rootfs files. (He might have more details) Besides, I remembered this kind of "nfs: server 192.168.7.1 not responding, still trying" thing used to occur even after a successful booting on qemuppc, when I did some disk IO intensive work such as "ls -R" in the target. So we guess it might be possible that somehow qemuppc/ppc kernel causes the loss of UDP packet on network for NFS. I tried to modify the poky-qemu-internal to boot it with tcp unfs instead of udp unfs. It seems the problem has gone. Though there was still a strange ethenet frame error during the booting process(see the following screen shot), I can get into the login terminal. We still need more investigation to root cause this problem. The screen snapshot is as the followings: ... ... ... ... VFS: Mounted root (nfs filesystem) on device 0:12 Freeing unused kernel memory: 256k init udev[50]: starting version 164 eth0: next frame inconsistency, 0x82 Yocto (Built by Poky 4.0) 0.9+snapshot-20100203 qemuppc ttyS0 qemuppc login: *** Bug 759 has been marked as a duplicate of this bug. *** (In reply to comment #30) > Qing had some interesting findings about this problem when he debugged another > zypper related bug. He changed the unfs to Ubuntu's kernel nfs server and > enabled the NFS logging. He found that the hang during the boot happened when > some gtk applications trying to access some rootfs files. (He might have more > details) When I was trying kernel servers, I found that three postinst scripts causing the block: matchbox-keyboard-im.sh sato-icon-theme.sh tasks.sh One of the similarities is that they all call some gtk command: gtk-update-icon-cache and gtk-query-immodules. After skipping them, the login prompt can be seen, although it later hangs again if I run some heavy command like zypper. (In reply to comment #30) > So we guess it might be possible that somehow qemuppc/ppc kernel causes the > loss of UDP packet on network for NFS. > > I tried to modify the poky-qemu-internal to boot it with tcp unfs instead of > udp unfs. It seems the problem has gone. Though there was still a strange > ethenet frame error during the booting process(see the following screen shot), > I can get into the login terminal. So this probably means that ppc kernel is dropping frames in tun/tap... I think NFS doesn't do ack/reconnect itself, does it? Bruce, This problem seems to be occurring independent of whether the userspace NFS server or kernel NFS is used to boot the image. Symptoms are pointing to this being a bug in the qemuppc kernel we're booting with. Can you offer any advice? This seems related to the "ppc boots slowly" defect that was reported after the update to 2.6.37. That issue was also related to X being on the image and accesses being made that couldn't be completed by qemu. The solution to that issue were to boot the minimal images that boot quickly and fine with both NFS and disk boot. Is this being seen with the minimal image ? I can't quite tell from the bug history. I'll do some testing with ppc minimal this afternoon and report what I find. Bruce, I was able to run into this error on my fifth boot of a minimal ppc image. It occurred right at the start of when init loaded. Haven't been able to get to this yet this week. It is very unlikely for M4. We have confirmed that this only happens in NFS booting, correct ? If so, the PCI emulation on qemppc isn't PCI, it's a hacked up ISA bus. Chances are that this is part of the cause. Agree with Bruce -- I can boot from the disk image, but not UNFS. This should go into the release notes, since it can't be fixed for 1.0. The issue stems from the non-standard emulated ISA bus on qemu ppc (since it isn't emulation of any 'real' board). The bus plus the interrupt handling in ppc prep emulation can't handle the traffic/load generated by a NFS boot. In particular any X based application increase the load to the point where this is seen much sooner. Not met on nightly build toolchain images 20110313-2: http://autobuilder.pokylinux.org/nightly/20110313-2/toolchain/x86_64/poky-lsb-eglibc-x86_64-powerpc-toolchain-gmae-0.9+snapshot-20110315.tar.bz2 boot info below : poky-qemu nfs /home/liangli/test_dir2/zImage-2.6.37+git1+212cae404e57ff9dc58c808035770d51325c3512_1+6f21e2e83dc93e108183256f3efebb63a6a11b6d-r16-qemuppc-20110314080355.bin /opt/poky-lsb/temp Assuming /opt/poky-lsb/temp is an nfs rootfs Set MACHINE to [qemuppc] based on kernel [/home/liangli/test_dir2/zImage-2.6.37+git1+212cae404e57ff9dc58c808035770d51325c3512_1+6f21e2e83dc93e108183256f3efebb63a6a11b6d-r16-qemuppc-20110314080355.bin] Continuing with the following parameters: KERNEL: [/home/liangli/test_dir2/zImage-2.6.37+git1+212cae404e57ff9dc58c808035770d51325c3512_1+6f21e2e83dc93e108183256f3efebb63a6a11b6d-r16-qemuppc-20110314080355.bin] ROOTFS: [/opt/poky-lsb/temp] FSTYPE: [nfs] Acquiring lockfile for tap0... Acquiring lockfile for tap1... Setting up tap interface under sudo Acquiring lockfile for tap2... poky-export-rootfs restart /opt/poky-lsb/temp No PID file, not stopping rpc.mountd No PID file, not stopping rpc.nfsd Creating exports file... Starting User Mode rpc.mountd /opt/poky-lsb/0.9+snapshot/sysroots/x86_64-pokysdk-linux/usr/bin/pseudo -P /opt/poky-lsb/0.9+snapshot/sysroots/x86_64-pokysdk-linux/usr /opt/poky-lsb/0.9+snapshot/sysroots/x86_64-pokysdk-linux/usr/sbin/rpc.mountd --allow-non-root --mount-pid /home/liangli/.poky-sdk/mount2.pid -f /home/liangli/.poky-sdk/exports2 --rmtab /home/liangli/.poky-sdk/rmtab2 --prog 21113 -r Starting User Mode nfsd /opt/poky-lsb/0.9+snapshot/sysroots/x86_64-pokysdk-linux/usr/bin/pseudo -P /opt/poky-lsb/0.9+snapshot/sysroots/x86_64-pokysdk-linux/usr /opt/poky-lsb/0.9+snapshot/sysroots/x86_64-pokysdk-linux/usr/sbin/rpc.nfsd --allow-non-root --nfs-pid /home/liangli/.poky-sdk/nfs2.pid -f /home/liangli/.poky-sdk/exports2 --prog 11113 -P 3051 -r On your target please remember to add the following options for NFS nfsroot=IP_ADDRESS:/opt/poky-lsb/temp,nfsvers=2,mountprog=21113,nfsprog=11113,udp Warning: distccd not present, no distcc support loaded. Running qemu-system-ppc... /opt/poky-lsb/0.9+snapshot/sysroots/x86_64-pokysdk-linux/usr/bin/qemu-system-ppc -kernel /home/liangli/test_dir2/zImage-2.6.37+git1+212cae404e57ff9dc58c808035770d51325c3512_1+6f21e2e83dc93e108183256f3efebb63a6a11b6d-r16-qemuppc-20110314080355.bin -net nic,vlan=0 -net tap,vlan=0,ifname=tap2,script=no,downscript=no -cpu 603e -M prep -bios powerpc_rom.bin -no-reboot -show-cursor -usb -usbdevice wacom-tablet -nographic -no-reboot --append "root=/dev/nfs console=ttyS0 console=tty0 nfsroot=192.168.7.1:/opt/poky-lsb/temp,nfsvers=2,mountprog=21113,nfsprog=11113,udp rw ip=192.168.7.6::192.168.7.5:255.255.255.0 mem=128M " Using QEMU machine description Linux version 2.6.37.3-yocto-standard+ (pokybuild@autobuilder) (gcc version 4.5.1 (GCC) ) #1 PREEMPT Mon Mar 14 01:40:11 PDT 2011 ...... NE*000 ethercard probe at 0x300:52:54:00:12:34:56 eth0: NE2000 found at 0x300, using IRQ 9. console [netcon0] enabled netconsole: network logging started mice: PS/2 mouse device common for all mice md: linear personality registered for level -1 md: raid0 personality registered for level 0 md: raid1 personality registered for level 1 md: raid10 personality registered for level 10 md: multipath personality registered for level -4 md: faulty personality registered for level -5 device-mapper: ioctl: 4.18.0-ioctl (2010-06-29) initialised: dm-devel@redhat.com oprofile: using timer interrupt. TCP cubic registered NET: Registered protocol family 17 Registering the dns_resolver key type IP-Config: Complete: device=eth0, addr=192.168.7.6, mask=255.255.255.0, gw=192.168.7.5, host=192.168.7.6, domain=, nis-domain=(none), bootserver=255.255.255.255, rootserver=192.168.7.1, rootpath= md: Waiting for all devices to be available before autodetect md: If you don't use raid, use raid=noautodetect md: Autodetecting RAID arrays. md: Scanned 0 and added 0 devices. md: autorun ... md: ... autorun DONE. VFS: Mounted root (nfs filesystem) on device 0:12. Freeing unused kernel memory: 256k init Yocto (Built by Poky 4.0) 0.9+snapshot-20110314 qemuppc ttyS0 qemuppc login: root root@qemuppc:~# df Filesystem 1K-blocks Used Available Use% Mounted on 192.168.7.1:/opt/poky-lsb/temp 120075472 100720208 13255760 88% / tmpfs 40 0 40 0% /mnt/.psplash none 62152 72 62080 0% /dev tmpfs 62152 32 62120 0% /var/volatile tmpfs 62152 0 62152 0% /media/ram nightly build 20110313-2 git info: a33a2cc024f9c9147c92c5da70d8fbb631b968f9 test env: Host: Ubuntu 10.10 X86_64 Kernel: 2.6.35-27-generic #48-Ubuntu SMP be reminded by jiajun and yi, I have tried this several times on qemuppc sato-sdk based on nightly build 20110313-2 ,3 times boot success and 2 times boot failed. yes, actually I have never succeed boot up in my machine. Always fail with latest cs. criping bumping to 1.1 Does this bug need a new owner? I'm still the placeholder for this one. We need hold on this until we get the 1.1 kernel version locked. At a minimum, trying this with our newer dev kernel would be a good first step. Bruce, can you update the target milestone to a specific milestone? Probably M2? This bug will be used to disable usermode NFS for qemu ppc. Which is targeted at the M3 milestone. update: this boots stably on core-image-minimal, but the X based images still have problems. Using 3.0 kernel, I can boot nfs with sato rootfs successfully. I tried 10 times, every time is OK. It's more stable than 2.6.37.
IP-Config: Complete:
device=eth0, addr=192.168.7.2, mask=255.255.255.0, gw=192.168.7.1,
host=192.168.7.2, domain=, nis-domain=(none),
bootserver=255.255.255.255, rootserver=192.168.7.1, rootpath=
md: Waiting for all devices to be available before autodetect
md: If you don't use raid, use raid=noautodetect
md: Autodetecting RAID arrays.
md: Scanned 0 and added 0 devices.
md: autorun ...
md: ... autorun DONE.
VFS: Mounted root (nfs filesystem) on device 0:13.
devtmpfs: mounted
Freeing unused kernel memory: 248k init
INIT: version 2.88 booting
Error opening /dev/fb0: No such file or directory
Starting udev
Starting Bootlog daemon: bootlogd.
mount: invalid number 'sys'
Configuring network interfaces... done.
Starting portmap daemon...
net.ipv4.conf.default.rp_filter = 1
net.ipv4.conf.all.rp_filter = 1
hwclock: can't open '/dev/misc/rtc': No such file or directory
Sun Aug 14 01:31:00 UTC 2011
hwclock: can't open '/dev/misc/rtc': No such file or directory
Running postinst /etc/rpm-postinsts/avahi-daemon.sh...
Adding system startup for /etc/init.d/avahi-daemon.
Running postinst /etc/rpm-postinsts/connman-gnome.sh...
Running postinst /etc/rpm-postinsts/dates.sh...
Running postinst /etc/rpm-postinsts/dbus-1.sh...
Adding system startup for /etc/init.d/dbus-1.
Running postinst /etc/rpm-postinsts/distcc.sh...
Adding system startup for /etc/init.d/distcc.
Running postinst /etc/rpm-postinsts/gdk-pixbuf-loader-gif.sh...
Running postinst /etc/rpm-postinsts/gdk-pixbuf-loader-jpeg.sh...
Running postinst /etc/rpm-postinsts/gdk-pixbuf-loader-png.sh...
Running postinst /etc/rpm-postinsts/gdk-pixbuf-loader-xpm.sh...
Running postinst /etc/rpm-postinsts/gthumb.sh...
Running postinst /etc/rpm-postinsts/liberation-fonts.sh...
Running postinst /etc/rpm-postinsts/libnss-mdns.sh...
Running postinst /etc/rpm-postinsts/matchbox-keyboard-im.sh...
Running postinst /etc/rpm-postinsts/matchbox-session-sato.sh...
Resolved address "xml::/etc/gconf/gconf.xml.defaults" to a writable configuration source at position 0
Resolved address "xml::/etc/gconf/gconf.xml.defaults" to a writable configuration source at position 0
Resolved address "xml::/etc/gconf/gconf.xml.defaults" to a writable configuration source at position 0
Resolved address "xml::/etc/gconf/gconf.xml.defaults" to a writable configuration source at position 0
Running postinst /etc/rpm-postinsts/pango-module-basic-fc.sh...
Running postinst /etc/rpm-postinsts/pango-module-basic-x.sh...
Running postinst /etc/rpm-postinsts/sato-icon-theme.sh...
Running postinst /etc/rpm-postinsts/settings-daemon.sh...
Running postinst /etc/rpm-postinsts/sudo.sh...
Running postinst /etc/rpm-postinsts/tasks.sh...
Running postinst /etc/rpm-postinsts/update-modules.sh...
Configuration for powerpc.generic not found, trying powerpc
WARNING: -e needs -E or -F
WARNING: Couldn't open directory /lib/modules/3.0.0-rc7-yocto-standard+: No such file or directory
FATAL: Could not open /lib/modules/3.0.0-rc7-yocto-standard+/modules.dep.temp for writing: No such file or directory
Running postinst /etc/rpm-postinsts/wpa-supplicant.sh...
Running postinst /etc/rpm-postinsts/xserver-nodm-init.sh...
Adding system startup for /etc/init.d/xserver-nodm.
INIT: Entering runlevel: 3
Starting system message bus: Unknown username "xuser" in message bus configuration file
dbus.
Starting Connection Manager
modprobe: FATAL: Could not load /lib/modules/3.0.0-rc7-yocto-standard+/modules.dep: No such file or directory
modprobe: FATAL: Could not load /lib/modules/3.0.0-rc7-yocto-standard+/modules.dep: No such file or directory
Starting Dropbear SSH server: Will output 1024 bit rsa secret key to '/etc/dropbear/dropbear_rsa_host_key'
Generating key, this may take a while...
Public key portion is:
ssh-rsa AAAAB3NzaC1yc2EAAAADAQABAAAAgmRrsp7tJSXMYG1ol12XMF3g4Kqlb2PVwy7Yah2kyP2NnCX//Z1jbe1jXmipNgPVplUvfek0vqU87nQzYwCRDMhcqCOiLJyyO6MlAH15/+r9cC38+6dQVNNFLSzMVWM7qVV2614tIS0q5nOtWu66Qr345HQB2KBo1g2W60l+XSMczWk= root@qemuppc
Fingerprint: md5 85:d4:13:90:4c:3f:9c:f2:e2:48:a2:9b:50:bd:bc:a2
modprobe: FATAL: Could not load /lib/modules/3.0.0-rc7-yocto-standard+/modules.dep: No such file or directory
modprobe: FATAL: Could not load /lib/modules/3.0.0-rc7-yocto-standard+/modules.dep: No such file or directory
dropbear.
Starting advanced power management daemon: No APM support in kernel
(failed.)
Starting Distributed Compiler Daemon: distcc.
creating NFS state directory: done
starting 8 nfsd kernel threads: modprobe: FATAL: Could not load /lib/modules/3.0.0-rc7-yocto-standard+/modules.dep: No such file or directory
rpc.nfsd: Unable to access /proc/fs/nfsd errno 2 (No such file or directory).
Please try, as root, 'mount -t nfsd nfsd /proc/fs/nfsd' and then restart rpc.nfsd to correct the problem
done
starting mountd: done
starting statd: done
exportfs: /etc/exports [1]: Neither 'subtree_check' or 'no_subtree_check' specified for export "*:/".
Assuming default behaviour ('no_subtree_check').
NOTE: this default has changed since nfs-utils version 1.0.x
Starting syslogd/klogd: done
* Starting Avahi mDNS/DNS-SD Daemon: avahi-daemon
...done.
Starting Telephony daemon
Stopping Bootlog daemon: bootlogd.
Yocto (Built by Poky 5.0) 1.0+snapshot-20110813 qemuppc ttyS0
qemuppc login: root
root@qemuppc:~# uname -a
Linux qemuppc 3.0.0-rc7-yocto-standard+ #2 PREEMPT Tue Aug 16 17:56:06 CST 2011 ppc GNU/Linux
root@qemuppc:~#
Liming, is this bug already fixed with the 3.0 kernel or are you still working on it? Thanks! no, this bug should be postponed to 1.1. We will provide a script to disable unfs boot mode for qemuppc now. boot sato image of qemuppc throught UNFS, now get the following output: liangli@liangli-OptiPlex-745:~$ runqemu nfs /home/liangli/test_dir_5/zImage-3.0.3+git1+5b535279e61197cb194bb2dfceb8b7a04128387c_1+c0da1d35dee9fb63ba6bfc7907a77b0feb89766e-r1-qemuppc-20110901052849.bin /opt/poky/temp Assuming /opt/poky/temp is an nfs rootfs Set MACHINE to [qemuppc] based on kernel [/home/liangli/test_dir_5/zImage-3.0.3+git1+5b535279e61197cb194bb2dfceb8b7a04128387c_1+c0da1d35dee9fb63ba6bfc7907a77b0feb89766e-r1-qemuppc-20110901052849.bin] Error: usermode NFS boot is not available for qemuppc. git info: 705ec50edda7f7fb8c132238fe3f1477384a470e git branch: poky/master test env : host : ubuntu 10.10 x86_64 kernel :2.6.35-30-generic #56-Ubuntu SMP test steps : 1.create toolchain using "bitbake meta-toolchain" command,get toolchain tarball from "tmp/deploy/sdk" subdir of build dir ,extract to /opt dir of the host machine. 2.set toolchain env and run "runqemu-extract-sdk" command. 3.run "runqemu nfs" command. since usermode NFS boot now not supported for qemuppc, so next weekly test will check again and change the bug status to verified. (In reply to comment #55) > boot sato image of qemuppc throught UNFS, now get the following output: > liangli@liangli-OptiPlex-745:~$ runqemu nfs > /home/liangli/test_dir_5/zImage-3.0.3+git1+5b535279e61197cb194bb2dfceb8b7a04128387c_1+c0da1d35dee9fb63ba6bfc7907a77b0feb89766e-r1-qemuppc-20110901052849.bin > /opt/poky/temp > Assuming /opt/poky/temp is an nfs rootfs > Set MACHINE to [qemuppc] based on kernel > [/home/liangli/test_dir_5/zImage-3.0.3+git1+5b535279e61197cb194bb2dfceb8b7a04128387c_1+c0da1d35dee9fb63ba6bfc7907a77b0feb89766e-r1-qemuppc-20110901052849.bin] > Error: usermode NFS boot is not available for qemuppc. > > git info: 705ec50edda7f7fb8c132238fe3f1477384a470e > git branch: poky/master > > test env : > host : ubuntu 10.10 x86_64 > kernel :2.6.35-30-generic #56-Ubuntu SMP > > test steps : > 1.create toolchain using "bitbake meta-toolchain" command,get toolchain tarball > from "tmp/deploy/sdk" subdir of build dir ,extract to /opt dir of the host > machine. > 2.set toolchain env and run "runqemu-extract-sdk" command. > 3.run "runqemu nfs" command. > > since usermode NFS boot now not supported for qemuppc, so next weekly test will > check again and change the bug status to verified. verified using M4 RC1 images(nightly build 20110908-5) from: http://autobuilder.pokylinux.org/nightly/20110908-5/ |
Using the nightly build 20101007-1 from auto builder, when booting the poky-image-sdk image of the PPC architecture through UNFS, a kernel panic happened displaying: ... ... <omit> ... ... IP-Config: Complete: device=eth0, addr=192.168.7.2, mask=255.255.255.0, gw=192.168.7.1, host=192.168.7.2, domain=, nis-domain=(none), bootserver=255.255.255.255, rootserver=192.168.7.1, rootpath= md: Waiting for all devices to be available before autodetect md: If you don't use raid, use raid=noautodetect md: Autodetecting RAID arrays. md: Scanned 0 and added 0 devices. md: autorun ... md: ... autorun DONE. Looking up port of RPC 11111/2 on 192.168.7.1 Looking up port of RPC 21111/1 on 192.168.7.1 VFS: Mounted root (nfs filesystem) on device 0:12. Freeing unused kernel memory: 180k init udev: starting version 158 nfs: server 192.168.7.1 not responding, still trying nfs: server 192.168.7.1 not responding, still trying nfs: server 192.168.7.1 not responding, still trying nfs: server 192.168.7.1 not responding, still trying nfs: server 192.168.7.1 not responding, still trying nfs: server 192.168.7.1 not responding, still trying nfs: server 192.168.7.1 not responding, still trying nfs: server 192.168.7.1 not responding, still trying nfs: server 192.168.7.1 not responding, still trying nfs: server 192.168.7.1 not responding, still trying INFO: task dropbearkey:583 blocked for more than 120 seconds. "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. dropbearkey D 10002168 0 583 581 0x00000000 Call Trace: [c7981b60] [c034d434] rpc_sleep_on+0x1a4/0x2dc (unreliable) [c7981c20] [c0008d74] __switch_to+0x48/0x90 [c7981c40] [c0369dc0] schedule+0x338/0x564 [c7981c90] [c036a02c] io_schedule+0x40/0x74 [c7981ca0] [c0095e54] sync_page+0x50/0x70 [c7981cb0] [c036a9e8] __wait_on_bit_lock+0x94/0x120 [c7981ce0] [c0095dd0] __lock_page+0x8c/0x9c [c7981d10] [c0096050] find_lock_page+0x70/0x88 [c7981d30] [c009699c] filemap_fault+0x16c/0x430 [c7981d70] [c00af290] __do_fault+0x64/0x4cc [c7981dc0] [c0018e00] do_page_fault+0x1e4/0x6fc [c7981f40] [c0012e28] handle_page_fault+0xc/0x80 --- Exception: 301 at 0x10002168 LR = (null)