Image location: http://autobuilder02.pokylinux.org/crownbay-noemgd-bernard/nightly/20110312-1/machines/crownbay/x86_32/poky-image-sato-sdk-live-crownbay-noemgd-20110313020701.hddimg.bz2 branch: bernard/poky master tree poky commit: a33a2cc024f9c9147c92c5da70d8fbb631b968f9 meta-intel commit: d9ac50894e55b780aaab43b96536e0ce678ff8f7 crownbay version: B0 We did not try S3 on crownbay before. The issue is that when I type "echo mem > /sys/power/state" to make system into S3, the X window will become black but the system is still live and does not stop running.
On my crownbay a0, "echo mem > /sys/power/state" doesn't make the screen black, or do anything else apparently.
cat /sys/power/state shows empty, even though CONFIG_SUSPEND is on and the other Power Management and ACPI options in the kernel config look ok to me as well. The S3 BIOS setting on my machine is also turned on. I also added the following debugging options to the kernel: CONFIG_PM_DEBUG=y CONFIG_PM_ADVANCED_DEBUG=y CONFIG_PM_VERBOSE=y CONFIG_CAN_PM_TRACE=y CONFIG_PM_TRACE=y CONFIG_PM_TRACE_RTC=y but nothing of interest showed up. Note that the CONFIG_PM_TRACE options can be useful for determining why the machine doesn't come back from suspend, so if suspend is working on your machine, but doesn't come back, it might be worth enabling that in your kernel. Anyway, digging further into why /sys/power/state is empty, I put debugging code in the kernel and found that the pm_states for mem and standby are there, but when they're tested, the suspend_ops->valid(state) for those returns 'invalid' and therefore /sys/power/state shows empty. I need to continue to find out which suspend_ops are actually being set and why they return the invalid value.
Dongxiao and I got more information: For A0 board, the BIOS ACPI table is broken, and thus kenrel doesn't think a valid S3 available which makes write to /sys/power/state a nop. For B0 board, S3 is available but there's kernel panic in the middle of suspending devices: --irq 16: nobody cared (try booting with the "irqpoll" option) This happens after all the devices sharing irq 16 have been suspended, however one of them still trigger an interrupt. Since all related drivers on irq 16 have been unloaded, nobody cares it then and thus kernel panic. The devices sharing irq 16 are: Intel high definition audio (HDA): 00:1b.0 eth0: 02:00.1 gpio: 02:00.2 OHCI1: 02:08.0 OHCI2: 02:08.1 OHCI3: 02:08.2 EHCI: 02:08.3 the BIOS only allows disabling HDA, which doesn't change the failure. We tried "nousb" kernel option by using a PS/2 keyboard. It looks approaching further w/o irq issue but still hangs around. It looks related to usb drivers but we only tried once with it because we borrow the board from other team. S3 issues are normally tricky since the error may occur along when devices are suspended, and thus many debug methods are not affordable. The best way we use so far is to add "no_console_suspend" boot option which doesn't suspend console or else nothing can be observed on the screen. MeeGo doesn't has this issue on same B0 board, which is based on 2.6.35 kernel. CrownBay BSP currently uses 2.6.34.7 kernel. Possibly later we can try 2.6.37 yocto kernel to see whether it's better. We may also contact MeeGo team for their experience, though they haven't tested S3 on this platform for quite long time.
If you want to try the 2.6.37 kernel with B0, you can switch over to 2.6.37 and try it using the following in the meta-intel repo: http://git.pokylinux.org/cgit/cgit.cgi/meta-intel/log/?h=tzanussi/crownbay-switch-to-2.6.37 If you do, I'd also appreciate it if you could tell me if networking works for you - this hasn't been pushed to master because I'm seeing a problem there on my A0 board.
Hi Tom, I switched to your branch and started building the new BSP. However when executing "do_kernel_checkout" task for linux-yocto, it fails like: ========================= error: pathspec 'yocto/standard/crownbay' did not match any file(s) known to git. ERROR: Function 'do_kernel_checkout' failed (see /distro/dongxiao/build-crownbay/tmp/work/crownbay-poky-linux/linux-yocto-2.6.37+git1+212cae404e57ff9dc58c808035770d51325c3512_1+e8685fd8c89f8e74cd24fdceeac52befc2dad355-r16/temp/log.do_kernel_checkout.27802 for further information) ========================= I manually step into the "/distro/dongxiao/build-crownbay/tmp/work/crownbay-poky-linux/linux-yocto-2.6.37+git1+212cae404e57ff9dc58c808035770d51325c3512_1+e8685fd8c89f8e74cd24fdceeac52befc2dad355-r16/linux" folder, and "git branch -a" shows there is no "yocto/standard/crownbay" branch. Could you help to check that? Thanks, Dongxiao
All the branches in the linux folder are showed as follows, however there is no crownbay... * master meta yocto/base yocto/standard/arm-versatile-926ejs yocto/standard/base yocto/standard/beagleboard yocto/standard/common-pc-64/base yocto/standard/common-pc-64/sugarbay yocto/standard/common-pc/atom-pc yocto/standard/common-pc/base yocto/standard/emenlow yocto/standard/fsl-mpc8315e-rdb yocto/standard/mti-malta32-be yocto/standard/mti-malta32-le yocto/standard/preempt-rt/base yocto/standard/preempt-rt/common-pc yocto/standard/preempt-rt/common-pc-64 yocto/standard/qemu-ppc32 yocto/standard/routerstationpro remotes/origin/master remotes/origin/meta remotes/origin/yocto/base remotes/origin/yocto/standard/arm-versatile-926ejs remotes/origin/yocto/standard/base remotes/origin/yocto/standard/beagleboard remotes/origin/yocto/standard/common-pc-64/base remotes/origin/yocto/standard/common-pc-64/sugarbay remotes/origin/yocto/standard/common-pc/atom-pc remotes/origin/yocto/standard/common-pc/base remotes/origin/yocto/standard/emenlow remotes/origin/yocto/standard/fsl-mpc8315e-rdb remotes/origin/yocto/standard/mti-malta32-be remotes/origin/yocto/standard/mti-malta32-le remotes/origin/yocto/standard/preempt-rt/base remotes/origin/yocto/standard/preempt-rt/common-pc remotes/origin/yocto/standard/preempt-rt/common-pc-64 remotes/origin/yocto/standard/qemu-ppc32 remotes/origin/yocto/standard/routerstationpro
I think you have an old kernel package in your cache somewhere that's getting picked up instead. A couple weeks ago, the kernel being served up by the autobuilder did indeed have this problem - both the crownbay and jasperforest branches were missing in the tarball being served up by the autobuilder. That has since been fixed, but you still need to clear out your local caches e.g. git2, git2_*, etc. That should fix the problem. A fresh build from this morning doesn't have the problem: trz@elmorro:/fedora/yocto/cbjunk/build/tmp/work/crownbay-poky-linux/linux-yocto-2.6.37+git1+212cae404e57ff9dc58c808035770d51325c3512_1+e8685fd8c89f8e74cd24fdceeac52befc2dad355-r16/linux$ git branch -a master meta yocto/base yocto/standard/arm-versatile-926ejs yocto/standard/base yocto/standard/beagleboard yocto/standard/common-pc-64/base yocto/standard/common-pc-64/jasperforest yocto/standard/common-pc-64/sugarbay yocto/standard/common-pc/atom-pc yocto/standard/common-pc/base * yocto/standard/crownbay yocto/standard/emenlow yocto/standard/fsl-mpc8315e-rdb yocto/standard/mti-malta32-be yocto/standard/mti-malta32-le yocto/standard/preempt-rt/base yocto/standard/preempt-rt/common-pc yocto/standard/preempt-rt/common-pc-64 yocto/standard/qemu-ppc32 yocto/standard/routerstationpro remotes/origin/master remotes/origin/meta remotes/origin/yocto/base remotes/origin/yocto/standard/arm-versatile-926ejs remotes/origin/yocto/standard/base remotes/origin/yocto/standard/beagleboard remotes/origin/yocto/standard/common-pc-64/base remotes/origin/yocto/standard/common-pc-64/jasperforest remotes/origin/yocto/standard/common-pc-64/sugarbay remotes/origin/yocto/standard/common-pc/atom-pc remotes/origin/yocto/standard/common-pc/base remotes/origin/yocto/standard/crownbay remotes/origin/yocto/standard/emenlow remotes/origin/yocto/standard/fsl-mpc8315e-rdb remotes/origin/yocto/standard/mti-malta32-be remotes/origin/yocto/standard/mti-malta32-le remotes/origin/yocto/standard/preempt-rt/base remotes/origin/yocto/standard/preempt-rt/common-pc remotes/origin/yocto/standard/preempt-rt/common-pc-64 remotes/origin/yocto/standard/qemu-ppc32 remotes/origin/yocto/standard/routerstationpro
Update the bug title becasue it exists on both emgd and noemgd image.
I just built out a 2.6.37 kernel based image for crownbay, however this issue still exists with same phenomenon.
Being worked on. Reclassify as a medium as I don't think this blocks release. - RP Agreed. This matches what we decided in our Top Issues meeting last week. This was Top Issue #12. - Julie
One interesting thing is, yesterday I tried the 2.6.37 image on a crownbay board from MeeGo IVI team, and S3 suspend works! S3 resume is also basically OK except the screen is still dark after awake. System should be OK since I can use keyboard to poweroff the machine. As posted in previous message, I met S3 suspend error when trying 2.6.37 image on the crownbay board which is for IDF demo. The demo board has been shipped to Beijing and I don't have a chance to try it again currently. I checked the version of the two boards, and they should be the same (shell bay Fab2). One known difference is the BIOS version, but I'm not sure if it is the cause.
(In reply to comment #11) > One interesting thing is, yesterday I tried the 2.6.37 image on a crownbay > board from MeeGo IVI team, and S3 suspend works! S3 resume is also basically OK > except the screen is still dark after awake. System should be OK since I can > use keyboard to poweroff the machine. > As posted in previous message, I met S3 suspend error when trying 2.6.37 image > on the crownbay board which is for IDF demo. The demo board has been shipped to > Beijing and I don't have a chance to try it again currently. > I checked the version of the two boards, and they should be the same (shell bay > Fab2). One known difference is the BIOS version, but I'm not sure if it is the > cause. that's a good progress. We can try upgrading BIOS version on the demo board after IDF, and the good thing is that we have one success now. :-) it at least proves two things: the kernel itself does matter and the BIOS is likely the reason. display is fine which is prone to broken in Linux (many times we require BIOS to do graphics card resume), as long as you verified other parts working after resume (such as network) btw, Dongxiao mentioned that with 2.6.37 kernel graphics is not launched successfully and he's using runlevel 3. Tom, did you observe same thing in your side? If yes, Dongxiao can help debug it
(In reply to comment #12) > (In reply to comment #11) > > One interesting thing is, yesterday I tried the 2.6.37 image on a crownbay > > board from MeeGo IVI team, and S3 suspend works! S3 resume is also basically OK > > except the screen is still dark after awake. System should be OK since I can > > use keyboard to poweroff the machine. > > As posted in previous message, I met S3 suspend error when trying 2.6.37 image > > on the crownbay board which is for IDF demo. The demo board has been shipped to > > Beijing and I don't have a chance to try it again currently. > > I checked the version of the two boards, and they should be the same (shell bay > > Fab2). One known difference is the BIOS version, but I'm not sure if it is the > > cause. > > that's a good progress. We can try upgrading BIOS version on the demo board > after IDF, and the good thing is that we have one success now. :-) > > it at least proves two things: the kernel itself does matter and the BIOS is > likely the reason. display is fine which is prone to broken in Linux (many > times we require BIOS to do graphics card resume), as long as you verified > other parts working after resume (such as network) > > btw, Dongxiao mentioned that with 2.6.37 kernel graphics is not launched > successfully and he's using runlevel 3. Tom, did you observe same thing in your > side? If yes, Dongxiao can help debug it No, I haven't seen any graphics problems with 2.7.37...
The error may be related with my error config with EMGD driver. I rebuild a image with EMGD 1742, and X works well on crownbay with 2.6.37 kernel. Besides, I tested suspend/resume within X environment, and everything works fine. (Network doesn't work since the image I build doesn't contain Tom's fix for network). Mark this bug as fixed.