Description of problem: Last night I installed rawhide on a Sony Vaio SZ3 which had been running FC6 previously. Under FC6 I was able to suspend to RAM and resume quite happily. Occasionally I had to restart NetworkManager to reconnect to wireless networks, but everything else seemed to work fine. However, with the latest rawhide kernel it goes quite seriously wrong. The suspend phase seems to work and the power LED flashes amber indicating this. However, when I switch it back on, the NumLock and CapsLock LEDs start flashing and it fails to respond to any commands, even the power switch. The only way to get it out of this state is to remove the battery. Version-Release number of selected component (if applicable): 2.6.20-1.3079 How reproducible: Every time Steps to Reproduce: 1. Suspend to RAM, using menu item 2. Attempt to resume 3. Actual results: Laptop is left in unusable state Expected results: Laptop resumes normally Additional info:
Created attachment 152836 [details] dmesg output
Created attachment 152837 [details] lspci output
I haven't had the chance to test this yet, but could the problem be related to what's described in http://lkml.org/lkml/2007/4/17/41?
the cpuidle code referenced there is only in -mm right now, so no, that's unrelated. The flashing keyboard LEDs that you see are an indication that the kernel 'oopsed'. Unfortunatly, getting debug info out of the kernel when it does this is extremely difficult, as not much of the system is 'up' at that stage, so networking, console etc aren't back. I'm debugging a number of other suspend related issues at the moment, so it's possible that at some stage this will get fixed as fallout from that, but without a serial console or similar, there's not much to go on from this bug.
have a look at this: http://people.freedesktop.org/~hughsient/quirk/
I have looked at that link. The easy fixes don't seem to help much, so I tried using 'pm_trace' with a patched 2.6.22-rc3 kernel. I had already put iwl3945 into /etc/pm/config.d/unload_modules. Here's the output I got: hash matches drivers/base/power/resume.c:57
This particular problem seems to have been solved. With kernel 2.6.22-8.fc7 the laptop suspends and resumes, though the backlight level is far too dim.
The laptop is now running rawhide and the on-resume oops hasn't re-appeared.