Fedora Core 4 test 3, with updates; tested with 2.6.11.8, 2.6.11-1.1329_FC4, 2.6.12-rc4-mm2, and 2.6.12-rc4-mm2 with the IDE GTM/STM patch in http://marc.theaimsgroup.com/?l=acpi4linux&m=111680155305976&w=2, though at this point I don't think the issue is necessarily IDE related. With all of the above tested kernels, the system enters S3 suspend (with a minimum of modules loaded, init=/bin/sh.. ata_piix & ext3 are the only things loaded, essentially). On unsuspend, I hear the dvd drive make a click, and the HDD light lights up, as does the wireless light. The LCD backlight does not turn on; the system is responsive to caps lock changes and Fn-F2 hardware wireless kill switch toggle. I'll attempt to get a serial console going and see if that can provide some more insight as to what's going on. I'm open to any ideas as far as debugging and collecting info.
Created attachment 5082 [details] acpidmp output
Created attachment 5083 [details] dmidecode output
Created attachment 5084 [details] lspci -vv output
Created attachment 5085 [details] /proc/interrupts
Created attachment 5086 [details] pre-suspend dmesg output, with acpi glue debug turned on
> the system is responsive to caps lock changes Sorry, I take that back -- the system is NOT repsonsive to caps lock/scroll lock/etc. I misremembered when filing the bug, just double checked. However, it does respond to things like the hardware RF toggle switch. I hooked up a serial console -- no output at all on unsuspend. It's just completely wedged.
Some more info.. swsusp works (with compiled-in ata_piix, instead of as modules); suspend2 does not.
see Documentation/power/video.txt for workarounds to attempt to get video to restore. Also, be sure you've got the latest BIOS and the latest X and video driver for the hardware.
Part of the problem was due to PCI-E bridge config space not being restored after suspend; this was fixed by a patch from Shaohua Li about a month ago (was in issue with tg3 etherntet on the acpi-devel list; MARC isn't working for me atm, so I can't provide a link). It's an nvidia video card, so I have to use the nvidia binary driver which complicates things somewhat. It's not clear whether their driver has to do anything special for PCIE support or not. However, with the pcie-portdrv patch, as well as a patch from Jens Axboe to support ACPI suspend on SATA, I'm now able to suspend the machine and bring it back as long as the nvidia kernel module is not loaded (and thus no X). I've unfortunately not heard anything back from nvidia yet.
As the problem here seems to be in nvidia binary driver and kernel issues addressed in this bugzilla are fixed now (Shaohua's and Jen's patch in the base) I think we can close the bug now. Please feel free to reopen if you think there is some other kernel issue that needs attention. Thanks.
2.6.14 stock does not have the SATA suspend bits in yet, so out of the box it has no chance at successfully resuming. Applying the latest version of jens' patch results in a failure to suspend due to kauditd not being able to be stopped. Booting with init=/bin/sh results in a supposedly successful suspend, and a resume that seems to work until the disk is touched (find / -> hd light lights up, lockup). However, the specific tg3 issue is fixed, and separate bugs should probably be filed for the remaining issues. I've really given up on this for now; there are too many pieces involved and I have no idea really where to start to even be able to help get this working in any reasonable amount of time. =/