Bugzilla – Bug 113624
comming back from suspend-to-disk fails
Last modified: 2007-06-05 11:20:30 UTC
Comming back from suspend-to-disk fails. It worked probably up to changing back to a graphical screen, at least I could see a "loading high mem" message, and then the monitor changed mode, and then turned off. The keyboard froze as well, so only a hard reset helped. ReiserFS was repairing itself with 3-4 screenfull of logs. The strange thing is, that it worked fine in beta 1 & 2.
Created attachment 47884 [details] powersave logs
Created attachment 47885 [details] hwinfo
I'd need kernel messages printed during suspend/resume.... Could you try taking kernel from beta2 and see if it works with beta3?
Sorry, I found the reason. I use the binary nVidia driver, and did not read documentation ( /usr/share/doc/packages/powersave/README.suspend-NVidia ).
so does it work if you use the open source nv driver? Or does it work if you follow the steps in README.suspend-NVidia? Or does it not work at all but you blame the nvidia driver :-) ?
It was just a blame :-) But now I have tried ťhe README.suspend-NVidia, and seems, that my blame was right. When I follow the instructions, suspend-to-disk works. I had to make only a slight change apart from the documentation: I had nothing related to agp in /etc/sysconfig/hardware. intel-agp loaded always automaticly (even if I started just runlevel 3), so I removed it from /lib/modules/... How do I change to the opensource nv driver?
Suspend to RAM works fine as well after this change. After 11 years of Linux I still have to learn again, that reading documentation helps a lot :-)
regarding the opensource driver: just change the driver to "nv", like in: | Section "Device" | BoardName "GeForce FX Go5600" | BusID "1:0:0" | Driver "nv" | Identifier "Device[0]" | VendorName "NVidia" | EndSection (i did not get around to installing the binary driver to test it yet ;-) I have to check who loads intel-agp, this may be a bug. Christian, any insight on this? intel-agp is loaded although nothing agp like is in /etc/sysconfig/hardware. This makes my documentation (README.suspend-NVidia) look bad ;-) Do we have to use /etc/hotplug/blacklist again for such stuff?
BTW: for suspend to ram, can you add a short report to bug #103930, describing what you had to do to make it work (regarding the NVidia driver, mentioning the README is enough :-) and on which system this is? Thanks!
The only remaining way to avoid to load a module is /etc/hotplug/blacklist. (maybe we'll move it to /etc/modprobe.d/blacklist, since modprobe can do this itself in between). If you want to learn, why and when a certain module is loaded, read README.Logging in docdir of newest udev package from STABLE (package for beta4).
ok, so after one release of /etc/sysconfig/hardware-controlled module loading, we are back to full hotplug. Good to know ;-) When will /etc/modprobe.d/blacklist go into production? I'll wait until then with my documentation. And do i have to put in intel-agp or intel_agp or both?
I see, I got a 'needinfo'. But what info?
i was not sure what worked for you. But after reading your reply with open eyes, i got it :-) Thanks.