Resume from suspend causes harddrive error

nukem996 · l33t Joined: 13 Nov 2003 Posts: 776

When ever I resume from suspend or hibernate my harddrive starts spinning and I cannt do anything that requires access to the harddrive I can only access things that are in memory. My dmesg is fill with the following in it after it resumes, this error only happens when I come back from resume, if I dont resume I never see this error.

davidgurvich · Veteran Joined: 23 Apr 2004 Posts: 1063

If you have a hitachi hard drive, it's probably going bad.

nukem996 · l33t Joined: 13 Nov 2003 Posts: 776

But I did the tests with two different harddrive testing programs and both said they were fine.

devsk · Posted: Thu Jul 13, 2006 10:36 pm Post subject:

neither you nor your drive is going crazy. Its the IDE driver which is not suspend-resume cycle safe. Head over to suspend2-users/devel list and I have reported similar problems with IDE drivers. One thing to see would be if DMA is disabled after resume (for me it was *mdma1 whereas it should be *udma5 in 'hdparm -I /dev/hda'). If it is , then enable it using hdparm and see if it makes any difference. The whole suspend thing is just not safe as yet, until of course the drivers become fully resumable.

nukem996 · l33t Joined: 13 Nov 2003 Posts: 776

On mine it stays at udma5. Disabling dma before I goto sleep mode helps a little and infact I thought it fixed it but after a few min of usage the error was back. The weird part about this is that when I first got this laptop I can Fedora which was fine with suspend I turned it into a Gentoo machine to take around the house I and I really didnt have a need for sleep mode. I noticed it wasnt working the other day which is how I discovered this error. Would you happen to know how to make it suspend aware?

devsk · Posted: Fri Jul 14, 2006 1:09 am Post subject:

There are other problems (apart from DMA settings) with IDE drivers. It seems to be fixed in 2.6.18-rc1, but that would unreliable in other ways. I have almost given up on suspend for now.

nukem996 · l33t Joined: 13 Nov 2003 Posts: 776

I used suspend on Fedora everyday for about a year and a half and never had a problem, I just havnt used it recently since my laptop is always on. Im going to need suspend again next September since im going to college, do you think it will be ready by then?

devsk · Posted: Fri Jul 14, 2006 1:49 am Post subject:

I used it during 2.6.9 days pretty reliably for about a year as well. Its just that upgrades break something or the other. I hope it gets better with 2.6.18. There is also a concerted effort to get suspend2 into mm and mainline kernel. Lets' how far that goes.

nukem996 · l33t Joined: 13 Nov 2003 Posts: 776

While experimenting with all this stuff I tired the suspend2-sources and I really didnt see a difference between them and the Gentoo sources. So if I may ask, what is the difference?

nukem996 · l33t Joined: 13 Nov 2003 Posts: 776

I was just looking through the 2.6.18-rc1 patches and it seems that there is a fix for power management going in. If you look at ide-io.c there is a hole bunch of stuff for power management. I might try just that patch tomarrow but it seems that pretty much every IDE driver has been updated(although according to the comments that has nothing to do with power management)

devsk · Posted: Fri Jul 14, 2006 3:45 am Post subject:

nukem996 · l33t Joined: 13 Nov 2003 Posts: 776

devsk · Posted: Fri Jul 14, 2006 4:32 am Post subject:

well if you are not using suspend2-sources or gentoo-sources patched with suspend2 patches, then you are using in-kernel swsusp. swsusp is more unreliable and has never worked for me. When I said I used it for a year, I used suspend2. suspend2 works better, faster and more reliably compared to in-kernel swsusp.

if any of your experiments with 2.6.18-rc1 succeed, please don't forget to post it here.

nukem996 · l33t Joined: 13 Nov 2003 Posts: 776

I just tried 2.6.18-rc1 with the suspend2 patch and I get the same error. It seem this is not fixed(at least yet). Is there a kernel bug or a link about this that you know of?

devsk · Posted: Fri Jul 14, 2006 7:15 pm Post subject:

darn!

one which related to my problem:

http://bugzilla.kernel.org/show_bug.cgi?id=2039

can you please file another bug report with your exact errors? It will be better to open a new issue becase your errors are slightly different from mine.

pgolik · Posted: Fri Jul 14, 2006 8:05 pm Post subject:

See my thread with similar problems. There is a patch in mm sources that helps with SATA (see the thread above), but I still have ATA isues on resume (like cdrom losing DMA). I'm glad to hear that there is an effort to resolve this in future kernels.

nukem996 · l33t Joined: 13 Nov 2003 Posts: 776

I posted a kernel bug which can be found at http://bugzilla.kernel.org/show_bug.cgi?id=6840

pgolik · Posted: Sun Jul 16, 2006 6:21 pm Post subject:

pgolik · Posted: Sun Jul 16, 2006 8:34 pm Post subject:

devsk · Posted: Mon Jul 17, 2006 2:53 pm Post subject:

you are right about that. There are some subtle differences in the freezer code (shared) in the suspend2 patch, which might make suspend2 slightly different. But I don't think it would be drastically different as in suspend-to-disk case.

pgolik · Posted: Fri Nov 10, 2006 8:39 pm Post subject:

Just bringing this old thread back to life. This patch finaly fixed my problem. It was the ide chipset driver allright. Now I have S3 suspend to ram fully working.[/url]

devsk · Posted: Sat Nov 11, 2006 8:57 am Post subject:

pgolik · Posted: Sat Nov 11, 2006 9:15 am Post subject:

devsk · Posted: Sat Nov 11, 2006 3:44 pm Post subject:

pgolik · Posted: Sat Nov 11, 2006 7:03 pm Post subject: