Red Hat Bugzilla – Bug 174958
Kernel Mischief: various problems, intermittantly
Last modified: 2007-11-30 17:11:18 EST
From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.12) Gecko/20050922 Fedora/1.0.7-1.1.fc4 Firefox/1.0.7 Description of problem: Unhindered by an abundance of money, I'm still running FC4 on the exact same hardware I was running RH8/9. Serious changes are starting to really impact my day, and they appear to be at the kernel level...the level that's historically been solid. 1. Keyboard From time to time, and without any visible reason, the keyboard "dies". I'm typing one minute and not, the next. If I unplug the keyboard and plug it back in, we're back on the road. It's a PS/2, in the simple, PS/2 port. This is almost an everyday occurance. I also have a problem with the left control key; it doesn't always respond, and I wind up typing in lower case until I notice it, and re-type it. It's also doing this on a spare keyboard, it's not the mechanism. 2. Mouse I have a very old, trusty Logitech Optical Mouse that's been solid for at least four years. But now it, too, stops working, requiring the USB plug to be unplugged, and plugged back in. This happens at least once a week, sometimes twice. It also responds multiple times to button clicks, but I'm sure that's just due to the age of the device. 3. Recent kernel versions have a worsening problem, from my point of view: a. 3-second stuttering in 3D applications. In UnrealTournament, Quake 3, and Enemy Territory all have an annoying 3-second delay on no particular schedule. Since these are real-time games, you can see what a broomstick-in-the-spokes it is. b. Streaming media is problematic. Both XMMS and RB until recently would kick into fast-forward. I'd have to stop it, and restart it to get it to go. It was part of the I/O Scheduler; it happens a lot less often using Anticipatory. In the last week, the problem has begun to STOP these programs, and in the case of RB, it's killing the program. (!) c. The machine hangs; no schedule, no clues. The machine here is an Athlon 1200, using the VIA chipset...not known for it's high throughput, but can that be the only problem? Maybe it's a problem with my hardware? Version-Release number of selected component (if applicable): kernel-2.6.14-1.1644_FC4 How reproducible: Sometimes Steps to Reproduce: Problem(s) occur in a random order. Actual Results: Most times, it works. Sometimes it hangs, and I'm rebooting daily, now. Expected Results: Everything normal. Additional info: I'm happy to set up the system to be investigated from the outside, and I'm happy to help in any way I'm able.
How's the memtest86 doing on that box? I saw systems going bad in the middle of an installation before. And I shudder to think what if ACPI does something bad. We'll never find it. But it may be worth it to try to disable ACPI with various kernel options... RHL9 didn't have it, IIRC.
Well, memtest86 ran fine for a few hours...though this isn't a new install; it's been installed for a couple of momths. This morning in the first two hours back online, it locked 4 times at random, locked the entire machine twice, and killed RB 4-5 times as well... And I've turned off ACPI to limit the number of possible sources....no joy....
More data: when "sitting still" running the TV capture card and watching Gkrellm, I noticed something: the TV delays "pulse" in time with the network access. I'm using NFS here, a shared /home directory and /shares (communally shared media, etc) with nothing special at the workstation. Are there any network-card/kernel bugs I'm not able to find here?
Well, there's egg on MY face. My trusty machine who's been with me for so long has finally died. I tested the same kinds of things on a similar box (one on to which I finally fell back) and none of the serious problems were there. The pulsing, the 3D delays, all of this happened on a machine that's been gradually getting worse for some time. Memtest showed nothing, the whole time. Is there any computer forensics project going on? My experience until now has been that computer problems tend to be "digital", being a problem or not...even with they're intermittant. It's really odd to see one creep up like rust. Thanks guys- sorry to have wasted your time.
My sympathies...