Page 3 of 4 FirstFirst 1234 LastLast
Results 21 to 30 of 36

Thread: random and persistant hard kernel lock, 10.10 (old and newer kernel) and 12.04 (live

  1. #21
    Join Date
    Jul 2010
    Beans
    58

    Re: random and persistant hard kernel lock, 10.10 (old and newer kernel) and 12.04 (l

    https://dl.dropbox.com/u/24530033/lo...mesg_10_4_2012
    https://dl.dropbox.com/u/24530033/lo...mod_10_04_2012

    these are with scaling enabled in BIOS. I assume you want sudo dmesg and sudo lsmod?
    the terminal seemed to not capture all of the dmesg output

  2. #22
    Join Date
    Nov 2009
    Beans
    Hidden!

    Re: random and persistant hard kernel lock, 10.10 (old and newer kernel) and 12.04 (l

    Quote Originally Posted by dave.kts View Post
    https://dl.dropbox.com/u/24530033/lo...mesg_10_4_2012
    https://dl.dropbox.com/u/24530033/lo...mod_10_04_2012

    these are with scaling enabled in BIOS. I assume you want sudo dmesg and sudo lsmod?
    the terminal seemed to not capture all of the dmesg output
    Then I guess it would be the ACPI cpufreq driver. The dmesg you provided is indicating that it's not nForce2 (it could have made an error!).

    This would give a very strong indication that you are using the generic driver which uses ACPI. That certainly makes some alarmbells ring since ACPI is known for it's diverse amount of 'implementations'.

    Yes, I really think you should test with cpufreq disabled inside your BIOS. Test it for 2 or 3 weeks, if the hangs are gone then we could look at bisecting it. Or I could give you steps to disable cpufreq using kernel compilation. This would allow you to use cpufreq under Windows, while avoiding it under Linux. That would be a suitable workaround.

    Even better would be to start bisecting, I looked at the amount of commits that were made to the acpi cpufreq driver between .35 and .36, there were only ten. Assuming that the occurence of you switching cpufreq driver between .35 and .36 has not occurred. And assuming that .36 allowed you to use cpufreq for this first time has not occurred.
    linux-git/fs/super.c: "Self-destruct in 5 seconds. Have a nice day...\n",

  3. #23
    Join Date
    Jul 2010
    Beans
    58

    Re: random and persistant hard kernel lock, 10.10 (old and newer kernel) and 12.04 (l

    Quote Originally Posted by Rexilion View Post
    Even better would be to start bisecting, I looked at the amount of commits that were made to the acpi cpufreq driver between .35 and .36, there were only ten. Assuming that the occurence of you switching cpufreq driver between .35 and .36 has not occurred. And assuming that .36 allowed you to use cpufreq for this first time has not occurred.
    I'm not sure I'm following.. could you clarify about the amount of commits..... between .35 and .36? If you are refering to kernel 2.6.35 vs. .36 then I may have given incorrect information, sorry. These are my currently installed 10.10 kernels:

    david@SuperU:~$ dpkg --list | grep linux-image
    ii linux-image-2.6.35-28-generic-pae 2.6.35-28.50 Linux kernel image for version 2.6.35 on x86
    ii linux-image-2.6.35-30-generic-pae 2.6.35-30.61 Linux kernel image for version 2.6.35 on x86
    ii linux-image-2.6.35-32-generic-pae 2.6.35-32.67 Linux kernel image for version 2.6.35 on x86
    ii linux-image-generic-pae 2.6.35.32.42 Generic Linux kernel image

    but keep in mind that lockups occured on 12.04 live installer CD, and other distros. This may have been a problem I've had ever since I've had this board. It replaced an AsRock board that all of the USB hub failed on, which replaced a board that the SATA controller failed on... 2 power suplies..3 HD's... winXP failed... new install: cloning failed... all while trying to make a demo for the band.

    I'm going salmon fishing this weekend so ill check back in monday. ill keep running it with scaling turned off in BIOS. but I will want to get to the bottom of this if we can. Thanks again

  4. #24
    Join Date
    Nov 2009
    Beans
    Hidden!

    Re: random and persistant hard kernel lock, 10.10 (old and newer kernel) and 12.04 (l

    Let me get this straight, from your observations I conclude the following:

    - 2.6.35 works, no lockups
    - above 2.6.35, lockups

    A prime suspect in these lock-ups is the cpufreq driver which you just disabled in the BIOS. To narrow it down, disable cpufreq in the BIOS and boot a kernel that previously caused lock-ups.

    Between 2.6.35 and 2.6.36 there were about 10 patches inside the cpufreq infrastructure and drivers.

    Furthermore, I quote:

    One more thing (in 10.10 only), I have a quad core and sometimes ill have a constant 25+% CPU use even tho sys monitor does not show the culprit of the use. It speads itself out over all of the cores, but will remain at 25-28% CPU use forever until I suspend or reboot. System doesnt seem any slower because of this, but happens frequently, randomly, and seems unrelated to predicting a kernel hang.
    To me, this also points to some cpu problem. Especially since it's 'fixed' after a reboot/resume/thaw. Since the kernel reïnitialises the CPU and freshly applies cpufreq on top of it.

    The fact that it seems unrelated doesn't mean it's relevant. It could be a race condition that occurs based on probability (which is also a bug btw).

    So, for now (and I assume you already did that before you went fishing) is that you have booted a Linux OS with a kernel version above 2.6.35 and have CPU frequency scaling disabled in the kernel.

    When you get back, check for both symptoms:

    - ghost cpu usage 25%
    - hard lock-up

    Hope they bite!
    linux-git/fs/super.c: "Self-destruct in 5 seconds. Have a nice day...\n",

  5. #25
    Join Date
    Jul 2010
    Beans
    58

    Re: random and persistant hard kernel lock, 10.10 (old and newer kernel) and 12.04 (l

    Incorrect sir, as for the .35 vs. .36 I only have 3 updates of .35 installed, have never had any .36 installed on 10.10 as i left this as a backup system a while ago to try other distros. (and now updating is broken.. more details to follow)
    I don't think there was a point that the lockups started happening in relation to kernel version, I think I had just initially blamed it on it because I hadnt noticed it happen until after one of the later kernels were installed. I had thought that booting with an earlier kernel would prevent the lockups but that wasnt the case, I still have lockups on my earliest version.
    So long story short, It seems to have nothing to do with the kernel version.

    As for the ghost process, I found it to be "apt-get" using TOP, and would stay at 100% (of 1 core) forever untill I would sudo kill the PID. after 10-15 seconds of heavy HD activity, the process would die and everything seemed fine, CPU usage back to normal. Something to do with CRON hourly, or daily, I have no idea what any of it is doing but I know that whenever I try to use an apt-get install command I get an error like: E: Unable to locate package bla/bla/bla or unable to lock something something..
    Not concerned with fixing this because it is only an issue in 10.10 and as soon as we figure out the lockup issue, I'm going to get the 12.04 to work with my video and not worry about the 10.10

    So far no lockups with scaling disabled in BIOS...

  6. #26
    Join Date
    Jul 2010
    Beans
    58

    Re: random and persistant hard kernel lock, 10.10 (old and newer kernel) and 12.04 (l

    Fail.
    just locked up with scaling in BIOS disabled.
    watching youtube, nothing else going on. everything froze and audio loops. next boot BIOS takes ~20 sec. longer to post.

  7. #27
    Join Date
    Oct 2006
    Location
    Canada
    Beans
    143
    Distro
    Ubuntu 10.04 Lucid Lynx

    Re: random and persistant hard kernel lock, 10.10 (old and newer kernel) and 12.04 (l

    hi dave.kts

    I'm no expert and i don't want to sidetrack your troubleshooting but it seems like a hardware issue to me. Random locks are typically more a hardware than a software issue. The problem may be more than just your BIOS settings.

    I believe you mentioned running windowsXP in an earlier post. This is something i would certainly try to determine whether it's a linux or in fact a hardware issue. If windowsXP also causes problems then at least you know it's a hardware issue.

    Good luck,

    Mardybear

  8. #28
    Join Date
    Jul 2010
    Beans
    58

    Re: random and persistant hard kernel lock, 10.10 (old and newer kernel) and 12.04 (l

    Doesnt hang in XP, altho i did have some hardware compatability/ driver issues that would cause BSOD.
    also worth noting that I used to use an M-audio Delta 1010lt PCI card in this, until I did a BIOS update and after that BIOS wouldnt even post with that card anymore. (I changed over to a firewire audio interface )
    Pretty sure all that is straightened out tho, but I havent used XP in months upon months. I use XP for music production, so it does not see internet. Now I'm not working on any projects, so I use Linux on it (and my laptop).

    I think I'm going to take this intel board out and pick off the capacitors one by one with my rifle this weekend. Thats what I call troubleshooting.

  9. #29
    Join Date
    Oct 2006
    Location
    Canada
    Beans
    143
    Distro
    Ubuntu 10.04 Lucid Lynx

    Re: random and persistant hard kernel lock, 10.10 (old and newer kernel) and 12.04 (l

    You mentioned a history of issues (ie. soundcard, black screen, query resolved) but haven't used windowsXP much lately. If it were my system, i would boot to windowsXP and run it solid for a couple days....complete all the windows updates, reboot as necessary, tax the system with your music software/whatever...and see if it's stable or also gives you random freezes or black screens. If still a problem then it's probably time to get your rifle loaded

    Take care,

    Mardybear

  10. #30
    Join Date
    Jul 2010
    Beans
    58

    Re: random and persistant hard kernel lock, 10.10 (old and newer kernel) and 12.04 (l

    I'm still wondering why I didnt have any lockups when I was running SolusOS 1.1 and yet tried 1.2 (and many others) and did have a lockup.
    I won't test in XP because I know the hardware works with SolusOS 1.1 and I dont want that OS using internet, which is mostly what I'm using this for right now. It already has virus issues. XP is just too vulnerable using internet, and too risky on a DAW
    Your probably wondering why I dont just use SolusOS if it works. Well I want a "studio" distro to work with since I discovered how low of a latency is possible with my firewire audio box. SolusOS 1.1 used some custom kernel that wasn't RT but I still had great performance.

Page 3 of 4 FirstFirst 1234 LastLast

Tags for this Thread

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •