On Tue, 2006-06-27 at 22:17 +0300, Georgios Petasis wrote:
> Dear Fedora users,
>
> I have an AMD-64 system which I have 32-bit fedora installed. The
> system was working
> perfectly with fedora core 4. When core 5 was released, I upgraded
> again to the 32-bit version.
> Since day one, I experienced random freezes. So, I did a clean install
> of 32-bit core 5.
> The freezes didn't stop. I get them so frequently, that it is
> impossible to work with the system.
>
> My system is in this state quite many months now (this means that I
> don't use it at all). I have tried
> to ask for help in this list on how to debug the freezes, but with no
> results.
> Of course I have tried to check my hardware, so I did a memtest86
> check on memory and a SMART
> check on my (only) disk. I even changed the graphics card. No hardware
> errors have been found but the
> problem persisted.
>
> Today, I came across a kernel boot argument that can help debug such
> freezes nmi_watchdog=1. So,
> I booted my system with this option, and I haven't yet got a freeze
> (so I added permenantly in my grub.conf :-))
> Looking in the output of dmesg, I can see strange entries, like the
> one at the end of this mail.
> Is there a problem with the cpuspeed process? Is it safe to disable
> it? What is its job?
> Is there anything I can set in my BIOS to prevent this?
> Finally, is it a bug that should be reported somewhere? (I have never
> reported a bug using bugzilla :-))
>
> Any help will be welcomed :-)
>
> Regards,
>
> George
>
> BUG: unable to handle kernel NULL pointer dereference at virtual
> address 00000000
> printing eip:
> c040b8bf
> *pde = 00000000
> Oops: 0000 [#1]
> last sysfs file: /devices/system/cpu/cpu0/cpufreq/scaling_setspeed
> Modules linked in: autofs4 sunrpc ip_conntrack_ftp
> ip_conntrack_netbios_ns ipt_REJECT xt_state ip_conntrack nfnetlink
> xt_tcpudp iptable_filter ip_tables x_tables dm_mirror dm_mod video
> button battery ac ipv6 lp parport_pc parport nvidia(U) floppy ehci_hcd
> ohci_hcd r8169 forcedeth i2c_nforce2 i2c_core ext3 jbd sata_nv libata
> sd_mod scsi_mod
> CPU: 0
> EIP: 0060:[<c040b8bf>] Tainted: P VLI
> EFLAGS: 00010246 (2.6.17-1.2139_FC5 #1)
> EIP is at powernowk8_target+0x4fb/0x742
> eax: 00000000 ebx: 00000008 ecx: c0010042 edx: 00000006
> esi: f7e7e5c0 edi: 00000000 ebp: 0000000a esp: f41e9e9c
> ds: 007b es: 007b ss: 0068
> Process cpuspeed (pid: 1239, threadinfo=f41e9000 task=f3c3faa0)
> Stack: 00000002 c06197f5 c0619d0e 00000000 0000000c 00000006 00000000
> 00000002
> 000c3500 f7e31200 0000000c 00000002 00000006 0000000a 00000000
> 001e8480
> 000c3500 c0600f00 00000002 c040b3c4 f7e31200 00000000 000c3500
> c059de73
> Call Trace:
> <c0600f00> __mutex_lock_slowpath+0x262/0x439 <c040b3c4>
> powernowk8_target+0x0/0x742
> <c059de73> __cpufreq_driver_target+0x52/0x61 <c059ee1f> store_speed
> +0xab/0xc0
> <c059ed74> store_speed+0x0/0xc0 <c059d2d3> store+0x31/0x41
> <c04957c0> sysfs_write_file+0xa6/0xcc <c049571a> sysfs_write_file
> +0x0/0xcc
> <c045c597> vfs_write+0xa8/0x150 <c045cacb> sys_write+0x41/0x67
> <c0402cb3> syscall_call+0x7/0xb
> Code: f8 ff ff 85 c0 0f 85 bb 01 00 00 8b 46 28 39 f8 74 19 89 7c 24
> 08 89 44 24 04 c7 04 24 4e 9d 61 c0 e8 09 11 01 00 e9 9b 01 00 00 <8b>
> 46 24 39 44 24 30 74 15 89 44 24 08 8b 44 24 30 c7 04 24 99
> EIP: [<c040b8bf>] powernowk8_target+0x4fb/0x742 SS:ESP 0068:f41e9e9c
> BUG: cpuspeed/1239, lock held at task exit time!
> [c06dfe80] {userspace_mutex}
> .. held by: cpuspeed: 1239 [f3c3faa0, 115]
> ... acquired at: store_speed+0x6b/0xc0
Hi George,
What motherboard are you using ? I used to have a Abit motherboard that
would do that to me at least once a day and had to be rebooted. Turned
out to be a capacitor problem -- the original motherboard had a flaw
which was that an under-rated capacitor was being used. This was
reported on some message boards, and there were 2 solutions: send the
board back to the manufacturer or replace the capacitor with another
that had the recommended rating.
Anyway, the frustrating thing was that you could not predict the
freezes, and on Windows, it worked fine (unless you started Photoshop).
You may want to check the message boards for your motherboard.
I am using an Nvidia motherboard, K8NGM2-FID, from MSI (the ONLY brand I
use now), with an Athlon AMD64 and running 64-bit FC5 with no problems
so far.
Regards,
--
Pascal Chong
email: chongym@xxxxxxxxxxxxxx
web: http://cymulacrum.net
pgp: http://cymulacrum.net/pgp/cymulacrum.asc
"La science ne connaît pas de frontière parce que la connaissance
appartient à l’humanité. et que c’est la flamme qui illumine le monde."
-- Louis Pasteur
Attachment:
signature.asc
Description: This is a digitally signed message part
