View previous topic :: View next topic |
Author |
Message |
Swanker n00b
Joined: 19 Jun 2005 Posts: 6
|
Posted: Tue Jul 26, 2005 5:30 pm Post subject: nF4 motherboards, X2 procs, and lost tick(s)! |
|
|
I am trying to determine whether it is just my motherboard/bios causing me headaches in Linux or if it could be a nF4/AMD64/dual-core issue in general.
After getting 2.6.12 and above kernels up and running, I noticed that sometimes my clock would drift quite a bit or I'd see warnings like these in the system logs:
Jul 19 20:41:36 4400 kernel: rtc: lost some interrupts at 1024Hz.
Jul 19 17:36:57 4400 kernel: Your time source seems to be instable or some driver is hogging interrupts
My keyboard also acted as if it was sticking in Xorg so my sentences would look lllliiiiikkkkkeeee ttthhhhiiiissssss.
I then started using the "report_lost_ticks" kernel parameter to get more detailed errors and have been getting warnings/errors such as:
time.c: Lost 10 timer tick(s)! rip setup_boot_APIC_clock+0x112/0x120)
time.c: Lost 1 timer tick(s)! rip __do_softirq+0x54/0xf0)
time.c: Lost 1 timer tick(s)! rip acpi_processor_idle+0x137/0x389)
Andi Kleen at SUSE believes it to be bad SMM code in my motherboard's BIOS -- the code is slow and causing problems.
Through some testing I found out that if I set TIMER_HZ to 100 the problem was less frequent. Using TIMER_HZ=1000 filled the logs with warnings. I have tried various 2.6.12 and 2.6.13 (including -mm) kernels with just about every kernel parameter possible with no success.
My system details:
DFI LP nF4 SLI-DR (6/23 official bios)
AMD X2 4400+
4GB (4x1GB) @ PC3200, 2T, dual channel
tried using onboard AC97 and PCI Audigy2
Using onboard nF4 SATA and SiI3114 SATA
Has anyone actually solved their "Your time source seems to be instable or some driver is hogging interrupts" problem and checked it via the "report_lost_ticks" kernel parameter? If so, could you post your hardware list? Better yet, are there any folks with similar hardware to mine that don't have this issue at all? |
|
Back to top |
|
|
crazycat l33t
Joined: 26 Aug 2003 Posts: 838 Location: Hamburg, Germany
|
Posted: Tue Jul 26, 2005 6:16 pm Post subject: |
|
|
Seems like dfi lanparty just plain sucks,also problems with segfaults for many people if you search the forum. Anyway here is thread you might be interested in https://forums.gentoo.org/viewtopic.php?t=191716 . Have a nice read. |
|
Back to top |
|
|
paulisdead Guru
Joined: 10 Apr 2002 Posts: 510 Location: Seattle, WA
|
Posted: Tue Jul 26, 2005 10:40 pm Post subject: |
|
|
crazycat wrote: | Seems like dfi lanparty just plain sucks,also problems with segfaults for many people if you search the forum. Anyway here is thread you might be interested in https://forums.gentoo.org/viewtopic.php?t=191716 . Have a nice read. |
The DFI board's been great for me. I had the lost ticks problem with both the neo4 and the DFI, and I posted a fix in that worked for me on the last page of that thread. For some reason it stopped booting one day with clock=pmtmr, but if you remove that part it'll work fine. _________________ "we should make it a law that all geeks have dates" - Linus |
|
Back to top |
|
|
cciechad n00b
Joined: 28 Jun 2005 Posts: 31
|
Posted: Wed Jul 27, 2005 2:18 am Post subject: |
|
|
Using latest ~amd64 kernel with timers enabled in the kernel has solved the problem for me.
Chad Simmons |
|
Back to top |
|
|
paulisdead Guru
Joined: 10 Apr 2002 Posts: 510 Location: Seattle, WA
|
Posted: Wed Jul 27, 2005 4:15 am Post subject: |
|
|
cciechad wrote: | Using latest ~amd64 kernel with timers enabled in the kernel has solved the problem for me.
Chad Simmons |
which kernel out of curiosity, the gentoo sources? I've been using 2.6.13-rc1, since rc3 gave me some problems with the nvidia drivers. _________________ "we should make it a law that all geeks have dates" - Linus |
|
Back to top |
|
|
Swanker n00b
Joined: 19 Jun 2005 Posts: 6
|
Posted: Wed Jul 27, 2005 5:34 am Post subject: |
|
|
For those who think their timer problems have been solved, could you try booting with 'report_lost_ticks' and see if you still get errors such as:
time.c: Lost 1 timer tick(s)! rip acpi_processor_idle+0x137/0x389)
I've gotten some configs where the kernel wouldn't complain much, but after adding that boot parameter I found I still had issues. I guess a single lost tick here and there isn't enough to get the kernel to issue warnings. |
|
Back to top |
|
|
paulisdead Guru
Joined: 10 Apr 2002 Posts: 510 Location: Seattle, WA
|
Posted: Wed Jul 27, 2005 6:07 am Post subject: |
|
|
Swanker wrote: | For those who think their timer problems have been solved, could you try booting with 'report_lost_ticks' and see if you still get errors such as:
time.c: Lost 1 timer tick(s)! rip acpi_processor_idle+0x137/0x389)
I've gotten some configs where the kernel wouldn't complain much, but after adding that boot parameter I found I still had issues. I guess a single lost tick here and there isn't enough to get the kernel to issue warnings. |
I'll have to look at that, but the real problem I had was that my clock was running way too fast, and would cause all sorts of havoc when under a high load. Videos would play too fast, tapping a key would result in it thinking I had been holding it down for awhile, etc. Since I've had such visible side affects from it, and they're gone now, I'm happy. _________________ "we should make it a law that all geeks have dates" - Linus |
|
Back to top |
|
|
KhanReaper n00b
Joined: 21 Sep 2004 Posts: 37
|
Posted: Thu Jul 28, 2005 4:01 am Post subject: |
|
|
Hi,
Speaking of keyboard anomalies on an nForce4 board with an X2 processor, I have experienced some issues other than those you described, but I still believe that what I have posted will be useful.
On occasion, there are governor problems that result in an 'oops' that only kills the keyboard but not the machine. For a while, I had thought that it was some KVM problem, but it became clear that it wasn't.
Take a look at some of the resources in this bugzilla report, and consider applying the patch listed in one of the mailing list threads to the kernel.
https://bugs.gentoo.org/show_bug.cgi?id=100132 _________________ Even the Politburo concurs with Process of Elimination. Shouldn't you? |
|
Back to top |
|
|
robnotts Guru
Joined: 15 Mar 2004 Posts: 405 Location: Nottingham, UK
|
Posted: Thu Jul 28, 2005 5:54 am Post subject: |
|
|
Khanreaper...
Have you applied the powernow-k8 patch? Since I applied that, I have a completely clean dmesg from boot to shutdown.
Rob. _________________ ---
Gentoo Phenom][ X4 955 on AMD790 + Geforce 220GT 8GB/1.75TB (Desktop)
+ MythTV (3xFreeview,1xFreesat HD) on 1080p
Gentoo Turion64 X2 Geforce 6150 2GB/120GB (Laptop) |
|
Back to top |
|
|
KhanReaper n00b
Joined: 21 Sep 2004 Posts: 37
|
Posted: Thu Jul 28, 2005 9:16 am Post subject: |
|
|
robnotts wrote: | Khanreaper...
Have you applied the powernow-k8 patch? Since I applied that, I have a completely clean dmesg from boot to shutdown.
Rob. |
Yes, I applied the patch, and dmesg has been henceforth clean of these notifications. There have been no more panics as well. The only complaint that I will lodge is that some driver (skb) or (skge) has been dumping a lot of garbage as of late:
Code: |
skb: pf=2 (unowned) dev=eth1 len=242
PROTO=17 192.168.0.2:138 192.168.0.255:138 L=242 S=0x00 I=25 F=0x4000 T=64
ip_local_deliver: bad skb: PRE_ROUTING LOCAL_IN LOCAL_OUT POST_ROUTING
skb: pf=2 (unowned) dev=eth1 len=236
PROTO=17 192.168.0.2:138 192.168.0.255:138 L=236 S=0x00 I=26 F=0x4000 T=64
ip_local_deliver: bad skb: PRE_ROUTING LOCAL_IN LOCAL_OUT POST_ROUTING
skb: pf=2 (unowned) dev=eth1 len=242
PROTO=17 192.168.0.2:138 192.168.0.255:138 L=242 S=0x00 I=27 F=0x4000 T=64
ip_local_deliver: bad skb: PRE_ROUTING LOCAL_IN LOCAL_OUT POST_ROUTING
skb: pf=2 (unowned) dev=eth1 len=236
PROTO=17 192.168.0.2:138 192.168.0.255:138 L=236 S=0x00 I=28 F=0x4000 T=64
ip_local_deliver: bad skb: PRE_ROUTING LOCAL_IN LOCAL_OUT POST_ROUTING
skb: pf=2 (unowned) dev=eth1 len=242
PROTO=17 192.168.0.2:138 192.168.0.255:138 L=242 S=0x00 I=29 F=0x4000 T=64
ip_local_deliver: bad skb: PRE_ROUTING LOCAL_IN LOCAL_OUT POST_ROUTING
skb: pf=2 (unowned) dev=eth1 len=236
PROTO=17 192.168.0.2:138 192.168.0.255:138 L=236 S=0x00 I=30 F=0x4000 T=64
|
_________________ Even the Politburo concurs with Process of Elimination. Shouldn't you? |
|
Back to top |
|
|
robnotts Guru
Joined: 15 Mar 2004 Posts: 405 Location: Nottingham, UK
|
Posted: Thu Jul 28, 2005 9:22 am Post subject: |
|
|
Ah right, I also applied the patch for the sk98lin driver, and am now using my Marvell ethernet port, rather than forcedeth... I was having a few problems with the forcedeth driver not locking the machine, but locking the net, with the transmission light constantly being lit.
Rob. _________________ ---
Gentoo Phenom][ X4 955 on AMD790 + Geforce 220GT 8GB/1.75TB (Desktop)
+ MythTV (3xFreeview,1xFreesat HD) on 1080p
Gentoo Turion64 X2 Geforce 6150 2GB/120GB (Laptop) |
|
Back to top |
|
|
ikshaar Veteran
Joined: 23 Jul 2002 Posts: 1339 Location: Baltimore, MD
|
Posted: Thu Aug 04, 2005 8:05 pm Post subject: |
|
|
Not sure if it is linked... new MB Neo4 X2 proc 4200... I have insane clock drift... clock going too fast around 10min per hour.
No message in dmesg about lost ticks though....
I used a config derived from LiveCD as I could not boot otherwise. Only change I did are driver related I think (sata, nv_sata, skge). I disabled CnQ (BIOS and kernel) as it does not work (I am not up for patching my kernel yet).
Linux 2.6.12-gentoo-r6 #2 SMP x86_64 AMD Athlon(tm) 64 X2 Dual Core Processor 4200+ AuthenticAMD GNU/Linux _________________ "May God stands between you and harm in all the empty places where you must walk" - Babylon 5 |
|
Back to top |
|
|
|