nV News Forums

 
 

nV News Forums (http://www.nvnews.net/vbulletin/index.php)
-   NVIDIA Linux (http://www.nvnews.net/vbulletin/forumdisplay.php?f=14)
-   -   Unmanaged IRQ with 2.6.15 and RT patch (http://www.nvnews.net/vbulletin/showthread.php?t=63620)

SNoiraud 01-20-06 10:44 AM

Unmanaged IRQ with 2.6.15 and RT patch
 
Hi,

Some more questions :
It seems I have plenty of I/O which are not managed with the nvidia driver.
I use the 8174 release and the rt6 patch for 2.6.15. the card is a FX1400.
The machine is an HP workstation xw8200.
I get the info like that :
-sh-2.05b# cat /proc/interrupts
CPU0 CPU1
0: 31071405 0 local-APIC-edge [........N/ 0] pit
1: 1311 5 IO-APIC-edge [........./ 0] i8042
2: 0 0 XT-PIC [........N/ 0] cascade
4: 340 1 IO-APIC-edge [........./ 1] serial
9: 0 0 IO-APIC-edge [........./ 0] acpi
12: 884 55 IO-APIC-edge [........./ 0] i8042
58: 2940 2999 IO-APIC-level [........./ 0] ioc0
66: 14 15 IO-APIC-level [........./ 0] ioc1
137: 282228 268327 IO-APIC-level [........./50555] nvidia
145: 534 623 IO-APIC-level [........./ 0] eth2, Intel ICH5
201: 0 0 IO-APIC-level [........./ 0] eth0
233: 197730 211659 IO-APIC-level [........./ 0] eth1
NMI: 0 0
LOC: 31064929 31061963
ERR: 0
MIS: 0


You can see 50555 unmanaged I/O for the nvidia driver either approximatively 10 per second.

What can cause this ? Is there an option to invalid that ?
The graphic works apparently correctly without major problem for the moment.
Whit this value, the server is on a XDM login and I am on the first virtual screen.
If the counter reach 99900, the driver is removed except if we set a kernel option :
CONFIG_X86_IO_APIC

I have also BUGS reported in the logs : the following is one of them. addresses are in the nv driver.
Jan 18 16:31:53 distrib2 kernel: BUG: scheduling while atomic: IRQ 169/0x00000001/6320
Jan 18 08:57:46 distrib2 kernel: [task_blocks_on_lock+112/208] (20)
Jan 18 08:57:46 distrib2 kernel: [<c01360a0>] (20)
Jan 18 08:57:46 distrib2 kernel: [schedule+58/240] (32)
Jan 18 08:57:46 distrib2 kernel: [<c0342c8a>] (32)
Jan 18 08:57:46 distrib2 kernel: [__down_mutex+375/544] (12)
Jan 18 08:57:46 distrib2 kernel: [<c0343ba7>] (12)
Jan 18 08:57:46 distrib2 kernel: [pci_bus_read_config_word+38/96] (88)
Jan 18 08:57:46 distrib2 kernel: [<c01dc676>] (88)
Jan 18 08:57:46 distrib2 kernel: [pg0+951727333/1069061120] (20)
Jan 18 08:57:46 distrib2 kernel: [<f90184e5>] (20)
Jan 18 08:57:46 distrib2 kernel: [pg0+949694306/1069061120] (20)
Jan 18 08:57:46 distrib2 kernel: [<f8e27f62>] (20)
Jan 18 08:57:46 distrib2 kernel: [pg0+949501740/1069061120] (16)
Jan 18 08:57:46 distrib2 kernel: [<f8df8f2c>] (16)
Jan 18 08:57:46 distrib2 kernel: [pg0+951169653/1069061120] (32)
Jan 18 08:57:46 distrib2 kernel: [<f8f90275>] (32)
Jan 18 08:57:46 distrib2 kernel: [pg0+951744465/1069061120] (16)
Jan 18 08:57:46 distrib2 kernel: [<f901c7d1>] (16)
Jan 18 08:57:46 distrib2 kernel: [pg0+949470754/1069061120] (16)
Jan 18 08:57:46 distrib2 kernel: [<f8df1622>] (16)
Jan 18 08:57:46 distrib2 kernel: [pg0+950809483/1069061120] (16)
Jan 18 08:57:46 distrib2 kernel: [<f8f3838b>] (16)
Jan 18 08:57:46 distrib2 kernel: [pg0+949484988/1069061120] (16)
Jan 18 08:57:46 distrib2 kernel: [<f8df4dbc>] (16)
Jan 18 08:57:46 distrib2 kernel: [pg0+949484967/1069061120] (16)
Jan 18 08:57:46 distrib2 kernel: [<f8df4da7>] (16)
Jan 18 08:57:46 distrib2 kernel: [pg0+949502495/1069061120] (32)
Jan 18 08:57:46 distrib2 kernel: [<f8df921f>] (32)
Jan 18 08:57:46 distrib2 kernel: [pg0+949502522/1069061120] (16)
Jan 18 08:57:46 distrib2 kernel: [<f8df923a>] (16)
Jan 18 08:57:46 distrib2 kernel: [pg0+951734772/1069061120] (48)
Jan 18 08:57:46 distrib2 kernel: [<f901a1f4>] (48)
Jan 18 08:57:46 distrib2 kernel: [tasklet_action+91/192] (20)
Jan 18 08:57:46 distrib2 kernel: [<c0123e4b>] (20)
Jan 18 08:57:46 distrib2 kernel: [ksoftirqd+284/448] (28)
Jan 18 08:57:46 distrib2 kernel: [<c012411c>] (28)
Jan 18 08:57:46 distrib2 kernel: [ksoftirqd+0/448] (40)
Jan 18 08:57:46 distrib2 kernel: [<c0124000>] (40)
Jan 18 08:57:46 distrib2 kernel: [kthread+147/192] (4)
Jan 18 08:57:46 distrib2 kernel: [<c0132233>] (4)
Jan 18 08:57:46 distrib2 kernel: [kthread+0/192] (20)
Jan 18 08:57:46 distrib2 kernel: [<c01321a0>] (20)
Jan 18 08:57:46 distrib2 kernel: [kernel_thread_helper+5/24] (16)
Jan 18 08:57:46 distrib2 kernel: [<c010117d>] (16)

dmetz99 01-21-06 09:23 AM

Re: Unmanaged IRQ with 2.6.15 and RT patch
 
I get similar behavior, although the number of unmanaged interrupts is much lower (<100) with 2.6.15-rt6. This is a UP/P4 HT system with a 6800 card. The disturbing thing is that I also get some random hard-locks with RT and nv8178 (and previous drivers) with no messages in the system logs. I think the problem is some incompatability between the hrtimers subsystem and the video drivers, since the problems seem to disappear with HR_TIMERS off. The problem then becomes intermittent long latencies in interrupt response. You could try disabling hrtimers and see if that helps you.


All times are GMT -5. The time now is 01:37 PM.

Powered by vBulletin® Version 3.7.1
Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Copyright 1998 - 2014, nV News.