23 Apr
2006
23 Apr
'06
3:14 a.m.
I'm working to get the best performance I can out of jackd working with
a Tascam US-122. Jack works well with an cheap on-board card
(frames/period 64/latency 2.9 mscec), but not so well with the US-122
(and apparently USB sound cards in general). The best frames/period I
can run it is 1024 with a lot of xruns.
I've got Ingo Molnar's realtime-preempt patch and I'm now attempting to
tune the IRQ threads. (My kernel: 2.6.15.2-rt-50studio) (Using modprobe
snd_usb_usx2y nrpacks=1)
This site describes what I'm trying to do pretty well:
http://tapas.affenbande.org/?page_id=6
I intend to tune the IRQ threads so that the usb-interrupt has highest
priority. Here are some details:
$ lsusb
Bus 002 Device 001: ID 0000:0000
Bus 001 Device 002: ID 1604:8007 Tascam US-122 Audio/Midi Interface
Bus 001 Device 001: ID 0000:0000
$ cat /proc/interrupts
CPU0 CPU1
0: 3832051 0 IO-APIC-edge timer
1: 9522 0 IO-APIC-edge i8042
7: 1 1 IO-APIC-edge parport0
8: 0 0 IO-APIC-edge rtc
9: 0 0 IO-APIC-level acpi
12: 117589 0 IO-APIC-edge i8042
14: 104 0 IO-APIC-edge ide0
50: 614644 0 IO-APIC-level ohci_hcd:usb1
58: 2 0 IO-APIC-level ehci_hcd:usb2
66: 2 0 IO-APIC-level ohci1394
217: 400222 0 IO-APIC-level libata, eth0
225: 290168 0 IO-APIC-level libata, NVidia CK804
233: 0 0 IO-APIC-level libata
NMI: 343 202
LOC: 3831594 3831830
ERR: 0
MIS: 0
I've been trying to increase the priority of IRQ 50 (ohci_hcd:usb1) like
so:
$ chrt -f -p 80 `pidof "IRQ 50"`
sched_getscheduler: No such process
failed to get pid 80's policy
It appears I have a problem getting the pidof any IRQ value.
$ echo `pidof "IRQ 0"`
$ echo `pidof "IRQ 50"`
$ echo `pidof "IRQ 9"`
pidof is installed and works...except with the IRQs.
I have another concern after reviewing another USX2Y rawusb post from
last year.
In response to a person having a similar problem as mine, one asked,
"Did you set PREEMPT_RT=y (complete real-time) on your kernel
configuration? It seems not... otherwise you would see those IRQ threads
around."
What can I do to verify the existence of the threads? What names do I
look for when running top? I don't see any command names with IRQ in
them...
Thanks,
Devin Venable
23 Apr
23 Apr
11:05 a.m.
In response to a person having a similar problem as mine, one asked,
"Did you set PREEMPT_RT=y (complete real-time) on your kernel
configuration? It seems not... otherwise you would see those IRQ threads
around."
What can I do to verify the existence of the threads? What names do I
look for when running top? I don't see any command names with IRQ in
them...
here with kernel 2.6.15-rt22
$ ps -A|grep IRQ
answers
17 ? 00:00:00 IRQ 9
389 ? 00:00:00 IRQ 12
419 ? 00:00:00 IRQ 14
421 ? 00:00:00 IRQ 15
438 ? 00:00:00 IRQ 1
912 ? 00:00:00 IRQ 6
1022 ? 00:00:00 IRQ 18
1149 ? 00:00:00 IRQ 19
1661 ? 00:00:01 IRQ 17
1825 ? 00:00:00 IRQ 7
1840 ? 00:00:00 IRQ 4
.
karsten
___________________________________________________________
Telefonate ohne weitere Kosten vom PC zum PC: http://messenger.yahoo.de
8:27 p.m.
"Did you set PREEMPT_RT=y (complete real-time) on
your kernel
> configuration? It seems not... otherwise you would see those IRQ threads
> around."
>
> What can I do to verify the existence of the threads?
here with kernel 2.6.15-rt22
$ ps -A|grep IRQ
answers
17 ? 00:00:00 IRQ 9
389 ? 00:00:00 IRQ 12
Nope. ps -A|grep IRQ returns nothing. Apparently I did miss the config
option. I just looked through the .config file I used and I could find
no PREEMPT_RT option.
These are the PREEMPT related values I can find in my .config.
# CONFIG_SCHED_SMT is not set
# CONFIG_PREEMPT_NONE is not set
# CONFIG_PREEMPT_VOLUNTARY is not set
CONFIG_PREEMPT=y
CONFIG_PREEMPT_BKL=y
I've read several howtos on adding Ingo Molnar's RT patch. None that
I've read say anything about having to explicitly adding this option to
the config file. What did I miss?
8:30 p.m.
On Sun, 2006-04-23 at 17:27 -0500, Devin Venable wrote:
I've read several howtos on adding Ingo
Molnar's RT patch. None that
I've read say anything about having to explicitly adding this option
to
the config file. What did I miss?
If the patch was applied properly you should be able to run "make
oldconfig" which will prompt you whether to enable this option
Lee
2:25 p.m.
On Sun, 2006-04-23 at 00:14 -0500, Devin Venable wrote:
I've been trying to increase the priority of IRQ 50 (ohci_hcd:usb1)
like
so:
$ chrt -f -p 80 `pidof "IRQ 50"`
sched_getscheduler: No such process
failed to get pid 80's policy
It appears I have a problem getting the pidof any IRQ value.
You must not have enabled realtime preemption when you configured the
kernel.
Lee
6791
days inactive
6791
days old
linux-audio-user@lists.linuxaudio.org
4 comments
3 participants
participants (3)
-
Devin Venable -
karsten wiese -
Lee Revell