[Fwd: Re: [PlanetCCRMA] Sound playback freezes system
after updating kernel-modules-alsa]
Fernando Lopez-Lezcano
nando@ccrma.Stanford.EDU
Thu Apr 7 15:52:01 2005
On Thu, 2005-04-07 at 15:43, Shayne O'Connor wrote:
> Fernando Lopez-Lezcano wrote:
> >Hmmm, that should have no consequences, the rtirq script is meaningless
> >> in the stable kernel as there are no separate processes and schedulling
> >> priorities for the interrupts. A newer/better version I just got from
> >> Rui is in the works.
> >
> >Done, it is in the repository... you could upgrade if you want to test
> >it. Could you check (before or after) what the priorities of the
> >interrupts are? (the new version has much better reporting of the status
> >- thanks Rui!):
> > /etc/rc.d/init.d/rtirq status
> >
>
> BEFORE changing rtirq:
>
> [mrmachine@localhost ~]$ cat /proc/interrupts
> CPU0
> 0: 359820 XT-PIC timer 0/59820
> 1: 199 XT-PIC i8042 0/199
> 2: 0 XT-PIC cascade 0/0
> 3: 1 XT-PIC ehci_hcd 0/1
> 4: 0 XT-PIC ohci_hcd 0/0
> 5: 0 XT-PIC EMU10K1 0/0
> 7: 243652 XT-PIC parport0 0/38087
> 8: 1 XT-PIC rtc 0/1
> 9: 0 XT-PIC acpi 0/0
> 10: 20148 XT-PIC nvidia 0/20148
> 11: 36087 XT-PIC eth0, ohci_hcd 0/36087
> 12: 15381 XT-PIC i8042 0/15381
> 14: 12517 XT-PIC ide0 0/12517
> 15: 22 XT-PIC ide1 0/22
> NMI: 0
> LOC: 359742
> ERR: 5565
> MIS: 0
>
>
> [mrmachine@localhost ~]$ /etc/rc.d/init.d/rtirq status
>
> PID CLS RTPRIO NI PRI %CPU STAT COMMAND
> 276 FF 80 - 120 0.0 S< IRQ 8
> 1457 FF 70 - 110 0.0 S< IRQ 5
> 1583 FF 60 - 100 0.0 S< IRQ 3
> 1649 FF 60 - 100 0.0 S< IRQ 4
> 1383 FF 59 - 99 0.1 S< IRQ 11
> 374 FF 50 - 90 0.0 S< IRQ 1
> 20 FF 49 - 89 0.0 S< IRQ 9
> 297 FF 49 - 89 0.0 S< IRQ 12
> 333 FF 46 - 86 0.0 S< IRQ 14
> 335 FF 45 - 85 0.0 S< IRQ 15
> 1267 FF 43 - 83 0.0 S< IRQ 6
> 3088 FF 38 - 78 0.0 S< IRQ 7
> 3984 FF 37 - 77 0.1 S< IRQ 10
>
>
> AFTER:
>
> [mrmachine@localhost ~]$ cat /proc/interrupts
> CPU0
> 0: 80920 XT-PIC timer 0/80920
> 1: 71 XT-PIC i8042 0/71
> 2: 0 XT-PIC cascade 0/0
> 3: 0 XT-PIC ehci_hcd 0/0
> 4: 0 XT-PIC ohci_hcd 0/0
> 5: 0 XT-PIC EMU10K1 0/0
> 7: 5702 XT-PIC parport0 0/672
> 8: 1 XT-PIC rtc 0/1
> 9: 0 XT-PIC acpi 0/0
> 10: 3449 XT-PIC nvidia 0/3449
> 11: 6140 XT-PIC ohci_hcd, eth0 0/6140
> 12: 1009 XT-PIC i8042 0/1009
> 14: 9160 XT-PIC ide0 0/9160
> 15: 22 XT-PIC ide1 0/22
> NMI: 0
> LOC: 80841
> ERR: 5030
> MIS: 0
> [mrmachine@localhost ~]$ /etc/rc.d/init.d/rtirq status
>
> PID CLS RTPRIO NI PRI %CPU STAT COMMAND
> 276 FF 80 - 120 0.0 S< IRQ 8 XT-PIC rtc
> 1425 FF 70 - 110 0.0 S< IRQ 5 XT-PIC EMU10K1
> 1551 FF 60 - 100 0.0 S< IRQ 3 XT-PIC ehci_hcd
> 1618 FF 60 - 100 0.0 S< IRQ 4 XT-PIC ohci_hcd
> 1593 FF 59 - 99 0.1 S< IRQ 11 XT-PIC ohci_hcd, eth0
> 374 FF 50 - 90 0.0 S< IRQ 1 XT-PIC i8042
> 20 FF 49 - 89 0.0 S< IRQ 9 XT-PIC acpi
> 297 FF 49 - 89 0.0 S< IRQ 12 XT-PIC i8042
> 333 FF 46 - 86 0.2 S< IRQ 14 XT-PIC ide0
> 335 FF 45 - 85 0.0 S< IRQ 15 XT-PIC ide1
> 1264 FF 43 - 83 0.0 S< IRQ 6
> 2936 FF 38 - 78 0.0 S< IRQ 7 XT-PIC parport0
> 3860 FF 37 - 77 0.2 S< IRQ 10 XT-PIC nvidia
>
> as you can see, nothing really changed except for the labelling ...
> and it still locks up my machine, even if i change the priority in
> qjackctl to 62 ...
>
> >You could install the old rtirq package to see if this is the problem
> >(all in one line):
> >
> >rpm -Uvh --oldpackage --test
> >http://ccrma.stanford.edu/planetccrma/mirror/fedora/linux/planetcore/3/i386/rtirq-20041118-2.rhfc3.ccrma.i386.rpm
> >
> >This will mess up dependencies but you could do this temporarily. See
> >what happens whith the old version (ie: what it prints with the "status"
> >argument). You could also send me the output of "ps -axuw" to see all
> >the processes you have running, or "ps -axuw|grep IRQ" to just see the
> >interrupt request processes.
> >
> >It could be a problem in the alsa build, for some reason (obviously
> >works here). You could try reverting to the old version. This should do
> >it:
> >
> > rpm -q -a | grep 1.0.9-0.2.rc1
> >that should list alsa-driver and hte kernel-module-alsa kernel module
> >packages, if that is correct then:
> >
> > rpm -q -a | grep 1.0.9-0.2.rc1 | xargs rpm -e --nodeps
> >that should get rid of the old alsa packages
> >
> >apt-get update
> >apt-get -f install
> >
> ok, i did this, and everything works fine
Good. Same happened to Bradley, after reverting things work again.
Something must have happened during the (re)build of the alsa packages
are they were built from the same source components as 0.1.rc1. I still
don't know exactly what is that happened.
> - except this version of alsa doesn't give me Lee Revell's multi-channel patch :(
Yes, that is post rc1, I think.
> so, it looks to me like it's the updated alsa modules that are causing
> the problem - which reminds me: do i remember Alsa 1.09 RC1 needing a
> patch from Lee to work properly with the EMU10k1? i'm sure Lee
> recommended for all Soundblaster users to use RC2, not RC1?
The rc1 I packaged has a couple of critical patches from Lee for the
Emu10k1 driver, but not all the stuff that later went into rc2.
Glad sound is working again...
-- Fernando