PROBLEM: Maestro sound module locks up the computer
Maestro sound module locks up the computer
When playing music through the maestro sounddriver, the computer will lock up at irregular intervals. These lock ups will cause the entire machine to freeze, but a single press on the keyboard will cause the machine to continue as if nothing had happened. The machine's services are still active though, and if I ssh in to the machine there are two apparent problems; the input/output 'lags' behind (that is, when I type 'ls', I have to press return twice to see the output, the two returns are still processed as two returns, but it seems like I lag one command behind all the time) and that the machines clock is acting weird. When typing 'date', I noticed that the time seemed to wrap around after 8 seconds; e.g. say the clock is '16:40:10'. After 6 seconds the clock is, not surprisingly, '16:40:16'; but after 8 seconds it's '16:40:10' again. I'm not sure it's exactly 8 seconds though, as I've only had the chance of verifying this once.
I've narrowed it down (I think) to the maestro driver as these lockups only happen when I play music. It doesn't matter if I use mplayer from console, or xmms from X, the lockups still happen. These lockups doesn't happen if I don't play music, so...
This started to happen a couple of months ago.
More information can be obtained if you ask me. It's very easy to reproduce the problem (as it happens several times a day), so I can test various hacks if you have any.
Please CC me any replies, I only read the list filtered through kernel-traffic. ;)
//F
The machine is an Compaq Armada M700, with an Maestro 2E soundcard:
maestro: Configuring ESS Maestro 2E found at IO 0x3000 IRQ 11
maestro: subvendor id: 0xb1120e11
maestro: not attempting power management.
maestro: AC97 Codec detected: v: 0x83847609 caps: 0x6940 pwr: 0xf
maestro: 1 channels configured.
maestro: version 0.15 time 08:33:02 Mar 5 2003
ver_linux output:
Linux wolverine.xmen 2.4.20-gentoo-r1 #2 Wed Mar 5 08:15:06 CET 2003 i686 Pentium III (Coppermine) GenuineIntel GNU/Linux
Gnu C 3.2.2 Gnu make 3.80 util-linux 2.11y mount 2.11y modutils 2.4.25 e2fsprogs 1.32 reiserfsprogs 3.6.4 Linux C Library 2.3.1 Dynamic linker (ldd) 2.3.1 Procps 2.0.10 Net-tools 1.60 Kbd 1.06 Sh-utils 2.0.15 Modules Loaded nls_iso8859-1 floppy thermal processor fan button
battery ac maestro hid uhci usbcore keybdev mousedev input
proc/iomem:
00000000-0009fbff : System RAM
0009fc00-0009ffff : reserved
000a0000-000bffff : Video RAM area
000c0000-000c7fff : Video ROM
000d0000-000d17ff : Extension ROM
000f0000-000fffff : System ROM
00100000-13feffff : System RAM
00100000-002c496c : Kernel code
002c496d-00348aff : Kernel data
13ff0000-13ff37ff : reserved
13ff3800-13ffffff : ACPI Non-volatile Storage 40000000-410fffff : PCI Bus #01
40000000-40ffffff : ATI Technologies Inc Rage Mobility P/M AGP 2x 41000000-41000fff : ATI Technologies Inc Rage Mobility P/M AGP 2x
41100000-41100fff : Texas Instruments PCI1450 41180000-41180fff : Texas Instruments PCI1450 (#2) 41200000-4121ffff : Intel Corp. 82557/8/9 [Ethernet Pro 100] 41280000-41280fff : Intel Corp. 82557/8/9 [Ethernet Pro 100]
41280000-41280fff : eepro100
41300000-41300fff : Lucent Microelectronics LT WinModem 50000000-53ffffff : Intel Corp. 440BX/ZX/DX - 82443BX/ZX/DX Host bridge
Quoted 4 lines · Fredrik Jagenheim
8 seconds; e.g. say the clock is '16:40:10'. After 6 seconds the clock is, not surprisingly, '16:40:16'; but after 8 seconds it's '16:40:10' again. I'm not sure it's exactly 8 seconds though, as I've only had the chance of verifying this once.
That kind of weirdness is useful info
Quoted 6 lines
I've narrowed it down (I think) to the maestro driver as these lockups only happen when I play music. It doesn't matter if I use mplayer from console, or xmms from X, the lockups still happen. These lockups doesn't happen if I don't play music, so...
This started to happen a couple of months ago.
What did you change at the time it happened - any new kernels etc ?
Quoted 3 lines
More information can be obtained if you ask me. It's very easy to reproduce the problem (as it happens several times a day), so I can test various hacks if you have any.
Does it occur if you turn off all ACPI and APM power management support in the kernel and in the BIOS ?
Quoted 2 lines · Fredrik Jagenheim
When playing music through the maestro sounddriver, the computer will lock up at irregular intervals. These lock ups will cause the entire machine to freeze,
[...]
Quoted 4 lines
I've narrowed it down (I think) to the maestro driver as these lockups only happen when I play music. It doesn't matter if I use mplayer from console, or xmms from X, the lockups still happen. These lockups doesn't happen if I don't play music, so...
Maybe I have a similar problem. ESS Maestro 2E, OSS maestro.o driver, kernel 2.4.18 (and all versions since 2.4.0), Toshiba 4100XCDt laptop. However, here first the sound becomes somehow distorted and a few minutes later the machine just powers off. For me, it looks like a hardware failure because the same problem was visible when I got the machine with WinNT preinstalled.
Could you please tell me if sound is distorted before the "hangs"? In that case, try pausing the output and then continue playing. The distortion should disappear and you should be safe from hangs until the next distortion period.
Regards,
Carl-Daniel
Signature · 1 line
http://www.hailfinger.org
Quoted 7 lines · Alan Cox
Quoted 4 lines · Fredrik Jagenheim
8 seconds; e.g. say the clock is '16:40:10'. After 6 seconds the clock is, not surprisingly, '16:40:16'; but after 8 seconds it's '16:40:10' again. I'm not sure it's exactly 8 seconds though, as I've only had the chance of verifying this once.
That kind of weirdness is useful info
I seen that on one computer too, with 2.5.59. For some strange reason IRQ0 from timer was stopped, and thus system reported jiffies + tsc / <host_cpu_freq> as current time, taking only 32bit tsc into account. After I disabled irqbalance, problem never reappered (it was two cpus PIII from Dell, with serverworks chipset & 8GB memory; I do not remember model number).
System worked in that state for several minutes, until it lost some NFS reply, which it never retried due to stopped timer...
Best regards,
Petr Vandrovec [email protected]