Hi, > Since rescent -CURRENT is stable enough, I have the chance to find > out remaining pcm problem. My MP box no more has double fatal fault > and turns into random sleep. The random sleep happens after pcm having > its own problem. > > uname -av is > FreeBSD cartier.home 5.0-CURRENT FreeBSD 5.0-CURRENT #1: Wed Dec 4 23:07:36 CST >2002 [EMAIL PROTECTED]:/usr/obj/usr/src/sys/GENERI i386 > > Kernel config is simply a modified GENERIC, with options SMP, > APIC_IO, and ident GENERI. > > Below is what I've got in my dmesg. Complete dmesg is at > http://fatpipi.cirx.org/~clive/cartier_dmesg.txt > > lock order reversal > 1st 0xc6992900 pipe mutex (pipe mutex) @ /usr/src/sys/kern/sys_pipe.c:465 > 2nd 0xc0524600 sigio lock (sigio lock) @ /usr/src/sys/kern/kern_sig.c:2225 > acquiring duplicate lock of same type: "pcm channel" > 1st pcm0:record:0 @ /usr/src/sys/dev/sound/pcm/sound.c:191 > 2nd pcm0:play:0 @ /usr/src/sys/dev/sound/pcm/sound.c:191 > wakeup from sleeping state (slept 00:02:06) > ata0: resetting devices .. > done
I think the real problem is in pcm on MP system as you said, IRQ 9 maybe shared with pcm0 and acpi. Could you show me this? # acpidump | grep SCI_INT Also, I'd like to see which acpi event was occured on ramdom sleep. Backtrace is helpful to track this down. Press Ctrl+Alt+Esc into DDB, enter db> b acpi_SetSleepState to set break point, then send me the output of db> t when the ramdom sleep is occurred. BTW, you can set ignoring some of acpi events like this: # sysctl hw.acpi.power_button_state=NONE # sysctl hw.acpi.sleep_button_state=NONE # sysctl hw.acpi.lid_switch_state=NONE > acpi0: <RCC RCCNILE > on motherboard # I worry about this... :P Thanks To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
