On 22/03/07 at 10:49 -0400, Alan Stern wrote: > On Wed, 21 Mar 2007, Lucas Nussbaum wrote: > > > On 21/03/07 at 08:41 -0800, Andrew Morton wrote: > > > On Wed, 21 Mar 2007 07:18:45 -0700 bugme-daemon@xxxxxxxxxxxxxxxxxxx wrote: > > > > > > > http://bugzilla.kernel.org/show_bug.cgi?id=8249 > > > > > > > > Summary: NMI watchdog detected LOCKUP when plugging USB devices > > > > with acpi=off > > > > Kernel Version: 2.6.21-rc4 > > > > Status: NEW > > > > Severity: high > > > > Owner: greg@xxxxxxxxx > > > > Submitter: lucas@xxxxxxxxxxxxxxxxxx > > > > > > > > > > > > Most recent kernel where this bug did *NOT* occur: none, as far as I know > > > > Problem Description: > > > > > > > > My kernel is running with acpi=off to workaround another bug. When I turn on my > > > > USB printer (a Canon PIXMA MP750), I get this on the serial console: > > It looks like your problem occurs either as a result of the USB printer > driver's actions or as a result of HAL. You could try turning off the HAL > daemon and renaming the usblp.ko file so that it won't get loaded > automatically. Then after plugging in the printer you could insmod the > renamed driver file by hand, and see whether that causes the oops. Then > start HAL by hand and see whether that causes the oops. Ah, good idea. A problem specific to usblp is unlikely, since it also fails with other USB devices (mouse, remote controler). If I kill hald before turning on the printer, the kernel doesn't deadlock. If then, I modprobe usblp, the module gets inserted ok. If then, I start hald (or if I start if with the module unloaded), it deadlocks. I straced hald. The output is on http://blop.info/strace-hald.log, but it's probably not very interesting, except for the end: [pid 2619] close(14) = 0 [pid 2619] munmap(0xb7f19000, 4096) = 0 [pid 2619] readlink("/sys/block/ram0/device", 0x8150f30, 256) = -1 ENOENT (No such file or directory) [pid 2619] open("/sys/block/ram0/slaves", O_RDONLY|O_NONBLOCK|O_LARGEFILE|O_DIRECTORY) = 14 [pid 2619] fstat64(14, {st_mode=S_IFDIR|0755, st_size=0, ...}) = 0 [pid 2619] fcntl64(14, F_SETFD, FD_CLOEXEC) = 0 [pid 2619] getdents64(14, /* 2 entries */, 4096) = 48 [pid 2619] getdents64(14, /* 0 entries */, 4096) = 0 [pid 2619] close(14) = 0 [pid 2619] open("//proc/apm", O_RDONLY|O_LARGEFILE) = 14 I also confirmed that "cat /proc/apm" works fine when the USB printer is off. But it deadlocks when the USB printer is on. With the printer off: # cat /proc/apm 1.16ac 1.2 0x03 0x01 0xff 0x80 -1% -1 ? -- | Lucas Nussbaum | lucas@xxxxxxxxxxxxxxxxxx http://www.lucas-nussbaum.net/ | | jabber: lucas@xxxxxxxxxxx GPG: 1024D/023B3F4F | - To unsubscribe from this list: send the line "unsubscribe linux-acpi" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
