I still see a variant of the numlock problem in 2.1.30. If I press numlock
repeatedly my computer soon locks solid for a few seconds. If several
processes are running (xquake and the xserver are a good combination)
the lockup happens faster. In the logfile I get:
irqlock2 stuck at c012927e, waiting for c0112732
(System.map: kmalloc do_gettimeofday)
another one:
irqlock2 stuck at c01a7809, waiting for c011273
(System.map: update_timeout do_gettimeofday)
Y
In SCSI code (between scsi_reset and scsi_malloc)
My setup is:
- Gigabyte 586DX motherboard (Intel 82439HX Triton II (rev 1))
with 2xP100.
- Adaptec 2940 UW on board (driver aic7xxx-2.1.23-Feb12)
- PS/2 mouse (produces many interrupts)
- linux 2.1.30 of cours
- RedHat 4.1. system
Hmmm... I think thats all that is necessarry.
Matthias
By the way, there are some symbols that need to be exported to get modules
working, these are:
global_irq_holder, local_irq_count, waitqueue_lock, __global_sti,
__global_cli, __global_save_flags, __global_restore_flags, tqueue_lock