bad: scheduling while atomic!

2.5.50 NCR5380/PAS16 Paul 2 participants 3 messages Dec 3, 2002 – Dec 5, 2002

Share & tools

Paul

author

2002-12-03 01:33 UTC #0 lore raw

	Hi;

	This happens just before the scsi cdrom would be detected

during boot. The bad: part and the call trace are endlessly emited (exact same trace) and its pretty much game over. (fortunately I had a serial console machine handy.)
The driver is compiled into the kernel, not a module. The card (pas16) works under 2.4. I can test things
or supply more information/debugging if anyone wants.

Thanks;
Paul
[email protected]

SCSI subsystem driver Revision: 1.00
scsi-pas16 : probing io_port 0388
scsi-pas16 : detected board.
scsi-pas16 : io_port = 0388
scsi0 : irq = 10
scsi0 : at 0x0388 irq 10 options CAN_QUEUE=32 CMD_PER_LUN=2 release=3 generic options AUTOPROBE_IRQ AUTOSENSE PSEUDO DMA UNSAFE USLEEP, USLEEP_POLL=200 USLEEP_SLEEP=20 generic release=7 scsi-pas16 : probing io_port 0384
scsi-pas16 : probing io_port 038c
scsi-pas16 : probing io_port 0288
scsi-pas16 : io_port = 0000
scsi0 : Pro Audio Spectrum-16 SCSI
bad: scheduling while atomic!

Call Trace:
 [<c0112a85>] schedule+0x3d/0x2c0
 [<c012045c>] worker_thread+0x144/0x2cc
 [<c0120318>] worker_thread+0x0/0x2cc
 [<c025c790>] NCR5380_main+0x0/0x178
 [<c0112d48>] default_wake_function+0x0/0x2c
 [<c0112d48>] default_wake_function+0x0/0x2c
 [<c0108935>] kernel_thread_helper+0x5/0xc

Alan Cox

2002-12-03 15:36 UTC #1 lore raw

Quoted 9 lines · Paul

	Hi;

	This happens just before the scsi cdrom would be detected
during boot. The bad: part and the call trace are endlessly emited (exact same trace) and its pretty much game over. (fortunately I had a serial console machine handy.)
The driver is compiled into the kernel, not a module. The card (pas16) works under 2.4. I can test things
or supply more information/debugging if anyone wants.

Im using generic 5380 on 2.5.x but not 2.5.50 yet. Can you turn on all the NCR5380 debugging and see where it errors from (the trace doesnt give an answer as gcc has been smartly inlining functions used in one place only I suspect)

Paul

author

2002-12-05 03:16 UTC #2 lore raw

Quoted 6 lines · Alan Cox <[email protected]>, on Tue Dec 03, 2002 [04:17:51 PM] said:

Im using generic 5380 on 2.5.x but not 2.5.50 yet. Can you turn on all the NCR5380 debugging and see where it errors from (the trace doesnt give an answer as gcc has been smartly inlining functions used in one place only I suspect)

Hi;

Paul
[email protected]

SCSI subsystem driver Revision: 1.00
scsi-pas16 : probing io_port 0388
scsi-pas16 : detected board.
scsi-pas16 : io_port = 0388
scsi : NCR5380_all_init()
scsi0 : irq = 10
scsi0 : at 0x0388 irq 10 options CAN_QUEUE=32 CMD_PER_LUN=2 release=3 generic options AUTOPROBE_IRQ AUTOSENSE PSEUDO DMA UNSAFE USLEEP, USLEEP_POLL=200 USLEEP_SLEEP=20 generic release=7 scsi-pas16 : probing io_port 0384
scsi-pas16 : probing io_port 038c
scsi-pas16 : probing io_port 0288
scsi-pas16 : io_port = 0000
scsi0 : Pro Audio Spectrum-16 SCSI
LINE:1160 Adding c3f34200 to 00000000
scsi0 : command added to tail of queue
scsi0 : not connected
MAIN tmp=c3f34200 target=0 busy=0 lun=0
LINE:1232 Removing: ffffffff->c3f34200 c3f34200->00000000
scsi0 : main() : command for target 0 lun 0 removed from issue_queue
scsi0 : starting arbitration, id = 7
scsi0 : arbitration complete
scsi0 : won arbitration
scsi0 : selecting target 0
bad: scheduling while atomic!

Call Trace:
 [<c0112a85>] schedule+0x3d/0x2c0
 [<c012045c>] worker_thread+0x144/0x2cc
 [<c0120318>] worker_thread+0x0/0x2cc
 [<c025caac>] NCR5380_main+0x0/0x2dc
 [<c0112d48>] default_wake_function+0x0/0x2c
 [<c0112d48>] default_wake_function+0x0/0x2c
 [<c0108935>] kernel_thread_helper+0x5/0xc

bad: scheduling while atomic!

Call Trace:
 [<c0112a85>] schedule+0x3d/0x2c0
 [<c012045c>] worker_thread+0x144/0x2cc
 [<c0120318>] worker_thread+0x0/0x2cc
 [<c025caac>] NCR5380_main+0x0/0x2dc
 [<c0112d48>] default_wake_function+0x0/0x2c
 [<c0112d48>] default_wake_function+0x0/0x2c
 [<c0108935>] kernel_thread_helper+0x5/0xc

......