Re: cdrecord dev=/dev/hdd in 2.9.0-test9-mm2: lots (LOTS) of error messages

From: Valdis . Kletnieks
Date: Fri Nov 07 2003 - 19:18:35 EST


On Thu, 06 Nov 2003 14:43:14 +0100, Jurriaan <thunder7@xxxxxxxxx> said:
> I tried to burn a CD using this command-line in 2.6.0-test9-mm2:
>
> sudo cdrecord -v dev="/dev/hdd" -dao -useinfo *.wav
>
> Result:
>
> [this about 5 times per second]
> Nov 6 14:37:17 middle kernel: arq->state 4
> Nov 6 14:37:17 middle kernel: Badness in as_put_request at drivers/block/as-
iosched.c:1783

I'm seeing this as well burning an ISO in TAO mode on /dev/hdb, but slightly different traceback:

Nov 6 16:03:06 turing-police kernel: arq->state 4
Nov 6 16:03:07 turing-police kernel: Badness in as_put_request at drivers/block/as-iosched.c:1783
Nov 6 16:03:07 turing-police kernel: Call Trace:
Nov 6 16:03:07 turing-police kernel: [as_put_request+113/140] as_put_request+0x71/0x8c
Nov 6 16:03:07 turing-police kernel: [elv_put_request+19/23] elv_put_request+0x13/0x17
Nov 6 16:03:07 turing-police kernel: [__blk_put_request+91/133] __blk_put_request+0x5b/0x85
Nov 6 16:03:07 turing-police kernel: [blk_put_request+35/67] blk_put_request+0x23/0x43
Nov 6 16:03:07 turing-police kernel: [sg_io+955/1072] sg_io+0x3bb/0x430
Nov 6 16:03:07 turing-police kernel: [scsi_cmd_ioctl+520/1204] scsi_cmd_ioctl+0x208/0x4b4
Nov 6 16:03:07 turing-police kernel: [avc_has_perm+57/67] avc_has_perm+0x39/0x43
Nov 6 16:03:07 turing-police kernel: [__copy_from_user_ll+76/90] __copy_from_user_ll+0x4c/0x5a
Nov 6 16:03:07 turing-police kernel: [cdrom_ioctl+29/3404] cdrom_ioctl+0x1d/0xd4c
Nov 6 16:03:07 turing-police kernel: [write_chan+432/451] write_chan+0x1b0/0x1c3
Nov 6 16:03:07 turing-police kernel: [default_wake_function+0/24] default_wake_function+0x0/0x18
Nov 6 16:03:07 turing-police kernel: [selinux_file_permission+289/300] selinux_file_permission+0x121/0x12c
Nov 6 16:03:07 turing-police kernel: [idecd_ioctl+55/66] idecd_ioctl+0x37/0x42
Nov 6 16:03:07 turing-police kernel: [blkdev_ioctl+797/816] blkdev_ioctl+0x31d/0x330
Nov 6 16:03:07 turing-police kernel: [sys_ioctl+512/583] sys_ioctl+0x200/0x247
Nov 6 16:03:07 turing-police kernel: [syscall_call+7/11] syscall_call+0x7/0xb
Nov 6 16:03:07 turing-police kernel: [pfkey_xfrm_state2msg+686/2757] pfkey_xfrm_state2msg+0x2ae/0xac5

(pfkey_xfrm_state2msg()??? But I'm not doing ipsec - is the traceback on crack? ;)

In addition, I end up with a bad burn because *something* is managing to
starve off reading the ISO off disk and we get a buffer underrun.

Under earlier kernels (-test7 or so), the input buffer stayed around 90+%,
here it would start off near empty, get up to about 50%, then slowly go down
till it hit zero, got an underrun, and croaked. One CD got 9M in, another
got 38M in. Decided to wait till I had more blanks handy before debugging
more. (I admit not knowing if the problem is the AS elevator, or
Con's swappiness patch).

Yes, the source disk hda and target CD/RW hdb are on the same IDE
controller, blame Dell.. ;) (though it's worked fine for a year till now).

Attachment: pgp00001.pgp
Description: PGP signature