Hi Jens,

I just hit a badness (actually, tons of badness like this) in as-iosched
while ripping
an audio CD with ripperX (with cdparanoia as a backend).
I was using 2.6.14 on an IBM Thinkpad R52. The kernel has been compiled with
gcc-4.0.2-2 (Debian testing).

The first badness in dmesg is:

cdrom: dropping to single frame dma
arq->state: 4
Badness in as_insert_request at drivers/block/as-iosched.c:1519
 [<c0237410>] as_insert_request+0x70/0x1d0
 [<c022dc25>] __elv_add_request+0xa5/0xe0
 [<c022dc8b>] elv_add_request+0x2b/0x40
 [<c0230fe6>] blk_execute_rq_nowait+0x46/0x60
 [<c023107a>] blk_execute_rq+0x7a/0xe0
 [<c0231310>] blk_end_sync_rq+0x0/0x30
 [<c0160b77>] bio_phys_segments+0x27/0x30
 [<c0232610>] blk_rq_bio_prep+0x40/0xb0
 [<c0230dc7>] blk_rq_map_user+0xb7/0xf0
 [<c026bc32>] cdrom_read_cdda_bpc+0x182/0x210
 [<c026bd1b>] cdrom_read_cdda+0x5b/0xc0
 [<c026d4b6>] mmc_ioctl+0x846/0xb20
 [<c0107bce>] timer_interrupt+0x3e/0x60
 [<c013affd>] handle_IRQ_event+0x3d/0x70
 [<c02352af>] scsi_cmd_ioctl+0xbf/0x510
 [<c01202f2>] __do_softirq+0x42/0xa0
 [<c0140f90>] buffered_rmqueue+0xc0/0x1c0
 [<c013f544>] generic_file_aio_write+0x84/0x110
 [<c026be66>] cdrom_ioctl+0xe6/0xe40
 [<c0146bdd>] lru_cache_add_active+0x2d/0x40
 [<c014c7b5>] do_anonymous_page+0x135/0x150
 [<c0252b75>] idecd_ioctl+0x85/0xa0
 [<c0233670>] blkdev_ioctl+0x140/0x1c0
 [<c016457b>] block_ioctl+0x2b/0x30
 [<c016e6a2>] do_ioctl+0x32/0x90
 [<c016e860>] vfs_ioctl+0x60/0x1f0
 [<c016ea35>] sys_ioctl+0x45/0x70
 [<c0103105>] syscall_call+0x7/0xb

It found similar reports in the archives
(http://lkml.org/lkml/2005/10/31/238
or http://lkml.org/lkml/2005/9/14/201) without any interesting conclusion.

_after_ reboot, hdparm looks like this:
/dev/hdc:
 IO_support   =  0 (default 16-bit)
 unmaskirq    =  0 (off)
 using_dma    =  0 (off)
 keepsettings =  0 (off)
 readonly     =  0 (off)
 readahead    = 256 (on)
 HDIO_GETGEO failed: Invalid argument

dmesg (without the badness flooding), lspci and .config are attached.

Regards,
Brice Goglin