cdrecord and 2.5 kernels

[Markus Plail <linux-kernel@gitteundmarkus.de>]

Hi there!

I just compiled a few kernels and realized that 2.5.45 was the kernel
that broke writing DVD/CD-Rs for me. The messages at the bottom are
from 2.5.70 plus the second sg_io patch.

regards
Markus


Waiting for reader process to fill input buffer ... input buffer ready.
BURN-Free is ON.
cdrecord: WARNING: Drive returns wrong startsec (-150) using -11077 from ATIP
Writing lead-in at sector -11077
cdrecord: Success. write_g1: scsi sendcmd: no error
CDB:  2A 00 FF FF D7 AD 00 00 1A 00
status: 0x2 (CHECK CONDITION)
Sense Bytes: 70 00 05 00 00 00 00 0E 00 00 00 00 21 02 00 00
Sense Key: 0x5 Illegal Request, Segment 0
Sense Code: 0x21 Qual 0x02 (invalid address for write) Fru 0x0
Sense flags: Blk 0 (not valid) 
resid: 63648
cmd finished after 0.006s timeout 40s
write leadin data: error after 1845792 bytes
cdrecord: Could not write Lead-in.
Writing  time:    1.819s
cdrecord: fifo had 64 puts and 0 gets.
cdrecord: fifo was 0 times empty and 0 times full, min fill was 100%.
[23:34:34]-[Sat May 31]-[~]
[plail@plailis_lfs]

dmesg:
May 31 23:33:54 localhost kernel: hdb: confused, missing data
May 31 23:33:54 localhost kernel: hdb: confused, missing data
May 31 23:33:54 localhost kernel: cdrom_newpc_intr: 63648 residual after xfer
May 31 23:34:08 localhost kernel: hdb: DMA disabled
May 31 23:34:34 localhost kernel: hdb: confused, missing data
May 31 23:34:34 localhost kernel: hdb: confused, missing data
May 31 23:34:34 localhost kernel: cdrom_newpc_intr: 63648 residual after xfer

Note: It was me who disabled DMA to see if it makes a difference.