1 |
-----BEGIN PGP SIGNED MESSAGE----- |
2 |
Hash: SHA1 |
3 |
|
4 |
On Wednesday 28 August 2002 02:09, Stuart Bouyer wrote: |
5 |
|
6 |
> Do you have scsi-cdrom support in the kernel or as a module? Otherwise |
7 |
> as normal user I can't use cdrecord as it looks for sg0. |
8 |
> |
9 |
> from my kernel .config |
10 |
> # |
11 |
> # SCSI support |
12 |
> # |
13 |
> CONFIG_SCSI=y |
14 |
> CONFIG_BLK_DEV_SD=y |
15 |
> CONFIG_SD_EXTRA_DEVS=40 |
16 |
> # CONFIG_CHR_DEV_ST is not set |
17 |
> # CONFIG_CHR_DEV_OSST is not set |
18 |
> CONFIG_BLK_DEV_SR=y |
19 |
> CONFIG_BLK_DEV_SR_VENDOR=y |
20 |
> CONFIG_SR_EXTRA_DEVS=2 |
21 |
> CONFIG_CHR_DEV_SG=y |
22 |
> CONFIG_SCSI_DEBUG_QUEUES=y |
23 |
> CONFIG_SCSI_MULTI_LUN=y |
24 |
> CONFIG_SCSI_CONSTANTS=y |
25 |
> # CONFIG_SCSI_LOGGING is not set |
26 |
> |
27 |
> Another thing to try is check in /lib/dev-state, where devfsd keeps all |
28 |
> its permissions and make sure that it isn't being set there. Not sure |
29 |
> why it changes after a SIGHUP, but not a boot time though |
30 |
|
31 |
Plot thickens |
32 |
|
33 |
I have it as a module: |
34 |
CONFIG_BLK_DEV_SR=m |
35 |
CONFIG_CHR_DEV_SG=m |
36 |
|
37 |
Init mouts devfsd long before modules gets loaded, I'll guess this is culprit. |
38 |
|
39 |
dev-state as only cd entry no generic |
40 |
|
41 |
Workaround is to add kill -HUP 'pidof devfsd` to last steps of default init |
42 |
sequence, fix is to recompile kernel an reduce these module thingys. |
43 |
|
44 |
Where on earth have I got this idea that modules should be used when ever |
45 |
possible. Thank you for your help. |
46 |
- -- |
47 |
Linux cartman.linux.home 2.4.19-gentoo-r7 #1 Fri Jul 5 00:45:50 EEST 2002 i686 |
48 |
AuthenticAMD |
49 |
-----BEGIN PGP SIGNATURE----- |
50 |
Version: GnuPG v1.0.7 (GNU/Linux) |
51 |
|
52 |
iD8DBQE9bHmZ6BYkNNMD2uARAm51AJ44IWHpUMtJLBHakFliua2upWsHZQCfZfci |
53 |
CX+9GOw7HDT/UEA8D1l8MIY= |
54 |
=pwsi |
55 |
-----END PGP SIGNATURE----- |