Missing SCSI command in the allowed list?
While cloning an audio CD using cdrdao 1.1.9 with vanilla linux-2.6.9 as a user, I see the following "errors":
ERROR: Read buffer capacity failed.
This refers to the buffer in the CD drive. The indiation of the drive buffer fill is missing. The clone sounds like the original. The command used for cloning:
cdrdao copy --device /dev/hdd --source-device /dev/hdb \
--drivergeneric-mmc-raw --speed 8 --on-the-fly --paranoia-mode 1 -v 1 \
--fast-toc
If I run that as root, there are no such "errors", and levels of both the software cdrdao buffer and hardware buffer in the drive are indicated.
Looks like the following SCSI command is missing from the allowed list:
cmd[0] = 0x5c; // READ BUFFER CAPACITY
(according to cdrdao source, Sony CDU920 uses 0xec instead, but I don't have Sony drives here). Please fix the list or explain why this cannot be done.
Signature · 1 line
Alexander E. Patrakov
Hi,
Quoted 4 lines · Alexander E. Patrakov
While cloning an audio CD using cdrdao 1.1.9 with vanilla linux-2.6.9 as a user, I see the following "errors":
ERROR: Read buffer capacity failed.
I submitted a patch for this a few days ago. It has been merged into Linus's tree.
Daniel
Quoted 10 lines · Daniel Drake
Hi,
Quoted 4 lines · Alexander E. Patrakov
While cloning an audio CD using cdrdao 1.1.9 with vanilla linux-2.6.9 as a user, I see the following "errors":
ERROR: Read buffer capacity failed.
I submitted a patch for this a few days ago. It has been merged into Linus's tree.
Yes, I see the patch, thanks:
http://www.kernel.org/pub/linux/kernel/v2.6/testing/cset/cset-axboe%40suse.de[torvalds]
ChangeSet|20041104154725|45958.txt
But the question remains: what should the users of not 100% MMC-compatible CR-RW drives (i.e. those which have a separate cdrado or cdrecord driver, not generic-mmc/generic-mmc-raw) do? Is the support for writing as non-root on such drives just dropped without any plans to "fix" it?
Signature · 1 line
Alexander E. Patrakov
Hi,
Quoted 4 lines · Alexander E. Patrakov
But the question remains: what should the users of not 100% MMC-compatible CR-RW drives (i.e. those which have a separate cdrado or cdrecord driver, not generic-mmc/generic-mmc-raw) do? Is the support for writing as non-root on such drives just dropped without any plans to "fix" it?
I'd also be interested to know the answer here. Jens?
Some Gentoo users have reported that commands such as ED/EB/E9/F5 are being rejected. When inspecting the cdrecord source code, it seems that these are specific to plextor drives. These drives are MMC but have a few vendor-specific extensions. How should we go about permitting cases like this in the command filter?
Thanks,
Daniel
Quoted 5 lines · Daniel Drake
Some Gentoo users have reported that commands such as ED/EB/E9/F5 are being rejected. When inspecting the cdrecord source code, it seems that these are specific to plextor drives. These drives are MMC but have a few vendor-specific extensions. How should we go about permitting cases like this in the command filter?
Someone posted a very nice patch that added a little fs so you could adjust the tables from user space. I'd suggest that. Unfortunately its not been merged with the base kernel yet
Quoted 8 lines · Daniel Drake
Hi,
Quoted 4 lines · Alexander E. Patrakov
But the question remains: what should the users of not 100% MMC-compatible CR-RW drives (i.e. those which have a separate cdrado or cdrecord driver, not generic-mmc/generic-mmc-raw) do? Is the support for writing as non-root on such drives just dropped without any plans to "fix" it?
Quoted 7 lines
I'd also be interested to know the answer here. Jens?
Some Gentoo users have reported that commands such as ED/EB/E9/F5 are being rejected. When inspecting the cdrecord source code, it seems that these are specific to plextor drives. These drives are MMC but have a few vendor-specific extensions. How should we go about permitting cases like this in the command filter?
See Alans post, that's the only real way to deal with the situation. Right now we are stuck with half a solution (which is better than none or the 5% initial solution), it would still be nice to have it finished. Search the archives, there were several posts on this.
Signature · 1 line
Jens Axboe