Re: 2.5.69-70 ide-cd to guarantee fault-free CD/DVD burning experience?

Jens Axboe (axboe@suse.de)
Wed, 4 Jun 2003 21:55:46 +0200


On Wed, Jun 04 2003, Andy Polyakov wrote:
> > ... accept ... patch which makes it possible to access the
> > sense data returned by IDE CD/DVD units from user-land with SG_IO ioctl.
>
> The originally proposed modifications were indeed sufficient to get
> DVD+RW units working, but apparently not DVD-RW ones:-( Note though
> that [another] problem discussed here is not specific to DVD-RW
> recordings. It's generic bug/deficiency. Once a packet commands is
> terminated with an error condition the whole bio should be purged at
> once and not only the first chunk as it's currently implemented.
>
> Attached patch should be considered as a "denoting" patch, not
> "final." Well, because it was verified with single application,
> growisofs of dvd+rw-tools, which uses mmap-ed, in other words
> page-aligned, buffer(s). I mean I'm not 100% sure if hard_nr_sectors
> is appropriate even for general case of 4-byte aligned buffers...
> Then if-statement should probably be extended even to REQ_PC case...
>
> Cheers. A.
> 8<--------8<--------8<--------8<--------8<--------8<--------8<--------
> --- ./drivers/ide/ide-cd.c.orig Tue Jun 3 12:21:56 2003
> +++ ./drivers/ide/ide-cd.c Wed Jun 4 16:14:41 2003
> @@ -657,6 +657,9 @@
> struct request *rq = HWGROUP(drive)->rq;
> int nsectors = rq->hard_cur_sectors;
>
> + if (rq->flags&REQ_BLOCK_PC)
> + nsectors = rq->hard_nr_sectors; /* purge it all ... */
> + else
> if ((rq->flags & REQ_SENSE) && uptodate) {
> /*
> * For REQ_SENSE, "rq->buffer" points to the original failed

The *sector* values don't really work well for REQ_BLOCK_PC, it doesn't
even have to be set at all. One solution would be to add more rq members
(yuck), a nicer one is probably to make cdrom_end_request return
ide_end_request ret value, and simply make the error locations kill the
requests... It's not very nice, but should work.

-- 
Jens Axboe

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/