[PATCH] 2.5.14 IDE CD-ROM PIO mode

[PATCH] 2.5.14 IDE CD-ROM PIO mode

[Osamu Tomita]

I could not use CD-ROM drive [PIO mode] on 2.5.14.
Error messeges are follows.

May  7 11:05:03 sarah kernel: hdc: cdrom_read_intr: data underrun (4294967292 blocks)
May  7 11:05:03 sarah kernel: end_request: I/O error, dev 16:00, sector 92
May  7 11:05:03 sarah kernel: Buffer I/O error on device ide1(22,0), logical block 22

This patch works fine for me. Please check it.

--- linux/drivers/ide/ide-cd.c.orig	Mon May  6 12:38:01 2002
+++ linux/drivers/ide/ide-cd.c	Tue May  7 16:43:51 2002
@@ -962,7 +962,7 @@
 
 	/* First, figure out if we need to bit-bucket
 	   any of the leading sectors. */
-	nskip = MIN(rq->current_nr_sectors - bio_sectors(rq->bio), sectors_to_transfer);
+	nskip = MIN((int)(rq->current_nr_sectors - bio_sectors(rq->bio)), sectors_to_transfer);
 
 	while (nskip > 0) {
 		/* We need to throw away a sector. */
-- 
Osamu Tomita
E-mail: tomita@cinet.co.jp

Re: [PATCH] 2.5.14 IDE CD-ROM PIO mode

[Martin Dalecki]

Uz.ytkownik Osamu Tomita napisa?:
> I could not use CD-ROM drive [PIO mode] on 2.5.14.
> Error messeges are follows.
> 
> May  7 11:05:03 sarah kernel: hdc: cdrom_read_intr: data underrun (4294967292 blocks)
> May  7 11:05:03 sarah kernel: end_request: I/O error, dev 16:00, sector 92
> May  7 11:05:03 sarah kernel: Buffer I/O error on device ide1(22,0), logical block 22
> 
> This patch works fine for me. Please check it.

Indeed this makes very much of sense. I will take it.
How did you see it "hawkeye"?

> --- linux/drivers/ide/ide-cd.c.orig	Mon May  6 12:38:01 2002
> +++ linux/drivers/ide/ide-cd.c	Tue May  7 16:43:51 2002
> @@ -962,7 +962,7 @@
>  
>  	/* First, figure out if we need to bit-bucket
>  	   any of the leading sectors. */
> -	nskip = MIN(rq->current_nr_sectors - bio_sectors(rq->bio), sectors_to_transfer);
> +	nskip = MIN((int)(rq->current_nr_sectors - bio_sectors(rq->bio)), sectors_to_transfer);
>  
>  	while (nskip > 0) {
>  		/* We need to throw away a sector. */

Re: [PATCH] 2.5.14 IDE CD-ROM PIO mode

[Martin Dalecki]

Uz.ytkownik Christoph Hellwig napisa?:
> On Tue, May 07, 2002 at 11:57:20AM +0200, Martin Dalecki wrote:
> 
>>>@@ -962,7 +962,7 @@
>>> 
>>> 	/* First, figure out if we need to bit-bucket
>>> 	   any of the leading sectors. */
>>>-	nskip = MIN(rq->current_nr_sectors - bio_sectors(rq->bio), sectors_to_transfer);
>>>+	nskip = MIN((int)(rq->current_nr_sectors - bio_sectors(rq->bio)), sectors_to_transfer);
>>
> 
> What about a s/MIN/min/g in the idea driver to easily catch such bugs?

Good idea partially already implemented :-).
At least the generic code and the host chip driver code are alread
switched to using those "chatch them" macros.

Re: [PATCH] 2.5.14 IDE CD-ROM PIO mode

[Christoph Hellwig]

On Tue, May 07, 2002 at 11:57:20AM +0200, Martin Dalecki wrote:
> > @@ -962,7 +962,7 @@
> >  
> >  	/* First, figure out if we need to bit-bucket
> >  	   any of the leading sectors. */
> > -	nskip = MIN(rq->current_nr_sectors - bio_sectors(rq->bio), sectors_to_transfer);
> > +	nskip = MIN((int)(rq->current_nr_sectors - bio_sectors(rq->bio)), sectors_to_transfer);

What about a s/MIN/min/g in the idea driver to easily catch such bugs?

Re: [PATCH] 2.5.14 IDE CD-ROM PIO mode

[Martin Dalecki]

Uz.ytkownik Bartlomiej Zolnierkiewicz napisa?:
> while we are here, I think that drivers shouldn't take
> requests < hardsect_size, it should be handled by block layer

There could be clever beasts which do (over)buffering...
like ide-cdrom indeed does.

Re: [PATCH] 2.5.14 IDE CD-ROM PIO mode

[Bartlomiej Zolnierkiewicz]

while we are here, I think that drivers shouldn't take
requests < hardsect_size, it should be handled by block layer

requests On Tue, 7 May 2002, Martin Dalecki wrote:

> Uz.ytkownik Christoph Hellwig napisa?:
> > On Tue, May 07, 2002 at 11:57:20AM +0200, Martin Dalecki wrote:
> >
> >>>@@ -962,7 +962,7 @@
> >>>
> >>> 	/* First, figure out if we need to bit-bucket
> >>> 	   any of the leading sectors. */
> >>>-	nskip = MIN(rq->current_nr_sectors - bio_sectors(rq->bio), sectors_to_transfer);
> >>>+	nskip = MIN((int)(rq->current_nr_sectors - bio_sectors(rq->bio)), sectors_to_transfer);
> >>
> >
> > What about a s/MIN/min/g in the idea driver to easily catch such bugs?
>
> Good idea partially already implemented :-).
> At least the generic code and the host chip driver code are alread
> switched to using those "chatch them" macros.