Re: Re: Re: PATCH for ide_floppy

["Manfred.Scherer.Mhm@t-online.de" <Manfred.Scherer.Mhm@t-online.de>]

yes, 
#define IDEFLOPPY_TICKS_DELAY  HZ/20
seems to be the solution.

when I've tested some values for IDEFLOPPY_TICKS_DELAY in December 2004,
I cannot found the best value for this. The Kernel version was 2.6.8
from the SuSE9.2 distribution.

I take a look in ide-cd.c and found there the function
cdrom_timer_expiry as
a possibility to handle timeout issues smoother. I tried this function
as 
idefloppy_timer_expiry in ide-floppy.c in combination with
IDEFLOPPY_TICKS_DELAY  60
as a best result for all cases. It seems to reach out to change
IDEFLOPPY_TICKS_DELAY
as suggested from you, idefloppy_timer_expiry is not really necessary.  


-----Original Message-----
Date: Fri,  1 Jul 2005 19:08:58 +0200
Subject: Re: Re: PATCH for ide_floppy
From: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
To: "Manfred.Scherer.Mhm@t-online.de" <Manfred.Scherer.Mhm@t-online.de>

On 7/1/05, Manfred.Scherer.Mhm@t-online.de
<Manfred.Scherer.Mhm@t-online.de> wrote:
> it's not really a performance issue but more a timeout issue.
> 'IDEFLOPPY_TICKS_DELAY  60' avoids error messages in /var/log/messages
> like 'reset ide ...'.
> Without the idefloppy_timer_expiry the data transfer to the ide-floppy
> is pending a long time between some transfer of data. The floppy LED
> indicated this too.
> With kernel 2.4.x I've never had this problem.

This seems related to 2.4 -> 2.6 HZ change.

> > @@ -317,7 +324,13 @@
> >         unsigned long flags;
> >  } idefloppy_floppy_t;
> >
> > +#if 0
> >  #define IDEFLOPPY_TICKS_DELAY  3       /* default delay for ZIP 100
> */
> > +#define IDEFLOPPY_TICKS_DELAY  6       /* default delay for ZIP 100
> > --ms 2005/01/01 */
> > +#define IDEFLOPPY_TICKS_DELAY  12      /* default delay for ZIP 100
> > --ms 2005/01/01 */
> > +#endif
> > +#define IDEFLOPPY_TICKS_DELAY  60      /* default delay for ZIP 100
> > --ms 2005/01/07 */
> > +

"ticks" delay should be expressed using HZ

#define IDEFLOPPY_TICKS_DELAY  HZ/20

for 50msec delay (N.B. the comment in the code about default delay
being 50msec also seems wrong - it was more like ~33msec in 2.4)

Could you please test if this fixes your problems?

Bartlomiej


-
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/