Subject: Re: Busspace sanity ...
To: None <grefen@hprc.tandem.com>
From: John Nemeth <jnemeth@cue.bc.ca>
List: port-i386
Date: 06/13/1998 19:44:36
On Jun 13,  1:33pm, Stefan Grefen wrote:
} In message <199806130859.BAA27553@cue.bc.ca>  John Nemeth wrote:
} > On Jun 10, 12:41am, Stefan Grefen wrote:
} > } In message <199806092202.PAA10864@lestat.nas.nasa.gov>  Jason Thorpe wrote:
} > } > On Tue, 09 Jun 1998 23:25:15 +0200 
} > } >  Stefan Grefen <grefen@hprc.tandem.com> wrote:
} > } > 
} > } >  > I aggree, but please with a additional switch. This is helpful for
} > } >  > driver development but not for daily use.
} > } > 
} > } > The DEBUG option isn't really "for daily use", either...  I guess you
} > } > and I disagree on the semantics of the DEBUG option.
} > } 
} > } The problem is the message goes away without DEBUG, but a 'fixed'
} > } driver stays. I can add a ifdef DEBUG in the driver around the 
} > 
} >      The message goes away when the driver is fixed too.
} 
} It is no fixed. This a copyout to the card. I've read it any way the
} way it is. All I con do is go to a slower copyout. It is the final use
} of the buffer so aligning doesn't buy a thing.

     In this case, I would agree with you.  The damage has already
been done and there is nothing you can do about it.  The upper layers
should have made sure the data (or buffer as the case may be) was
aligned before handing it to you.

     Anyways, my comments weren't about your CS8900 driver in
particuler, they were meant to be general comments on all data used
anywhere in the kernel.  There may be good reasons for having some
data not aligned, but in most cases it should be.

} > } You stop doing that if you play with bigger machines in your day-time job.
} > 
} >      This is very unfortunate.  I think about performance issues all
} > the time.  Just because you have a big machine doesn't mean you can
} > get sloppy.  Bigger machine generally have bigger problems to solve,
} > and being sloppy means you need a much bigger machine.
} 
} Depending on what you do, concentrating on elimiating latency is more
} important, you can always buy a faster machine if your short on CPU-cycles.
} (That doesn't i don't care about them, just that the importance of latency
} issues is much higher).
} Or as a well known performace guru said:
} You can buy bandwidth, but latency is here to stay ....

     Misaligned data will cause you problems with latency (i.e.
causing the processor to do two memory reads instead of one).

}-- End of excerpt from Stefan Grefen