Re: yet more newbie questions

[Atle]

Richard wrote:
>
>   Good.  I think there are two independent factors explaining your
> following report:
> 
> > I did get a large number of
> >
> > hd0: read_intr: status=0x59 { DriveReady SeekComplete DataRequest Error }
> > hd0: read_intr: status=0x10 { SectorIdNotFound }, LBAsect=95049239, 
> > sector=28
> >
> > This pair of messages repeats a few times and then
> >
> > ide0: reset: success
> 
>   Note: GNUmach uses Linux's device drivers.  I've gotten errors like
> this before with Linux, and in each case the drive eventually gives
> out.  Don't trust me on this one, though: do you get these errors
> under Linux? 
I have even gotten them (semantically equivalent) under FreeBSD!
Now, my idea about this are based on two vague memories from MC'ing
around the /share, /doc, HOWTO and so on:
It seems there are two different Linux modules for IDE disks, one uses
an 'old and safe' and the other a 'new,improved,but-doesn't-always-work'
approach, and if you get errors that say 'read_intr: status=0xsth {
DriveRead Seek diddeli-doo-doo ... use the old one.

Now, I even have a theory :-)
I think it may have to do with PIO vs. [U,W]DMA modes and (maybe)
interupt latencies.

I think FreeBSD is very clear about what is does:

... timeout, trying to reset ...
... timeout, trying again ...
... no go, trying to _fallback_to_PIO_mode
... success!

This is to take with two pinches of salt, but I am quite sure there is
something to it.

Atle