Re: VFAT kernel considerations

[q-funk@megacom.net (Martin-Eric Racine)]

 >> Btw. I'm a lot more worried about the wb-cache, which is what has
 >> caused most data-loss under MagiC. Can anybody (Frank? Konrad?)
 >> confirm that "our" implementation is safer than MagiC's?
 >
 >What do you understand under safe? If you turn off the power before the 
 >cache is flushed you have always a inconsistent filesystem.

How much time before the cache is flushed and current write completed? 
Is there any "write complete" signal so Shutdown or N.Closure don't 
reboot the system until everything in the cache has been written to disk?

 >Same with minix filesystem (turbo mode = writeback mode).

Good to know.  Btw, where can I set the size of the Minix cache?

***

Generally speaking, the only drive that would require a cache (IMHO)
is A, because its access is always blocking.  Most modern hard-disks
already cache the data by design, so it should be fast enough.

What's really needed is a way to speed up drive A access, either by
rewriting the driver so it becomes non-block/background or by adding
a floppy-specific cache.

Ideas on how that could be done?


-----------------------------------------------------------------
From: Martin-Eric Racine FUNKYWARE http://www.megacom.net/~q-funk
-----------------------------------------------------------------
 Unsollicited material will be proof-read for $100 USD per word,
 payable within 10 days.  Submission of your advertisement to my
 address constitutes your formal acceptance of these terms.
-----------------------------------------------------------------
 <TITLE><BLINK><H1>No HTML Messages Please!</H1></BLINK></TITLE>