-----BEGIN PGP SIGNED MESSAGE-----

On Thu, 11 Jun 1998, Erik E. Fair wrote:

>My notion was to disable write cache *before* the last write...
>
>That may or may not *also* flush the cache. It's something we should ask
>Western Digital, Quantum, and Seagate...

And (admittedly a reasonably small) n other vendors  And assure that
this is true not just for all current drives but all past drives.  And
hope it remains true for all future drives.  Perhaps we should
have a quirks table with the capability to indicate whether a drive
should:

    a) not use write caching at all (i.e. the drive provides _no_ way
       to flush the cache)

    b) do synchronous write when syncing to flush async data (the drive
       behaves as you describe)

    c) use ATA-4 cache flushing (the drive is ATA-4 _and_ implements the
       optional flush command)

Just a thought...

- -- 
				Jim Wise
				jwise@unicast.com

-----BEGIN PGP SIGNATURE-----
Version: PGPfreeware 5.0i for non-commercial use
Charset: noconv

iQEVAwUBNYBH1xg+dMhCouwfAQECTQgAlwfTm1kotknMdqiIkP/ObVesi8uW8+Un
sKda/2+12fGRq3aKlVU1YNqwu5g5V2zCuIeIeHAtX/L5KbpdRro6Q1U4sWX80Kbj
PPbk69vbQr7RHTbFu/Z5+YnG7d7fK1CGsDzBj1SYBVf13KkxIFwx44vUai/idqTr
iSAbjnz64ZUzwW/onTd3Y2rZ4+ssRhUrN3gDsNc036okeckwgFNtNvHqB/o3DehV
JdPSY5MeitI9urWJ2YC3MDHJaAu0bdCwgtCs9Qqvwn5qfylNovA8GO3VA5PmsXHR
7ix9M41v+3xV7j1i9RTVxCcNpnCNh9tPoJmt8RBM/2sSm3/+vYDMpQ==
=1ftQ
-----END PGP SIGNATURE-----