On Wed, 01 Jul 2009 21:29:10 +0200
Christian Franke wrote:
> My first assumption was probably wrong. The devices with 48-bit support
> I tested use a JM20336 or -9 (not -7). The SATA-only chips support it
> probably because there is no much between 28- and 48-bit commands in the
> SATA protocol.
> I changed the USB ID table in scsiata.cpp such that the SATA chips use
> '-d usbjmicron,x'.
> >> BTW: Would is make sense to fall back to '-l error/selftest' if only '-l
> >> xerror/xselftest' is specified but does not work?
> > I don't know, maybe, maybe not.
> > It can clutter up the output if the user does something like
> > smartctl -l selftest -l xselftest /dev/sda
> > I'd rather get a selftest log and a message that GPL is unsupported,
> > than two identical selftest logs. On the other hand, this is a highly
> > hypothetical situation. :)
> > Maybe just printing something like
> > "-l xselftest: GPL unsupported, please try -l selftest instead" would be ok too.
> Such a hint makes sense.
> Hmm.. or a new option like '-l x,selftest' or '-l xselftest,selftest'
> for "try xselftest first" ?
Well, -l xselftest,selftest sounds nice and self-explanatory (at least to me).
A hint is probably still ok for -l xselftest.
Same goes for xerror / error.
I remember you telling me something about the drives which only write
to xerror log but not error log. In these cases the "-a" option would
probably produce output without any error log at all, and the user would
think that there are no errors (I'm assuming it's just empty, I could be wrong...).
Maybe it would make sense to make "-a" use "-l xerror,error" (and maybe
even "-l xselftest,selftest")?
This may possibly break some third-party parsers though (but not
gsmartcontrol, which doesn't use -a).