Port-amd64 archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: Does mfi write-back cache work for anyone?

On Thu, Jan 10, 2013 at 12:55:45AM -0500, Thor Lancelot Simon wrote:
> I think something still *must* be wrong with the driver -- here is what
> I see, including output from the Linux MegaCLI utility, which is pretty
> sure the battery is just fine:

Thinking about it -- it's almost as if the adapter firmware starts
up requiring confirmation from the OS that the battery is good and
that it should run the cache WB.  Like we're missing some kind of
acknowledgement or something.

Offhand, I don't see any code like this in the Linux driver, but I
will dig a little harder.  I'm also wondering whether it might be
feasible to reverse engineer a few of the battery related megacli
ioctls and just put them in the kernel.  By using the cache mode
override we could manage WB/WT on the cache "by hand" if we had
to, if we detect the adapter doing the wrong thing when the
battery is actually good.

Performance with this adapter is actually surprisingly good, if
I force the cache to WB mode.  In WT mode of course it's awful.


Home | Main Index | Thread Index | Old Index