[Coco] My Drivewire Problems Are Solved! Happy Happy Joy Joy

[Aaron Wolfe]

There is a 200ms timeout implemented on the coco side, if the server
doesn't respond within this time then the coco considers the operation to
have failed.  The Drivewire server I/O thread asks for the sector and gets
suspended while the OS wakes up the drive to retrieve it, if this takes
longer than 200ms then we have a failure on the coco.  Then, the drive
finally wakes up and the server thread returns the data,  the server sends
a bunch of bytes the coco isn't expecting and more bad things could
happen.  At some point I added code to abort server side operations if they
exceed the timeout, but at best that just keeps one error from turning into
several.  There is really no mechanism the server can use to tell the coco
"hey, this is going to take awhile" in the current protocol.  Maybe we can
address that at some point.  In any case, glad it's working better now.
On Mar 24, 2015 7:39 PM, "K. Pruitt" <pruittk at roadrunner.com> wrote:

> <snip>
>
>>
>> But the  good news is, the BIOS update fixed whatever was conflicting
>> with Drivewire (Java really) and now drivewire boots in seconds, runs
>> smoothly, and appears very stable.
>>
>>
> I've had a chance to go over the release notes for the BIOS update I
> mentioned earlier and there were some patches to the routines that deal
> with sleep and hibernation modes. And it was drives coming out of sleep
> mode which took drivewire down 99% of the time. So it looks like that was
> where the source for my drivewire troubles originated.  Why the drives
> themselves were going to sleep when specifically set to stay on remains a
> mystery that I can't duplicate.
>
> I never even considered the BIOS as a possible source for my dw problems.
> It was a fluke that I even ran across this BIOS update.  Debugging via
> coincidence.  Ya gotta love it.
>
>
> --
> Coco mailing list
> Coco at maltedmedia.com
> https://pairlist5.pair.net/mailman/listinfo/coco
>