EQL bugfix

Dancer (dancer@brisnet.org.au)
Sun, 03 May 1998 03:14:27 +1000

Well, I had some finally had some success with the EQL driver, and
getting rid of the bug. I spent a couple hours at the machine hub and
rewrote the whole queue handling/scheduling code. It's messy, and ugly
and nasty, but it _works_. No more forgetting that slaved devices exist.

There's still some omissions of mine to correct (I forget to clear the
slave flag on devices, but that's a minor tweak). Also the scheduler is
_seriously_ stupid right now. It's just round-robinning the devices,
assuming they are all running at the same speed, and ignoring the actual
packet-sizes written to the devices.

However, the site admin there is in awe. It's reliable, and is handling
the load across three modems smoothly, and without hitches. It's been up
48 hours with no sign of problems. The other end is still running eql
1.9, but comparitively little data comes back that way. I'll be changing
the other end in a few days when there's time.

Once I've got it tidied up, and improved the scheduler I'll throw it at
people, and if it's ugliness doesn't turn them away, what say we fold it
in as the working version?


Version: 3.1
GAT d- s++: a C++++$ UL++++B+++S+++C++H++U++V+++$ P+++$ L+++ E-
W+++(--)$ N++ w++$>--- t+ 5++ X+() R+ tv b++++ DI+++ e- h-@ 

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu