> As the one who put in fast interrupts, specifically for the serial
> driver (I wanted a 40 MHz 386 to be able to handle multiple 115kbps
> serial transfers), I agree. SA_INTERRUPT served a purpose long
> ago, but it's probably best that it go away now.
Whilst nobody is going to deny it doesn't make sense on modern
machines, please be aware that it would still be really nice to be
able to use low-end 386 boxes for async. PPP, etc.
> However, can we please *not* make this change before 2.2 goes out?
> Once we start the 2.3 development effort, yes, let's nuke
> SA_INTERRUPT, and fix any device drivers that still use
> SA_INTERRUPT. But if we nuke it now, I'm afraid we may end up
> destablizing the tree somewhat.
I agree - but I'd also like 2.2.x to be able to function as a router
with 4 NICS without dropping packets on the floor when someone copies
a file from one machine to another...
... under normal circumstances, such a copy should have packets close
to MTU, so it won't be an issue, but with windows machines on a
network, it pretty hard to define normal.
> I'm *pretty* sure the serial driver will work fine with
> SA_INTERRUPT gone --- but would I bet a million dollars on that
> fact? Not without doing some serious testing, which I'm afraid we
> might not have time to do before 2.2 ships.
Even on old 16Mhz 386s?
-cw
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/