Re: Efficient IPC mechanism on Linux

From: Luca Veraldi
Date: Wed Sep 10 2003 - 10:26:45 EST


> Can you forward-port your work to 2.6 kernel?
> Can you benchmarkt it against the same primitives in 2.6 kernel?

I could do it only if there was an effective reason to do it.
And there isn't.

They have just posted me a message with pipe latency under kernel 2.4.
And it is exactly the same (apart some minor variations in misurements
that are natural enough).

> You have just started your work - are you going to finish it? Or it
> was just-another-academical-study?

I consider my work finished.
I studied an efficient IPC mechanism and I tried to implement it
on an existing Operating System.

I tried some other IPC primitives in benchmark tests
and reported the comparisons between completion times.

I got my goal.

> If you can try to develop new sematics for old syntax (shm* & etc) it
> can be welcomed too.
> And if it would be poll()able - it would be great. Applications which
> do block on read()/getmsg() in real-life not that common, and as I've
> understood - this is the case for your message passing structure.

I'm not a kernel developer. Ask Linus Torvalds.

Bye,
Luca
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/