Re: Thread implementations...

Chris Wedgwood (chris@cybernet.co.nz)
Mon, 29 Jun 1998 12:42:54 +1200


> Well, I decided I might as well implement it. Appended is a stupid example
> program (and I mean _really_ stupid), and the diff against 2.1.107. It was
> pretty much exactly as I expected it to be: a small amount of judicious
> re-organization made the whole system call be less than 100 lines of code,
> and most of that is just checking the inputs to the system call rather
> than actual work.

Very cool...

> sendfile(int outfd, int infd, size_t size);

Can this API _PLEASE_ not be set in stone?

If were are going to invent APIs (which isn't necessarily a bad thing) then
why not make it compatible with something else.

In think case, I say the HP-UX 10.30 API which allows an iovec to be
appended or prepend to the data is very useful (not just because of the
syscall over head, but because it might save one TCP packet).

Objections?

-Chris

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