lkml.org 
[lkml]   [1999]   [Mar]   [3]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectRe: OSDI paper - IO-Lite: A Unified I/O Buffering and Caching System
Date
> dense but if I understand the paper correctly from the (casual I
> admit) glance I had:
>
> 1. Most of the gains come from less copying along pipes.

Its worth noting Linus tried page tricks for pipe copying including
doing copy-on-write stuff briefly and the only thing it improved where
benchmarks

> 3. The gains come from copying by reference.

And you can do that with some kind of mmap() ring buffer protocol set
up in advance in user space and avoid continually remapping

> the peer is transparent. I.e. the kernel converts it back to a normal
> pipe on the fly. If that is the case couldn't this be also achieved by
> some hands hake between the two processes?

Perhaps what you need is a mix - user->user is done via user handshaking
protocols mostly (think about blocking I/O - there you want tiny bits
of kernel help). user<->socket with some kind of ring buffer protocol would
be the kernel talking it in that case


-
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/

\
 
 \ /
  Last update: 2005-03-22 13:50    [from the cache]
©2003-2011 Jasper Spaans