Messages in this thread | | | Subject | Re: > 15,000 Simultaneous Connections | Date | Wed, 15 Sep 1999 00:28:02 -0700 | From | Jason Nordwick <> |
| |
>Advantanges over sigwaitinfo(): > > - can dequeue multiple events in one call >
This is very useful to avoid context switches.
> - Queues would be automatically shared -- iff file descriptors are shared. >
The same can be said about signal queues.
> - Since there is no implicit per-process state (e.g. signal handlers), we > avoid the need for library and application code to fight it out over > who gets what signals. >
I don't see this as a big enough issue to outweigh the benefits. It is fairly easy to pick out a signal and use it. (It it more difficult to ensure that other things don't stomp on your signal handlers).
The benefits that I see of POSIX stuff is that there is already a precedent and API out there. It allows you to queue I/O requests and let the OS worry about shoving the data out the pipe.
Somethings that I think would be very nice additions is the ability to grab multiple events from the queue, aio_{read,write}v(), and aio_sendfile().
I think that an efficient Asynchronous I/O daemon (or handling algorithms) could beat out the userland data pushing.
> >gid >
-jason
- 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/
| |