lkml.org 
[lkml]   [1998]   [Jun]   [26]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: Thread implementations...
   From: torvalds@transmeta.com (Linus Torvalds)
Date: 26 Jun 1998 09:20:46 GMT

That is just broken.

Not entirely...

If their system call overhead is so big that it makes sense to have
the iovec's be part of "sendfile()", they must be doing something
wrong.

This isn't it.

And if the system call overhead isn't that big, then they just made
sendfile() uglier for no good reason.

Neither is this...

Not that I've been very impressed by HP-UX before either. The
people working on it seem to have no taste (WNT is even worse in
this regard).

Well you pinhead, if you'd stop and think just one moment about what a
web server must do to send files back to the client, you'd see that
the iovec addition isn't so silly after all.

The problem is that you must send a http header up front (built in
userspace) and then tack the file data right afterwards. This is why
apache uses writev() at the moment in the first place.

It doesn't work to send the header first in a seperate small write()
call because then that goes out to the network in a tiny-gram before
the sendfile() call could even be made, they won't get coalesced into
full sized frames.

So please think twice before crucifying, oh mighty god... :-)

Later,
David S. Miller
davem@dm.cobaltmicro.com

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

\
 
 \ /
  Last update: 2005-03-22 13:43    [W:1.101 / U:0.504 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site