[lkml]   [2000]   [Nov]   [30]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: TCP push missing with writev()
On Thu, 30 Nov 2000, David S. Miller wrote:

> vector[ 1 ].iov_base = NULL;
> vector[ 1 ].iov_len = 0;
> "Don't do that" :-)
> I'll doubt I will bother to fix it (it's not %100 trivial to fix and
> it will put a cost into a hot path). Even if I did, you're going to
> need to fix Zeus not to do writev's with empty iovec's like this so
> that you get the PSH's in current kernels.

Its OK, our web server currently doesn't have this problem :)

I ran into the problem developing code that wrote data out of a circular
buffer - so the data being written may potentially be in two sections if
the buffer wraps around. A generic writev() call then emptied as much data
out of the buffer as possible, and so tripping the bug. I guess what I'm saying
is that you don't have to have plain dumb code (like the example code I
gave) in order to hit this problem.

Paranoia check: can this bug occur with writev() and other combinations of
sizes, or is it only when there is a 0-length buffer in there?


To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to
Please read the FAQ at

 \ /
  Last update: 2005-03-22 12:49    [W:0.048 / U:1.336 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site