mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
* Re: Is the linux networking code broken?
       [not found] ` <20020531024623.K4707-101000@ranger.pns.bellsouth.net>
@ 2002-05-31  8:45   ` Alex Riesen
  0 siblings, 0 replies; 2+ messages in thread
From: Alex Riesen @ 2002-05-31  8:45 UTC (permalink / raw)
  To: erc; +Cc: linux-kernel

On Fri, May 31, 2002 at 02:47:21AM -0500, Ed Carp wrote:
> On Fri, 31 May 2002, Alex Riesen wrote:
> 
> > Neither of them. How do you think the things are working here? 8-)
> > rsync, wget, apache, a whole lot other programs transfer megabytes.
> 
> Most of them use short writes.  Here's the source I'm using...

Most of them are the biggest possible writes.

~/compile/ercftp ./sender localhost ../nfs-kaboom.c
Using 2048 byte packets
Connecting to localhost
Resolved, opening TCP socket
Doing connect
Connected, waiting for OK...
Waiting for OK...
Sending ../nfs-kaboom.c
... a while
Waiting for ack...
caught signal 14!


~/compile/ercftp ./receiver 
Getting TCP socket
Doing bind
Doing listen
Accepting connection
Writing 23987 to ../nfs-kaboom.c
... another while

caught signal 14! flen=23987, inlen=0, pos=0, packets=12, packet.seqno=0


Neither of the programs finished because of awful lot of bugs within,
but the output clearly shows the facts.
Surely, no problems with networking observed.

What kernel do you use? Also, you may want check your system libraries
(libc, libresolv*).

-alex

^ permalink raw reply	[flat|nested] 2+ messages in thread

* Is the linux networking code broken?
@ 2002-05-31  0:12 Ed Carp
  0 siblings, 0 replies; 2+ messages in thread
From: Ed Carp @ 2002-05-31  0:12 UTC (permalink / raw)
  To: linux-kernel

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Or have I been smoking too much crack? ;)

I've tried this code on several different linux machines, running 2.0.36 -
2.4.7 kernels, and it always acts the same.  Doing a write() of more than
1024 bytes on a TCP socket always disables all network connectivity on the
sending machine.  (A reboot fixes things, naturally).  I've tried the same
code on FreeBSD 4.5 machines with no problems.  Is this a limitation in
the kernel networking code, in the IP fragmentation layer, or what?  I'm
scratching my head here...thanks in advance for any replies.  Off-list, if
you would be so kind...
- -- 
Ed Carp, N7EKG          http://www.pobox.com/~erc               214/986-5870
Director, Software Development
Escapade Server-Side Scripting Engine Development Team
Pensacola - Dallas - London - Dresden
http://www.squishedmosquito.com


-----BEGIN PGP SIGNATURE-----
Version: PGP 6.5.8
Comment: Made with pgp4pine 1.76

iQA/AwUBPPax+EbhwAGg7YRjEQK6ogCcCldlimuRlXlHMp5ltTgCycE3B7EAoNwo
QunzakQoiAMJyaWm8qyZi1RB
=qbSK
-----END PGP SIGNATURE-----



^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2002-05-31  8:46 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
     [not found] <20020531064959.GD1402@riesen-pc.gr05.synopsys.com>
     [not found] ` <20020531024623.K4707-101000@ranger.pns.bellsouth.net>
2002-05-31  8:45   ` Is the linux networking code broken? Alex Riesen
2002-05-31  0:12 Ed Carp

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®