Bug 16510

Summary:	User processes locked on SMP box while burst packet receiving and tcpdumping
Product:	Base System	Reporter:	amagai <amagai>
Component:	kern	Assignee:	freebsd-bugs (Nobody) <bugs>
Status:	Closed FIXED
Severity:	Affects Only Me
Priority:	Normal
Version:	3.4-STABLE
Hardware:	Any
OS:	Any

Description amagai 2000-02-06 03:10:01 UTC

on HostS;  ping -f -l 5000000 -s 1 hostD >/dev/null
on HostD;  tcpdump >/dev/null

  No tcpdump running on hostD;
hostD reciving packets about 60000 Packets per Second (pps),
hostS sending  packets about 60000 pps
hostD CPU interrupt about 46 %  (by /usr/bin/top)

  while tcpdump running on hostD;
on HostD, user processes (netstat, top ...) are locked (or are eager for 
CPU time silce) .
hostS sending about 80000 pps!!
Kill the ping process on hostS, netstat resume, and reports many input packtes
received.

  So, while tcpdump (or packet duplicate in BPF routine), all CPU time 
slice may consume in interrupt routines.

  Without SMP configuration on hostD, user processes are not locked.

  No tcpdump running on hostD;
hostD reciving packets about 46000 pps
hostS sending  packets about 46000 pps
hostD CPU interrupt about 60 %

  while tcpdump running on hostD;
hostD reciving packets about 44000 pps
hostS sending  packets about 44000 pps
hostD CPU interrupt about 60 %

Fix: 

Sorry, I don't know.
How-To-Repeat: as above.

Comment 1 iedowse freebsd_committer

2001-08-27 21:39:27 UTC

State Changed
From-To: open->feedback


Can you reproduce this problem with newer releases?

Comment 2 amagai 2001-08-28 04:37:32 UTC

> Can you reproduce this problem with newer releases?
> http://www.FreeBSD.org/cgi/query-pr.cgi?pr=16510

Hi, I update Host D to FreeBSD 4.4-RC.
The problem is not occur on the releases.

Please close the PR. 
Thank you.

AMAGAI Yoshiji
NUE

Comment 3 iedowse freebsd_committer

2001-09-26 20:35:08 UTC

State Changed
From-To: feedback->closed


Submitter says that this problem is not reproducable with 4.4-RC.