Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 19 Aug 2004 02:28:05 GMT
From:      Suihong Liang <s2liang@uwaterloo.ca>
To:        freebsd-gnats-submit@FreeBSD.org
Subject:   kern/70649: system clock slows down when heavily loaded
Message-ID:  <200408190228.i7J2S5fv066240@www.freebsd.org>
Resent-Message-ID: <200408190230.i7J2UN0k015908@freefall.freebsd.org>

next in thread | raw e-mail | index | archive | help

>Number:         70649
>Category:       kern
>Synopsis:       system clock slows down when heavily loaded
>Confidential:   no
>Severity:       serious
>Priority:       high
>Responsible:    freebsd-bugs
>State:          open
>Quarter:        
>Keywords:       
>Date-Required:
>Class:          sw-bug
>Submitter-Id:   current-users
>Arrival-Date:   Thu Aug 19 02:30:23 GMT 2004
>Closed-Date:
>Last-Modified:
>Originator:     Suihong Liang
>Release:        5.2.1
>Organization:
Univeristy of Waterloo
>Environment:
FreeBSD hurricane.cs.uwaterloo.ca 5.2.1-RELEASE FreeBSD 5.2.1-RELEASE #2: Fri Jul  2 16:25:54 GMT 2004     root@hurricane.cs.uwaterloo.ca:/usr/src/sys/i386/compile/ORIGIN  i386
>Description:
      It takes "SHORTER" time in scp a file while heavily loaded, in
fact, the receiver shows that it takes LONGER!

I think the kernel does not handle the system clock well when it is heavily loaded.

I've tried the built-in time, GNU time, and my simple home-brewed timer programs, same result was yielded.

>How-To-Repeat:
      1. cpu workload program can be a infinite loop program.
2. In my testing, I run 32 instances of such workload programs.
3. Use command "time" to time the scp in an ethernet environment. Do NOT trust the time reported by scp.
4. I used the same set of 30 files (each of 100MB).
5. reboot the sender after each set of scp is done.
>Fix:
      
>Release-Note:
>Audit-Trail:
>Unformatted:



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200408190228.i7J2S5fv066240>