Comment 1:

I guess this is my fault, but I don't know why.
http://build.golang.org/ looks fine.
Which version of FreeBSD? (if that matters... maybe?)

Owner changed to @bradfitz.

Status changed to Accepted.

Comment 2 by jgrahamc:

I don't know if the FreeBSD version matters but I am running:
$ uname -a
FreeBSD freebsd 9.0-RELEASE FreeBSD 9.0-RELEASE #0: Tue Jan  3 07:46:30 UTC 2012    
root@farrell.cse.buffalo.edu:/usr/obj/usr/src/sys/GENERIC  amd64
Happy to help debug if there's something useful I can do.

Comment 3:

I use exactly this kernel, but unfortunately, I can't reproduce the test failure.

Comment 4:

I haven't been able to reproduce this yet, could those on freebsd systems please try
running this script over your lunch break.
There was a prior commit,
https://code.google.com/p/go/source/detail?r=1315abc581ed1d8aa7f3e12f8cd5f250099198cc,
which should have fixed this problem.
% cat stress.bash 
#!/bin/bash
set -e
while true ; do 
        export GOMAXPROCS=$[ 1 + $[ RANDOM % 128 ]]
        go test std
done

Comment 6:

The test looks broken to me. It runs a single iteration, but there can always be minor
first-use noise. It should do something like 10 or 100 iterations and divide by the
count to throw away the noise. 
Changing the bug to be about adding a function for counting mallocs to package testing.
The TestCountMallocs in package fmt is a good template.
Then we should use the function in this test and others.

Labels changed: added priority-later, removed priority-triage.

Owner changed to ---.

Comment 7:

Labels changed: added size-m.

Comment 8:

Labels changed: added suggested.

Comment 9:

This issue was closed by revision 9bfd3c3.

Status changed to Fixed.