kern/49046: udf_renamerace test randomly fails

["gson%gson.org@localhost"@gson.org (Andreas Gustafsson) (Andreas Gustafsson)]

>Number:         49046
>Category:       kern
>Synopsis:       udf_renamerace test randomly fails
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Mon Jul 28 14:55:00 +0000 2014
>Originator:     Andreas Gustafsson
>Release:        NetBSD-current >= 2013.08.14.11.29.57
>Organization:
>Environment:
System: NetBSD
Architecture: i386
Machine: i386
>Description:

The fs/vfs/t_renamerace:udf_renamerace test case is failing in about
half the runs the TNF testbed, with the error "create: Invalid
argument".  This happens on all three architectures being tested
(i386, amd64, sparc).

The first recorded such failure on the testbed is from CVS date
2013.08.14.11.29.57, on amd64.  The HTML report from that test run is
at:

  
http://releng.netbsd.org/b5reports/amd64/build/2013.08.14.11.29.57/test.html#fs_vfs_t_renamerace_udf_renamerace

For about a month before that date, the test was consistently passing
on the testbed.  It's not clear what triggered the transition, but one
candidate is the commit of rumpuser.c 1.54 on 2013.08.14.08.29.25,
which according to the commit message "might shake out some bugs from
the qemu test runs".

Intriguingly, the test was already failing for reinoud@ about a month
before it started failing on the TNF testbed, as reported in private
email on 2013-07-22, and also evidenced by the marking of the test as an
"expected failure" on 2013.07.10.18.55.00 (without a reference to a PR).

For completeness, here's the HTML report from a recent failure:

  
http://releng.netbsd.org/b5reports/amd64/build/2014.07.27.10.19.12/test.html#fs_vfs_t_renamerace_udf_renamerace

>How-To-Repeat:

Run the ATF tests multiple times.

>Fix: