I observed a timeout in the following test during my JDK7-B143
baseline test on Solaris X86:

    java/util/logging/LoggingDeadlock3.java

The fix for the following bug:

    7041595 4/4 add lost test for 6487638

added the test.

This test has a timeout value of 15 which is too short.

I ran the test in a loop on my Solaris X86 test machine:

1005 2011.05.25 11:50:56 $ uname -a
SunOS frankenputer 5.10 Generic_141445-09 i86pc i386 i86pc

1006 2011.05.25 11:54:31 $ psrinfo -v
Status of virtual processor 0 as of: 05/25/2011 11:54:35
  on-line since 05/06/2011 07:31:49.
  The i386 processor operates at 2800 MHz,
        and has an i387 compatible floating point processor.
Status of virtual processor 1 as of: 05/25/2011 11:54:35
  on-line since 05/06/2011 07:32:12.
  The i386 processor operates at 2800 MHz,
        and has an i387 compatible floating point processor.

I let the loop run for 25000 iterations (almost 46 hours)
while frankenputer did its normal work (mirror downloads,
repo updates, builds...)

Here are the stats:

ksh do_analysis.ksh
total sample cnt: 25001
min value: 0.97
max value: 18.91
avg value: 1.56
<= 5 seconds: 24674
<= 10 seconds: 289
> 10 seconds: 38

Normally I use 4X the average for a timout, but that
clearly won't work in this case.

I think I'm going to go with 4X the max and round up
to 80 seconds. That should account for running the
test on slower machines.

I think the default is 120 seconds so 80 is still better...