Shouldn’t test/sun/management/jmxremote/bootstrap/solaris-sparc/launcher be removed as part of this change?
/Staffan On 18 Nov 2013, at 11:09, Jaroslav Bachorik <jaroslav.bacho...@oracle.com> wrote: > Could I get this reviewed, please? > > -JB- > > On 4.11.2013 14:07, Jaroslav Bachorik wrote: >> Please, review the following test change: >> >> Issue: https://bugs.openjdk.java.net/browse/JDK-8027163 >> Webrev: http://cr.openjdk.java.net/~jbachorik/8027163/webrev.00/ >> >> Currently, the test is designed to work only on solaris-32bit. With the >> support removed it is necessary to adapt the test as well. This includes >> generating the solaris-sparcv9 and solaris-amd64 binaries for the >> launcher the test is using. >> >> The test itself needs changing the way it retrieves the ARCH name for >> sparcv9 architecture. Also, since it is forbidden to check in any files >> with executable permissions the test must account for this and adjust >> the file permissions temporarily while executing. It needs to revert to >> the original permissions when it's finished to prevent reporting the >> launcher files as modified in HG. >> >> While testing the changes the >> sun/management/jmxremote/bootstrap/LocalManagementTest.java kept failing >> intermittently (50%) when run in agentvm mode together with >> CustomLauncherTest.java on sparcv9 machines. Forcing both of these tests >> to run in othervm mode seems to fix the problem. >> >> There is a small debugging improvement in >> ProcessTools.java#startProcess() method - when the phaser timesout or is >> interrupted all the stack traces are dumped to stderr to help diagnosing >> any locking situations. >> >> Thanks, >> >> -JB- >
