The test is checking to make sure SA will use what is in the bcp register to determine the current line number rather than just use frame->bcp, which often is not up to date. It spawns a debuggee that sits in a loop that is 10 lines of source. The test executes jstack on the debuggee and determines the current line number from the output. It does this 25 times and expectesto see at least 5 of the 10 line numbers being hit. There's no guarantee it will hit at least five. The chances of this not happening are very very low, but apparently the test has failed a couple of times because only 4 unique lines were hit. I decided to do the simple fix and just lower the threshhold to 4 lines.
Ran locally on linux-x64. Tier1 in progress. ------------- Commit messages: - lower required number of line numbers from 5 to 4 Changes: https://git.openjdk.org/jdk/pull/22117/files Webrev: https://webrevs.openjdk.org/?repo=jdk&pr=22117&range=00 Issue: https://bugs.openjdk.org/browse/JDK-8253440 Stats: 8 lines in 1 file changed: 4 ins; 0 del; 4 mod Patch: https://git.openjdk.org/jdk/pull/22117.diff Fetch: git fetch https://git.openjdk.org/jdk.git pull/22117/head:pull/22117 PR: https://git.openjdk.org/jdk/pull/22117
