On Sat, 9 May 2026 01:07:52 GMT, Chris Plummer <[email protected]> wrote:
>> The test sees the following output and is supposed to detect the "main[1]" >> prompt to indicate it is done with the "locals" command that was issued, and >> then issue a "cont" command: >> >> [9:13:15.40] Sending command: locals >> [9:13:15.560] reply[0]: Method arguments: >> [9:13:15.561] reply[1]: args = instance of java.lang.String[3] (id=669) >> [9:13:15.561] reply[2]: Local variables: >> [9:13:15.561] reply[3]: main[1] >> [9:13:15.561] Sending command: cont >> >> However, the output instead looks like this: >> >> [21:15:18.114] Sending command: locals >> [21:15:18.515] reply[0]: Method arguments: >> [21:15:18.515] reply[1]: args = instance of java.lang.String[3] (id=686) >> [21:15:18.515] reply[2]: Local variables: >> [21:15:18.515] Sending command: cont >> [21:15:18.716] reply[0]: main[1] > >> >> The JdbTest.findPrompt() code looks for a pattern of characters, followed by >> '[', then a number, then ']'. Unfortunately it matches the String[3] text >> you see in the output. Because of that the test thought the "locals" command >> had completed, and issued the "cont" command too soon, which gets the test >> out of sync. >> >> Apparently some tests have had this same issue before and a solution was >> already available. You just need to set compoundPromptIdent to the prompt >> that the test expects (sans the square brackets part). >> >> I also fixed a couple of comment typos I noticed in JdbTest while debugging >> this. >> >> Tested by running kill003 a couple hundred times on the failing platform and >> with the failing JVM args. >> >> --------- >> - [x] I confirm that I make this contribution in accordance with the >> [OpenJDK Interim AI Policy](https://openjdk.org/legal/ai). > > Chris Plummer has updated the pull request incrementally with two additional > commits since the last revision: > > - minor comment fix > - more test fixes Looks good modulo some nit and question. test/hotspot/jtreg/vmTestbase/nsk/jdb/kill/kill003/kill003.java line 80: > 78: > 79: // Make sure we stop in jdb when the NPE is thrown or rethrown. > 80: reply = jdb.receiveReplyFor(JdbCommand._catch + "all > java.lang.NullPointerException"); Q: I wonder if ever there is any need to check what is in the `reply`. The style of this file is that the `reply` is always set but is not checked. Then why do we need to ever set it. test/hotspot/jtreg/vmTestbase/nsk/jdb/kill/kill003/kill003.java line 100: > 98: reply = jdb.receiveReplyFor(JdbCommand.cont); > 99: > 100: // jdb got notified of the NPE for the kill command. continue > and jdb will get Nit: Suggestion: s/continue/Continue/ ------------- Marked as reviewed by sspitsyn (Reviewer). PR Review: https://git.openjdk.org/jdk/pull/31048#pullrequestreview-4280130975 PR Review Comment: https://git.openjdk.org/jdk/pull/31048#discussion_r3232969888 PR Review Comment: https://git.openjdk.org/jdk/pull/31048#discussion_r3232980789
