On Wed, 1 Feb 2023 23:51:02 GMT, Chris Plummer <[email protected]> wrote:

> This test started failing after 
> [JDK-8300811](https://bugs.openjdk.org/browse/JDK-8300811), which made the 
> jdb ThreadStartRequest and ThreadDeathRequest use SUSPEND_NONE instead of 
> SUSPEND_ALL. This sped up the handling of these events a lot, which has 
> impacted the timing of this test and exposed a bug.
> 
> The 2nd (last) breakpoint this test sets up uses SUSPEND_NONE. The debuggee 
> quickly exits after hitting the breakpoint because it is not suspended. This 
> puts the test at risk of having the debuggee exit before jdb can produce the 
> full output for the breakpoint that includes the line number information. The 
> test is looking for this output When it is not there, the test fails. 
> Although the risk was always there, it was made worse by speeding up the 
> handling of the ThreadDeath event. This is why this issue is also only seen 
> on product builds (jdb needs to be running fast).
> 
> This failure is somewhat noisy so I'd like to push it quickly, but it should 
> still have two reviews.

Marked as reviewed by amenkov (Reviewer).

-------------

PR: https://git.openjdk.org/jdk/pull/12378

Reply via email to