RFR: 8337410: The makefiles should set problemlist and adjust timeout basing on the given VM flags

[Magnus Ihse Bursie]

On Thu, 1 Aug 2024 18:26:55 GMT, Leonid Mesnik <lmesnik at openjdk.org> wrote:

> There jtreg tests have several additional problemlists 
> ProblemList-Xcomp.txt
> ProblemList-generational-zgc.txt
> ProblemList-zgc.txt
> Each of them is bound to corresponding execution mode (Xcomp/ZGC) and it makes sense to treat them like standard problemlist when tests are executed with -Xcomp or ZGC enabled.
> Currently, it is needed to set them manually and it is often forgotten. So engineers waste time analyzing known failures.
> 
> Additionally, the **default** timeoutFactor is increased when Xcomp is enabled because of slowness of this mode. 
> 
> The jtreg allows to add the same problemlist twice so it is not needed to update any execution system that set problemlists.
> 
> Later it might makes sens to set 'JTREG_ALL_OPTIONS' by asking java about actually set mode. So it is possible to adjust  options for fastdebug/slowdebug/product modes and/or different options that are set during compilation (saying different default GC).

make/RunTests.gmk line 844:

> 842: 
> 843:   ifneq ($$(findstring -Xcomp, $$(JTREG_ALL_OPTIONS)), )
> 844:     JTREG_AUTO_PROBLEM_LISTS := $$(JTREG_AUTO_PROBLEM_LISTS) ProblemList-Xcomp.txt

Please use `+=` instead.

make/RunTests.gmk line 845:

> 843:   ifneq ($$(findstring -Xcomp, $$(JTREG_ALL_OPTIONS)), )
> 844:     JTREG_AUTO_PROBLEM_LISTS := $$(JTREG_AUTO_PROBLEM_LISTS) ProblemList-Xcomp.txt
> 845:     JTREG_AUTO_TIMEOUT_FACTOR ?= 10

Since you assign this to be 4 a couple of lines up, this code will never do anything.

make/RunTests.gmk line 849:

> 847: 
> 848:   ifneq ($$(findstring -XX:+UseZGC, $$(JTREG_ALL_OPTIONS)), )
> 849:     ifneq ($$(findstring -XX:-ZGenerational, $$(JTREG_ALL_OPTIONS)), )

Is this the only way that zgc can be run in a non-generational mode? I understand the idea to automatically pick the right set of problem list files, I just want to make sure it is robust.

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

PR Review Comment: https://git.openjdk.org/jdk/pull/20430#discussion_r1704323910
PR Review Comment: https://git.openjdk.org/jdk/pull/20430#discussion_r1704324902
PR Review Comment: https://git.openjdk.org/jdk/pull/20430#discussion_r1704328291