On 2020-08-08 14:25, David Holmes wrote:
Seems reasonable - at least until jtreg is fixed.
I agree; when jtreg is fixed we should revert this, since it's writing
output that's not strictly needed, something I've been trying hard to fight.
But until then: looks good.
/Magnus
Where does the hs_err file from a crashing VM end up?
Thanks,
David
On 8/08/2020 7:11 am, Mikael Vidstedt wrote:
Please review this small change which adds a sanity check run of the
JDK under test before running jtreg tests.
JBS: https://bugs.openjdk.java.net/browse/JDK-8251316
<https://bugs.openjdk.java.net/browse/JDK-8251316>
webrev:
https://cr.openjdk.java.net/~mikael/webrevs/8251316/webrev.00/open/webrev
<https://cr.openjdk.java.net/~mikael/webrevs/8251316/webrev.00/open/webrev>
* Background (from JBS)
If the JDK under test is severely broken, if for example it crashes
on startup, jtreg will typically just report "Error: cannot determine
version for JDK: /path/to/bin/java" without any additional
information. To make it faster to diagnose the problem it would be
useful to do some sanity checking of the JDK before invoking jtreg.
* Testing
I verified locally that a crashing VM does produce the expected
output both on the console and in the sanity.log log file
Running tier1 now for good luck.
Cheers,
Mikael
