RFR(xxs): 8205925: Print correct and more information about secondary errors.

Thomas Stüfe thomas.stuefe at gmail.com
Wed Jun 27 16:18:24 UTC 2018

Hi all,

could I ask you for reviews for this small improvement:


Daniel, when solving JDK-8205648, remarked on the lack of error
information for secondary errors, i.e. errors which occur during error

Currently we see:

[error occurred during error reporting (test secondary crash 1), id 0xb]

Which does not tell much, especially since the printed id is wrong, it
is the id of the primary error, not the secondary one. It would be
helpful to see actually what error occurred during error reporting.

This patch adds the ability to see the correct secondary error.

Looks like this, for secondary signals:

[error occurred during error reporting (test secondary crash 1), id
0x8, SIGSEGV (0xb) at pc=0x00007fb877815ca6]

and for secondary asserts:

[error occurred during error reporting (test secondary crash 1), id
0xe0000000, Internal Error


Implementation note: when printing information about the secondary
error, I follow closely the already existing STEP("printing
exception/signal name"). I avoided any unnecessary refactoring since I
wanted to get this thru the door before 11 closes. Note that
https://bugs.openjdk.java.net/browse/JDK-8196507 also touches these
areas and under that bug I have some cleanups for 12 planned.

Thanks & Regards, Thomas

More information about the hotspot-runtime-dev mailing list