RFR: JDK-8170878: JDK 9 fails to build when enabling Hotspot code coverage
Magnus Ihse Bursie
magnus.ihse.bursie at oracle.com
Tue Dec 13 11:27:38 UTC 2016
Looks good to me.
On 2016-12-09 14:37, Erik Joelsson wrote:
> The bundles and the jprt_bundles targets fail when enabling native
> code coverage in the build. Currently enabling code coverage is the
> only way to generate data for the "symbols" image (because we are
> still putting the actual debug symbols in the regular jdk/jre images).
> One of the problems fixed is to add prerequisites to the symbols-image
> target so that it doesn't fail when code coverage is enabled.
> The other, more serious problem was that when enabling code coverage,
> the symbols bundle is now constructed with files from two different
> base directories, but the recipe logic is only able to handle one base
> dir. I chose to fix the recipe logic to handle multiple base dirs. The
> recipe already had a couple of different paths/methods depending on
> the type of bundle to be built. Not all of them could easily be
> converted to support multiple roots, so I reordered them and adjusted
> the conditionals so that the most optimal method is chosen given the
> new valid constraint.
> Bug: https://bugs.openjdk.java.net/browse/JDK-8170878
> Webrev: http://cr.openjdk.java.net/~erikj/8170878/webrev.01/
More information about the build-dev