RFR: 8049411: Minimal VM build broken after gcId.cpp was added
stefan.karlsson at oracle.com
Mon Jul 7 10:58:31 UTC 2014
On 2014-07-07 13:01, David Holmes wrote:
> On 7/07/2014 6:52 PM, Stefan Karlsson wrote:
>> On 2014-07-07 10:43, David Holmes wrote:
>>> Looks okay, but how did the original change get through JPRT ???
>> I guess we don't build and run minimal in JPRT? We used to build the
>> kernel version on Windows, but I can't find any reference to minimal in
>> the jprt.properties.
> It is built as part of the embedded builds in a standard hotspot
> integration job. How was the problem detected? Nightly builds or test
> failures? It's conceivable the omission got through the build but
> caused a runtime link failure.
I was doing a minimal build before pushing the G1 Class Unloading
changes. The build went fine, but the JVM couldn't be started. I've
heard from Mikael Gerding that there is supposed to be a linker flag to
prevent this kind of failures from happening.
>>> On 7/07/2014 6:29 PM, Stefan Karlsson wrote:
>>>> Hi all,
>>>> Please, review this change to fix a build problem with the minimal VM.
>>>> When gcId.cpp was introduced, it wasn't added to the list of files to
>>>> keep in the gc_implementation/shared directory.
More information about the hotspot-dev