JDK-7095698 : jdk7u2 does not work as a bootstrap after 7091141
The Version table provides details related to the release that this issue/RFE will be addressed.
Unresolved : Release in which this issue/RFE will be addressed. Resolved: Release in which this issue/RFE has been resolved. Fixed : Release in which this issue/RFE has been fixed. The release containing this fix may be available for download as an Early Access Release or a General Availabitlity Release.
7u2-b08 cannot build itself (boot cycle issue)
See more detail in comments section.
This CR is resolved by backing out fix of # 7091141 from 7u2.
7091141 jaxp 1.4.5 update 1 integration has been backed out. Please verify.
We need to understand what's the difference between running a single build (which does not fail) and build in a loop. What does the build process do differently?
How about "jdk7u2b08 -jaxp"? This way we could isolate non-jaxp issues causing this failure and also set up a starting point for regression?
I did some bisection and narrowed it to this changeset in the jaxp repo:
date: Fri Sep 16 12:40:34 2011 -0700
summary: 7091141: JAXP 1.4.5 update 1 for 7u2
I tried these variants:
jdk7u2-b09 fails as bootdir
jdk7u2-b08 fails as bootdir
jdk7u2-b07 works as bootdir
jdk7u2-b09 - jaxp works as bootdir
The last one is jdk7u2-b09 with 239:db8ccf2cbb33 backed out from the jaxp repo. So 7091141 (jaxp 1.4.5 update 1) triggered the failure. It may also be the cause of the failure, but that's not certain.
Does it happen if the build is run with a prior drop of JAXP? Is JAXP changeset triggering this issue or the issue is with JAXP? Could it be that the binary dependencies of JAXP that we are pulling from Apache are compiled with a different VM?
Not an RE script failure, but a problem in the openjdk hotspot builds themselves. Assigning to John Coomes, who can transfer it to an appropriate jvm/hotspot category if need be.