Switching to HotSpot 14 on OpenJDK6

[Andrew John Hughes]

2009/7/29 Daniel D. Daugherty <Daniel.Daugherty at sun.com>:
> Andrew John Hughes wrote:
>
> 2009/7/23 Mark Reinhold <mr at sun.com>:
>
>
> From: Mark Reinhold <mr at sun.com>
> Date: Wed, 22 Jul 2009 14:48:10 -0700
>
>
> ...
>
> Your repo looks fine to me -- thanks for creating it.
>
> Shall I go ahead and replace the current jdk6/hotspot repo with this?
> (That is, does anyone have any concerns?)
>
>
> Here's something (perhaps) to be concerned about: A recently-reported
> regression in Sun's proprietary 6u14 (and hence also in hsx14) causes
> debugging breakpoints to be ignored in some circumstances.  The bug is
> reproducible with both the client and server VMs, but only using the
> serial GC.
>
> The primary Sun bugid is 6862295 [1]; see also the downstream bug first
> reported in Eclipse [2].  Dan Daugherty is working on a fix, which we'll
> publish when available, but in the meantime hsx14 has this bug.
>
> Is the existence of this bug in hsx14 a reason to delay replacing the
> current jdk6/hotspot repo with Andrew's newer version?  Let me know.
>
> - Mark
>
>
> [1] http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=6862295
> [2] https://bugs.eclipse.org/bugs/show_bug.cgi?id=279137
>
>
>
> It's a concern, no doubt about that, but not moving to hs14 because of
> it is a bit like shutting the stable door after the horse has bolted;
> both IcedTea and Sun's proprietary releases have been shipping hs14
> for some time.  I think the better course of action is to get the
> patch in as soon as it is available and then roll it out in new
> releases.
>
> Besides, the proposed update doesn't (yet) move OpenJDK6 to HotSpot
> 14.  It rebases the repository so it's now at revision 0 of the
> current HotSpot7 repositories (plus the OpenJDK6 fixes), allowing
> changesets from the hs14 master and also the main HotSpot7 repository
> to be merged (as they all then have a common parent).  This is
> actually hs12b0 I believe (from b24 of OpenJDK7).
>
>
> The fix for 6862295 was pushed to the hotspot-rt repo for HSX-16
> last night; it should be integrated in HSX-16-B08. I have the
> same fix queued up for HSX-14.1 or HSX-14.2 depending on whether
> the fix is approved for 6u15 or 6u16.
>
> I also have the fix ready for the
> http://hg.openjdk.java.net/jdk6/jdk6/hotspot repo. The only odd
> thing I see is that the changeset ID is different for openjdk6
> version:
>
> changeset:   58:52f2c77b32e1
> tag:         tip
> user:        dcubed
> date:        Tue Jul 28 13:35:00 2009 -0600
> files:       src/share/vm/memory/genCollectedHeap.cpp
> src/share/vm/prims/jvmtiExport.cpp src/share/vm/runtime/globals.hpp
> description:
> 6862295: JDWP threadid changes during debugging session (leading to ingored
> breakpoints)
> Summary: Correctly count full GC operations for framework collectors. Add
> ForceFullGCJVMTIEpilogues as a future work around if needed.
> Reviewed-by: jcoomes, alanb, ysr
>
>
> Here is the changeset from the HSX-14.[12] and HSX-16 versions:
>
> changeset:   558:2b4230d1e589
> parent:      501:26bc4770e671
> user:        dcubed
> date:        Tue Jul 28 13:35:00 2009 -0600
> files:       src/share/vm/memory/genCollectedHeap.cpp
> src/share/vm/prims/jvmtiExport.cpp src/share/vm/runtime/globals.hpp
> description:
> 6862295: JDWP threadid changes during debugging session (leading to ingored
> breakpoints)
> Summary: Correctly count full GC operations for framework collectors. Add
> ForceFullGCJVMTIEpilogues as a future work around if needed.
> Reviewed-by: jcoomes, alanb, ysr
>
>
> I used "hg export" from the HSX-14.1 version to generate the
> patch for the openjdk6 version and I used "hg import" on the
> patch file. I expected the changeset ID to be the same.
>
> For the HSX-14.[12] and HSX-16 versions, I found the common
> ancestor changeset (501:26bc4770e671), made the changes
> relative to that changeset and merged the changes forward
> to each release's repo. The common changeset for HSX-14.1
> and openjdk6 goes all the way back to changeset 1. Ouch!
>
> Maybe there's a better way to do this...
>
> Dan
>
> P.S.
> What's the process for getting the fix into openjdk6?
>

Hi Dan,

As I mentioned in my last mail, OpenJDK6's repo is still as it was
before: at hs12b0.  It and the other HotSpot repositories now have a
common ancestor, but it's not yet a copy of hs14.
I plan to post a webrev moving it forward to hs14 (via a pull from
hs14/master).  It might be easier to just wait until this is applied,
or, if the patch you mention is already in hs14/master it will get
applied as part of this merge anyway.

Hope that helps,
-- 
Andrew :-)

Free Java Software Engineer
Red Hat, Inc. (http://www.redhat.com)

Support Free Java!
Contribute to GNU Classpath and the OpenJDK
http://www.gnu.org/software/classpath
http://openjdk.java.net

PGP Key: 94EFD9D8 (http://subkeys.pgp.net)
Fingerprint: F8EF F1EA 401E 2E60 15FA  7927 142C 2591 94EF D9D8