[7u6] Request for phase 2 approval for CR 7188168: 7071904 broke the DEBUG_BINARIES option on Linux

[Andrew Hughes]

----- Original Message -----
> Andrew,
> 
> Unfortunately this request comes very late in the 7u6 stabilization
> phase. Thorough testing & analysis is almost complete and we're
> preparing for release. 

Hmmm... was this announced somewhere?  All I see in my archives are the initial
announcement of the branch and I vaguely recall something about "August", nothing
more specific, but I may have missed something.

Is there a webpage that tracks the release process?  I don't see anything on
http://openjdk.java.net/projects/jdk7u/  It would be easier for us to work
to deadlines if we knew what they were :-)

> As previously communicated, only critical bug
> fixes are being considered. 

>From our perspective, this is critical.  Without this fixed, tools
can't work with the debug information on x86.

We can work around this by patching in IcedTea (we've actually already
applied it there) but we're working towards a scenario where we will be
able to use OpenJDK directly and avoid IcedTea.  If we're not able to get
fixes in during the latter stages of a release, then making such a switch
becomes unfeasible.

> With this issue not affecting the product
> JDK bits,

I'm not sure what you mean.  Can you explain?  This does affect what we ship.
A product build for us includes debug information.  This is why DEBUG_BINARIES
was added to begin with.

> can I ask that you push this fix to the jdk7u-dev repo
> instead.

Well, I was going to propose it for the main tree too, but I went for u6 first
as I was aware there was less time for that.  Apparently, there's even less than
I thought... :-)  Do I need to send a fresh mail for this or can I just push?

> 
> I'll create a 7u8 bug report for tracking purposes.
> 

We already have a bug ID from the 8 commit, if that's any help.

> Regards,
> Sean.
> 
> On 10/08/12 13:47, Andrew Hughes wrote:
> > Bug: http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=7188168
> > JDK8:
> > http://hg.openjdk.java.net/hsx/hotspot-rt/hotspot/rev/282abd0fd878
> >
> > Applies cleanly to 7u6.
> >
> > Original discussion:
> > http://mail.openjdk.java.net/pipermail/hotspot-dev/2012-July/006247.html
> > Original review:
> > http://mail.openjdk.java.net/pipermail/hotspot-runtime-dev/2012-August/004184.html
> >
> > Reasoning:
> >
> > This is a regression caused by the application of 7071904 to jdk7u.
> >  Prior to 7071904, a
> > build with DEBUG_BINARIES=true would result in binaries with DWARF
> > debug output, built using "-g".
> > After this change, bianries on platforms other than x86_64, ia64,
> > arm and ppc (such as x86 and
> > SPARC) are built using the STABS debug format.  This causes issues
> > with using the GNU/Linux
> > toolchain to debug/trace the JDK binaries e.g.
> >
> > https://bugzilla.redhat.com/show_bug.cgi?id=841057
> >
> > DWARF is the standard format used on GNU/Linux distributions, so
> > support for STABS is either
> > not as well tested or, in the case of newer tools, non-existent.
> >
> > Applying this fix shouldn't have any effect on builds which don't
> > use DEBUG_BINARIES (the change
> > just moves the existing code into the else block of a
> > DEBUG_BINARIES=true check), but will
> > allow us to build the JDK with the same debug information as we did
> > prior to 7071904.
> >
> > Thanks,
> 
> 

-- 
Andrew :)

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

PGP Key: 248BDC07 (https://keys.indymedia.org/)
Fingerprint = EC5A 1F5E C0AD 1D15 8F1F  8F91 3B96 A578 248B DC07