code review for second hotspot FDS gobjcopy work around (7165598)

Karen Kinnear karen.kinnear at
Thu May 24 14:28:46 UTC 2012


Code looks good.

Thank you for the work-around. 

So is this HDR_FLAGS workaround also needed in all the other jdk repositories?

And I assume the windows/makefiles/defs.make change is because you determined
that the failing tests actually pre-dated the zipped debuginfo. Did you find the root cause for that
yet, or is that an orthogonal topic?

And this also appears to have additional modifications for the earlier workaround add_gnu_debuglink.
I has assumed in reviewing 7165060 that you only used that for makefiles that built with DOF sections,
i.e. and that you had chosen not to use the simpler add_gnu_debuglink in all cases, even though
it is generally less risky. Did you change your approach here? Your comment was changed from SUNW_DOF to
SUNW_*. That makes sense to me. And are you also changing that for
the other jdk repositories?


On May 23, 2012, at 3:59 PM, Daniel D. Daugherty wrote:

> Greetings,
> This is a hotspot code review request for the second of a pair of
> Full Debug Symbols gobjcopy work arounds on Solaris. The first
> hotspot FDS gobjcopy work around was reviewed using bug 7165060
> and that fixed the dtrace test failures.
> The gobjcopy utility also crashes due to empty sections with the
> SHF_ALLOC flagset on Solaris X64 objects. This causes build
> failures.
> The first new temporary work around tool is add_gnu_debuglink
> and it was added by 7165060.The second new temporary work around
> tool is:
> fix_empty_sec_hdr_flags - removes the SHF_ALLOC flag from empty
>    sections in ELF objects.
> These temporary work arounds are only needed until the proper
> Solaris 10 Update 6 patches are made available. The two patches
> are independent of one another which is why there are two
> separate temporary work arounds. However, we're putting the
> temporary work arounds in place because the 7u6/HSX-23.2 project
> window is closing fast.
> Here is the webrev URL for the HSX-24 version:
> This fix will also be backported to 7u6/HSX-23.2 and I expect the
> changes to virtually identical.
> Thanks, in advance, for any reviews!
> Dan

More information about the build-dev mailing list