harmony-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Dmitry Irlyanov" <irlya...@gmail.com>
Subject Re: [build] HWA doesn't work
Date Fri, 01 Dec 2006 13:20:59 GMT
Ivan, Geir

I've tested your proposition on revision: 480581. Result is negative.
File modifying: harmony/trunk/working_vm/vm/vmcore/include/classloader.h
(
-    void SetBCPElement(const char *path, apr_pool_t *tmp_pool);
+    inline void SetBCPElement(const char *path, apr_pool_t *tmp_pool);
)
The result is:
/harmony/trunk/working_vm/build/lnx_ia32_icc_release/deploy/jre/bin/java -cp
. Hello
Failed to open JVM DLL:
/nfs/ims/proj/drl/mrt/users/irlyanov/harmony/trunk/working_vm/build/lnx_ia32_icc_release/deploy/jre/bin/default/harmonyvm
(/nfs/ims/proj/drl/mrt/users/irlyanov/harmony/trunk/working_vm/build/lnx_ia32_icc_release/deploy/jre/bin/default/libharmonyvm.so:
undefined symbol: _ZN20BootstrapClassLoader13SetBCPElementEPKcP10apr_pool_t)

Nevertheless, I propose to apply the JIRA-2375 issue due to there is no
performance waste, but the raise of universality.
This is release version characteristic only.
It is the compiler peculiarity (if you will, the bug) but I can't see the
motive to stop the patch committing.
Note also that this bug is the reason of windows icl release building, so
the priority of this issue is higher.


2006/12/1, Geir Magnusson Jr. <geir@pobox.com>:
>
> That's a reasonable argument.  Move the definition to the header file
> then.
>
> geir
>
>
> Ivan Volosyuk wrote:
> > I will call it a bug in source base, because if we define inline
> > function in cpp file it should only be used inside it. Any public
> > inline functions should be defined in header files.
> >
> > Intel compiler is much more aggressive for optimizations, this require
> > a cleaner code to be written in order to use it. Some of the
> > optimizations trigger bugs in source which are happily invisible by
> > other compilers, I recall one such a case which I have fixed only
> > after attentive examination of generated machine code.
> > --
> > Ivan
> >
> > On 11/30/06, Mikhail Fursov <mike.fursov@gmail.com> wrote:
> >> On 11/30/06, Geir Magnusson Jr. <geir@pobox.com> wrote:
> >> >
> >> >
> >> > I absolutely don't understand how you can claim a "hint" like
> "inline"
> >> > can be considered a bug in the source when the sourcebase is being
> >> > compiled by the same compiler.  can you explain?
> >> >
> >> > The only logical reason I can find is a miscommunication between
> >> compiler
> >> and linker.
>



-- 
_______________
With Best Regards,
Irlyanov Dmitry
Intel Corporation
Middleware Products Division

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message