trace_bytecode
David Holmes
david.holmes at oracle.com
Sun Jun 5 23:42:47 UTC 2016
On 5/06/2016 8:02 AM, yumin qi wrote:
> I mean the inside 'ifndef PRODUCT' is not needed.
Ah - right! Hardly worth the overhead of a bug to fix it though - can be
fixed when/if any functional change is needed in this code.
David
> On Saturday, June 4, 2016, David Holmes <david.holmes at oracle.com
> <mailto:david.holmes at oracle.com>> wrote:
>
> Hi Yumin,
>
> On 5/06/2016 6:12 AM, yumin qi wrote:
>
> Hi,
>
> In JDK8/hotspot: sharedRuntime.cpp, there is an embedded macro
> checking:
>
> #ifndef PRODUCT
>
> JRT_ENTRY(intptr_t, SharedRuntime::trace_bytecode(JavaThread*
> thread,
> intptr_t preserve_this_value, intptr_t tos, intptr_t tos2))
>
> const frame f = thread->last_frame();
>
> assert(f.is_interpreted_frame(), "must be an interpreted frame");
>
> #ifndef PRODUCT
>
> methodHandle mh(THREAD, f.interpreter_frame_method());
>
> BytecodeTracer::trace(mh, f.interpreter_frame_bcp(), tos, tos2);
>
> #endif // !PRODUCT
>
> return preserve_this_value;
>
> JRT_END
>
> #endif // !PRODUCT
>
> It looks not necessary --- JDK9 has changed.
>
>
> Why do you think this is unnecessary? In 9 it has moved to
> interpreterRuntime.cpp
>
> David
> -----
>
> I could not file bug (still could not logon to bugjira), could
> someone file
> a bug for it?
>
> Thanks
>
> Yumin
>
More information about the hotspot-runtime-dev
mailing list