On 12/1/06, Mikhail Fursov <[EMAIL PROTECTED]> wrote:
On 12/1/06, Xiao-Feng Li <[EMAIL PROTECTED]> wrote:
>
> Agree with Ivan, probably this next-to-high-bit thing is outdated, I
> remember Mikail Fursov said the finalize property will be checked
> outside the allocation invocation. (So agree with Gregory as well
> this may have nothing to do with JVMTI.)
>
JIT checks finalizable property for an object only for inlining of
allocation helpers.
If object is finalizable the allocation helper is not inlined and usual VM
helper is called.
This is all what JIT does for finalizable objects today. Usual VM helper
must check finalizable property by itself. Today this check is done inside
of GC_CC - fast_alloc fails if object is finalizable so VM have to call
slow_alloc.
Mikhail, thanks. It's ok for GCv5 to add the check in its allocation sequence.
At the same time, I wonder if that's better for JIT to check the
finalizable property for VM helper as well. I guess this can be easier
for VM helpers since they are GC independent with standard interface.
Only guess anyway, how do you think, Mikhail?
(Btw, I am now trying to build the gc_alloc Java helper for GCv5.)
Thanks,
xiaofeng
--
Mikhail Fursov
