Gregory Shimansky wrote:
There is a patch that fixes HARMONY-5510. This is a regression after
commit in HARMONY-5490. The single line change in encoder that didn't
relate to crash handler actually caused a problem in JVMTI when decoding
some call instructions. The decoder has to be changed to use information
about register sizes from encoder tables, not rely on REX.W prefix which
is not mandatory for some instructions (like calls).
I want to commit this patch to fix this regression before M5 is
released. What do you think?
+1
(I don't think the testing has started in earnest yet)
Regards,
Tim
