RFR(S): 8224580: Matcher can cause oop field/array element to be reloaded

[Roman Kennke]

Hi Roland,


> Hi Dean,
> 
> Thanks for looking at this.
> 
>> It seems to do what you want, but I'm curious, what's the worse that can 
>> happen if we call set_shared() on a node that doesn't need it?
> 
> Some complex match rules wouldn't trigger anymore. Let's say we have
> (CmpN (LoadN ... and we erroneously call set_shared on the LoadN then
> this:
> 
> instruct testN_mem(rFlagsReg cr, memory mem, immN0 zero)
> %{
>   predicate(CompressedOops::base() != NULL);
>   match(Set cr (CmpN (LoadN mem) zero));
> 
> won't match and instead of a single instruction we would have a load
> instruction followed by a comparison.
> 
> So worse case, correctness is not affected but code quality can be.

In general? Or only in offset==0 case?

Roman

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: OpenPGP digital signature
URL: <https://mail.openjdk.java.net/pipermail/hotspot-compiler-dev/attachments/20190603/0726b46a/signature.asc>