RFR: JDK-8217909: Make unused r12 register (without compressed oops) available to regalloc in C2
rkennke at redhat.com
Tue Feb 12 09:45:04 UTC 2019
Yeah, I'll see what I can do. Better to cut the number of classes in
half (also addressing rbp) than to double them. :-)
> The reg classes is getting out of control. For rbp, it was acceptable,
> but with r12 as well, it has to be refactored.
> I would prefer solution that extends the ad-format. Deans suggestion is
> better than duplicating the reg classes, but modifying the reg class
> definitions without any coupling to the ad-files might cause some
> unwanted suprises in the future.
> // Nils
> On 2019-02-11 21:33, dean.long at oracle.com wrote:
>> There is a way to do it without cloning all the register class variants.
>> For the arm64 port we massaged the register masks in
>> It might be worth considering, as the number of variants grows.
>> On 2/11/19 3:02 AM, Roman Kennke wrote:
>>> When running with compressed oops, the r12 register holds the heapbase,
>>> and thus is not available to register allocation in C2. However, when
>>> *not* running with compressed oops, it is still not available and
>>> remains unused. It should be made available to register allocation in
>>> this case.
>>> This patch implements this by introducing with_r12 and no_r12 variants
>>> of basically all register classes, and add dynamic reg classes to select
>>> one or the other, based on current settings. I needed to add UseZGC in
>>> those flags because ZGC asserts to not get r12 in its barriers. Not sure
>>> that this is necessary.
>>> Note that we might want to use the r12 register in Shenandoah later to
>>> keep GC state. In this case, we'd need to add UseShenandoahGC or such to
>>> the test. Do we want to abstract this whole check? Not sure that this is
>>> possible/feasible to do in .ad though...
>>> Testing: tier1 no regressions locally, eyeball generated code, yes it
>>> does use r12 now.
>>> Can I please get reviews?
More information about the hotspot-compiler-dev