RFR(M): 8195112: x86 (32 bit): implementation for Thread-local handshakes

Robbin Ehn robbin.ehn at oracle.com
Wed Jan 17 21:04:41 UTC 2018


Hi Martin,

On 01/15/2018 05:31 PM, Doerr, Martin wrote:
> Hi,
> 
> I think supporting Thread-local handshakes on all platforms will make future 
> changes easier, because some new features may require it.

Note that there is a fallback path using handshakes to safepoint if TLH is not supported.
So you can still use handshake one thread but with safepoint penalty.

> 
> At least, I prefer to treat 32 bit like the 64 bit version. So I just 
> implemented it for x86 (32 bit).
> 
> Please review:
> 
> http://cr.openjdk.java.net/~mdoerr/8195112_x86_local_poll/webrev.00/
> 
> I’ve ran a couple of tests on linux and Windows. Maybe someone else still 
> supports 32 bit and wants to run additional tests?

I'll do a sanity run.

> 
> I haven’t done performance comparisons, because it’s not so critical for us on 
> 32 bit.

This guy will hurt:
712       masm.get_thread(pollReg);

Since we are not shipping 32-bit but RedHat is (at least in Fedora) and they don't object (or anyone else) I'm fine with this.

> 
> I will also need a sponsor.

I'll do the sanity testing and I sponsor when we have some more reviews, preferable from RH since they are affected(?).

/Robbin

> 
> Best regards,
> 
> Martin
> 


More information about the hotspot-runtime-dev mailing list