RFR(M): 8050942 : PPC64: implement template interpreter for ppc64le

[Lindenmaier, Goetz]

Hi Sasha,

we ran our nightly tests on big-endian with this change.  They're all green.
reviewed.

Best regards,
  Goetz.


-----Original Message-----
From: Lindenmaier, Goetz 
Sent: Freitag, 18. Juli 2014 10:13
To: 'Alexander Smundak'
Cc: HotSpot Open Source Developers
Subject: RE: RFR(M): 8050942 : PPC64: implement template interpreter for ppc64le

Hi Sasha, 

thanks, now it works.  I just ran jvm98/javac.
Comprehensive tests will be executed tonight.

Best regards,
  Goetz.





-----Original Message-----
From: Alexander Smundak [mailto:asmundak at google.com] 
Sent: Freitag, 18. Juli 2014 02:58
To: Lindenmaier, Goetz
Cc: HotSpot Open Source Developers
Subject: Re: RFR(M): 8050942 : PPC64: implement template interpreter for ppc64le

On Thu, Jul 17, 2014 at 3:20 AM, Lindenmaier, Goetz
<goetz.lindenmaier at sap.com> wrote:
> I tested your change.  Unfortunately it breaks our port.  You need to fix Unsigned to
> Signed:
>
> --- a/src/cpu/ppc/vm/templateTable_ppc_64.cpp   Wed Jul 16 16:53:32 2014 -0700
> +++ b/src/cpu/ppc/vm/templateTable_ppc_64.cpp   Thu Jul 17 12:14:18 2014 +0200
> @@ -1929,7 +1929,7 @@
>    // default case
>    __ bind(Ldefault_case);
>
> -  __ get_u4(Roffset, Rdef_offset_addr, 0, InterpreterMacroAssembler::Unsigned);
> +  __ get_u4(Roffset, Rdef_offset_addr, 0, InterpreterMacroAssembler::Signed);
>    if (ProfileInterpreter) {
>      __ profile_switch_default(Rdef_offset_addr, Rcount/* scratch */);
>      __ b(Lcontinue_execution);
Oops. Fixed. Which test was broken by this, BTW?

> If you want to, you can move loading the bci in this bytecode behind the loop.
Done.

> Could you please fix indentation of relocInfo::none in call_c? Should
> be aligned to call_c.
Done.

The revised patch is at
http://cr.openjdk.java.net/~asmundak/8050942/hotspot/webrev.01/
please take another look.

Sasha