[10] RFR: 8044829: UseAES warning message generation differs on CPU architecture

[Rahul Raghavan]

Hi,

> You don't need to add  '&& UseAES' check to asserts in assembler_x86.cpp ...
> The fix in vm_version_x86.cpp is correct.

Thank you Vladimir for review comments.
So I will push only the vm_version_x86.cpp fix.

Thanks,
Rahul

> -----Original Message-----
> From: Vladimir Kozlov
> Sent: Saturday, June 17, 2017 5:48 AM
> To: hotspot-compiler-dev at openjdk.java.net
> Subject: Re: [10] RFR: 8044829: UseAES warning message generation differs on CPU architecture
> 
> Hi Rahul
> 
> You don't need to add  '&& UseAES' check to asserts in assembler_x86.cpp
> because these instructions generation is guarded by UseAES checks in
> other places.
> 
> The fix in vm_version_x86.cpp is correct.
> 
> Thanks,
> Vladimir
> 
> On 6/16/17 1:24 PM, Rahul Raghavan wrote:
> > Hi,
> >
> > Please review the following fix proposal.
> >
> > <webrev.00> - http://cr.openjdk.java.net/~rraghavan/8044829/webrev.00/
> >      src/cpu/x86/vm/vm_version_x86.cpp   -   3 lines changed: 0 ins; 3 del; 0 mod; 1436 unchg
> >      src/cpu/x86/vm/assembler_x86.cpp    -   8 lines changed: 0 ins; 0 del; 8 mod; 8843 unchg
> >
> > No issues with jprt (-testset hotspot) testing.
> >
> > Please refer <jbs> comments - https://bugs.openjdk.java.net/browse/JDK-8044829
> >
> > 1. Proposed fix changes in 'vm_version_x86.cpp' fixes the wrong warning generated for options -XX:-UseAES -XX:+UseAESIntrinsics.
> >        void VM_Version::get_processor_features() {
> >        .........
> >        -  if (!UseAES && !FLAG_IS_DEFAULT(UseAES))
> >        -    _features &= ~CPU_AES;
> >        -
> >
> > .e.g.: wrong warning without fix for x64, WITH AES support -
> >        Java HotSpot(TM) 64-Bit Server VM warning: AES intrinsics are not available on this CPU
> >
> > .e.g.: correct warning with fix - (same as already fixed for sparc)
> >        Java HotSpot(TM) 64-Bit Server VM warning: AES intrinsics require UseAES flag to be enabled. Intrinsics will be disabled.
> >
> > 2. Seems additional changes as proposed maybe required in  'assembler_x86.cpp' after changes in 'vm_version_x86.cpp'.
> >
> > 3. Found issues reported earlier for sparc was solved with 8131778 fix in 9-b96.
> > Found no similar issues with other targets.
> >
> >
> > Thanks,
> > Rahul
> >