RFR(L): 8008407: remove SPARC V8 support

Morris Meyer morris.meyer at oracle.com
Fri Jun 7 14:11:12 PDT 2013


I have addressed the mult and get_pc() and argument.cpp issues from the 
review and have successfully pushed those through JPRT.  Removing 
v9_only() seems to crash on JPRT and will be left for another day, along 
with the remaining few V8 instructions.

Thanks,

         --mm

WEBREV - http://cr.openjdk.java.net/~morris/8008407.03

On 6/1/13 2:54 PM, Morris Meyer wrote:
> I removed these things accordingly - and got failure on JPRT.
>
> I left the v9_dep changes in as a marker to the future instructions 
> that need to be removed.
>
>         --mm
>
>
> JPRT - 
> http://prt-web.us.oracle.com//archive/2013/06/2013-06-01-172500.mameyer.8008407//JobStatus.txt
>
> WEBREV - http://cr.openjdk.java.net/~morris/8008407.02
>
> On 5/31/13 5:57 PM, Christian Thalinger wrote:
>> src/cpu/sparc/vm/assembler_sparc.hpp:
>>
>> Can you also remove these:
>>
>> v9_only, v9_dep, and (if possible) v8_no_cc.
>>
>> src/cpu/sparc/vm/macroAssembler_sparc.cpp:
>>
>> Also remove these (now useless) wrappers:
>>
>>   void MacroAssembler::mult(Register s1, Register s2, Register d) {
>>   void MacroAssembler::mult(Register s1, int simm13a, Register d) {
>>
>> src/cpu/sparc/vm/macroAssembler_sparc.inline.hpp:
>>
>> Could you also get rid of get_pc like in my changes for 8007415:
>>
>> 1) src/cpu/sparc/vm/macroAssembler_sparc.cpp:
>>
>> -  inc(L4, get_pc(L4) + 2 * BytesPerInstWord); // skip getpc() code + 
>> inc + st_ptr to point L4 at call
>> +  rdpc(L4);
>> +  inc(L4, 3 * BytesPerInstWord); // skip rdpc + inc + st_ptr to 
>> point L4 at call
>>
>> 2) src/cpu/sparc/vm/sparc.ad:
>>
>> -    __ get_pc(L2);
>> +    __ rdpc(L2);
>>
>> src/os_cpu/solaris_sparc/vm/atomic_solaris_sparc.inline.hpp:
>>
>> Since _Atomic_move_long_v8 is not used anymore you can removed it from:
>>
>> src/os_cpu/solaris_sparc/vm/solaris_sparc.il
>>
>> -- Chris
>>
>> On May 31, 2013, at 12:50 PM, Morris Meyer <morris.meyer at oracle.com> 
>> wrote:
>>
>>> Could I get a review for this issue?
>>>
>>> This patch is based on work originally by Christian Thalinger with 
>>> several chunks of V8 removal from me.  There are a few V8 
>>> instructions left - LDD, STD, CASA and a few others.  Follow on bugs 
>>> will be assigned to rework each specific instruction.
>>>
>>> These changes have been tested with the vm regression suite and JPRT.
>>>
>>> Thanks in advance.
>>>
>>>         --morris
>>>
>>> WEBREV - http://cr.openjdk.java.net/~morris/8008407.01
>>> BUG - https://jbs.oracle.com/bugs/browse/JDK-8008407
>>>
>>>
>



More information about the hotspot-compiler-dev mailing list