[Mesa-dev] [PATCH 2/2] gk110/ir: always use limm form for log ops with immediates

[Karol Herbst]

On Fri, Mar 16, 2018 at 11:10 PM, Ilia Mirkin <imirkin at alum.mit.edu> wrote:
> You got the subjects backwards.
>
> On Fri, Mar 16, 2018 at 6:05 PM, Karol Herbst <kherbst at redhat.com> wrote:
>> in the short imm form 0x80000 was sign-extended to 0xfff80000 which leads to
>> wrong results.
>>
>> Signed-off-by: Karol Herbst <kherbst at redhat.com>
>> ---
>>  src/gallium/drivers/nouveau/codegen/nv50_ir_emit_nvc0.cpp | 2 +-
>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/src/gallium/drivers/nouveau/codegen/nv50_ir_emit_nvc0.cpp b/src/gallium/drivers/nouveau/codegen/nv50_ir_emit_nvc0.cpp
>> index 58594f02c7f..644bd11ec28 100644
>> --- a/src/gallium/drivers/nouveau/codegen/nv50_ir_emit_nvc0.cpp
>> +++ b/src/gallium/drivers/nouveau/codegen/nv50_ir_emit_nvc0.cpp
>> @@ -887,7 +887,7 @@ CodeEmitterNVC0::emitLogicOp(const Instruction *i, uint8_t subOp)
>>        }
>>     } else
>>     if (i->encSize == 8) {
>> -      if (isLIMM(i->src(1), TYPE_U32)) {
>> +      if (i->src(1).getFile() == FILE_IMMEDIATE) {
>>           emitForm_A(i, HEX64(38000000, 00000002));
>>
>>           if (i->flagsDef >= 0)
>
> Should probably assert i->src(1).mod == 0?

well maybe we could just do the fancy imm.applyTo() thing or something
as well as we do in the gk110 case. Sounds like a reasonable thing to
do instead putting asserts all over the code.