[Mesa-dev] [PATCH 08/11] intel/compiler: fix brw_negate_immediate for 16-bit types

[Chema Casanova]

El 24/04/18 a las 23:55, Jason Ekstrand escribió:
> On Wed, Apr 11, 2018 at 12:20 AM, Iago Toral Quiroga <itoral at igalia.com
> <mailto:itoral at igalia.com>> wrote:
> 
>     From: Jose Maria Casanova Crespo <jmcasanova at igalia.com
>     <mailto:jmcasanova at igalia.com>>
> 
>     16-bit immediates are replicated in each word of a 32-bit value
>     so we need to negate both.
>     ---
>      src/intel/compiler/brw_shader.cpp | 5 ++---
>      1 file changed, 2 insertions(+), 3 deletions(-)
> 
>     diff --git a/src/intel/compiler/brw_shader.cpp
>     b/src/intel/compiler/brw_shader.cpp
>     index 9cdf9fcb23d..c7edc60b63d 100644
>     --- a/src/intel/compiler/brw_shader.cpp
>     +++ b/src/intel/compiler/brw_shader.cpp
>     @@ -581,7 +581,8 @@ brw_negate_immediate(enum brw_reg_type type,
>     struct brw_reg *reg)
>            return true;
>         case BRW_REGISTER_TYPE_W:
>         case BRW_REGISTER_TYPE_UW:
>     -      reg->d = -(int16_t)reg->ud;
>     +   case BRW_REGISTER_TYPE_HF:
>     +      reg->ud ^= 0x80008000;
> 
> 
> This is not correct for integers.  We need to keep two separate cases.

That's true, I've forgotten about two's complement representation. For
this series v2 I will send:

    case BRW_REGISTER_TYPE_UW:
-      reg->d = -(int16_t)reg->ud;
+      uint16_t value = -(int16_t)reg->ud;
+      reg->ud = value | value << 16;
+      return true;
+   case BRW_REGISTER_TYPE_HF:
+      reg->ud ^= 0x80008000;

I'm including for v2 also a new patch for solving a problem with
negative values at brw_imm_w that is replicating the 16-bit value
wrongly because of sign extension.

>  
> 
>            return true;
>         case BRW_REGISTER_TYPE_F:
>            reg->f = -reg->f;
>     @@ -602,8 +603,6 @@ brw_negate_immediate(enum brw_reg_type type,
>     struct brw_reg *reg)
>         case BRW_REGISTER_TYPE_UV:
>         case BRW_REGISTER_TYPE_V:
>            assert(!"unimplemented: negate UV/V immediate");
>     -   case BRW_REGISTER_TYPE_HF:
>     -      assert(!"unimplemented: negate HF immediate");
>         case BRW_REGISTER_TYPE_NF:
>            unreachable("no NF immediates");
>         }
>     -- 
>     2.14.1
> 
>     _______________________________________________
>     mesa-dev mailing list
>     mesa-dev at lists.freedesktop.org <mailto:mesa-dev at lists.freedesktop.org>
>     https://lists.freedesktop.org/mailman/listinfo/mesa-dev
>     <https://lists.freedesktop.org/mailman/listinfo/mesa-dev>
> 
> 
> 
> 
> _______________________________________________
> mesa-dev mailing list
> mesa-dev at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/mesa-dev
>