[Mesa-dev] [PATCH] gallium: add more exceptions to tgsi_util_get_inst_usage_mask

[Roland Scheidegger]

Reviewed-by: Roland Scheidegger <sroland at vmware.com>

Albeit the way those masks are derived looks quite error-prone in
general (especially for new opcodes).

Am 19.10.2017 um 18:40 schrieb Tim Rowley:
> A number of double/int64 operations don't have matching
> read and write usage masks, which the fallthrough case of
> tgsi_util_get_inst_usage_mask assumes for componentwise
> tagged instructions.
> 
> No regressions in llvmpipe piglit; fixes a large number of
> swr regressions.
> ---
>  src/gallium/auxiliary/tgsi/tgsi_util.c | 12 ++++++++++++
>  1 file changed, 12 insertions(+)
> 
> diff --git a/src/gallium/auxiliary/tgsi/tgsi_util.c b/src/gallium/auxiliary/tgsi/tgsi_util.c
> index cfce59093c..afe5690ce0 100644
> --- a/src/gallium/auxiliary/tgsi/tgsi_util.c
> +++ b/src/gallium/auxiliary/tgsi/tgsi_util.c
> @@ -230,13 +230,25 @@ tgsi_util_get_inst_usage_mask(const struct tgsi_full_instruction *inst,
>        read_mask = TGSI_WRITEMASK_XYZ;
>        break;
>  
> +   case TGSI_OPCODE_DSEQ:
> +   case TGSI_OPCODE_DSNE:
> +   case TGSI_OPCODE_DSLT:
> +   case TGSI_OPCODE_DSGE:
>     case TGSI_OPCODE_DP4:
>     case TGSI_OPCODE_PK4B:
>     case TGSI_OPCODE_PK4UB:
>     case TGSI_OPCODE_D2F:
> +   case TGSI_OPCODE_D2I:
> +   case TGSI_OPCODE_D2U:
>     case TGSI_OPCODE_I2F:
>     case TGSI_OPCODE_U2F:
> +   case TGSI_OPCODE_U64SEQ:
> +   case TGSI_OPCODE_U64SNE:
> +   case TGSI_OPCODE_U64SLT:
> +   case TGSI_OPCODE_U64SGE:
>     case TGSI_OPCODE_U642F:
> +   case TGSI_OPCODE_I64SLT:
> +   case TGSI_OPCODE_I64SGE:
>     case TGSI_OPCODE_I642F:
>        read_mask = TGSI_WRITEMASK_XYZW;
>        break;
>