[Mesa-dev] [PATCH 2/2] intel/compiler: Use null destination register for memory fence messages

[Kenneth Graunke]

On Tuesday, February 6, 2018 5:09:10 PM PST Anuj Phogat wrote:
> From Message Descriptor section in gfxspecs:
>  "Memory fence messages without Commit Enable set do not return
>   anything to the thread (response length is 0 and destination
>   register is null)."
> 
> It fixes a piglit GPU hang in simulation environment.
> Piglit test: arb_shader_image_load_store-shader-mem-barrier
> 
> Signed-off-by: Anuj Phogat <anuj.phogat at gmail.com>
> Cc: mesa-stable at lists.freedesktop.org
> ---
>  src/intel/compiler/brw_eu_emit.c | 8 +++++++-
>  1 file changed, 7 insertions(+), 1 deletion(-)
> 
> diff --git a/src/intel/compiler/brw_eu_emit.c b/src/intel/compiler/brw_eu_emit.c
> index 1fb9aab51c..c66b813af8 100644
> --- a/src/intel/compiler/brw_eu_emit.c
> +++ b/src/intel/compiler/brw_eu_emit.c
> @@ -3290,7 +3290,13 @@ brw_memory_fence(struct brw_codegen *p,
>      */
>     insn = next_insn(p, BRW_OPCODE_SEND);
>     dst = retype(dst, BRW_REGISTER_TYPE_UW);
> -   brw_set_dest(p, insn, dst);
> +
> +  /* From Message Descriptor section in gfxspecs:
> +        "Memory fence messages without Commit Enable set do not return
> +         anything to the thread (response length is 0 and destination
> +         register is null).
> +   */
> +   brw_set_dest(p, insn, retype(vec1(brw_null_reg()), BRW_REGISTER_TYPE_UW));
>     brw_set_src0(p, insn, dst);
>     brw_set_memory_fence_message(p, insn, GEN7_SFID_DATAPORT_DATA_CACHE,
>                                  commit_enable);
> 

This seems wrong - it looks like you're doing this for messages with
commit enable set, which do return things...

Cc'ing Curro.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: This is a digitally signed message part.
URL: <https://lists.freedesktop.org/archives/mesa-dev/attachments/20180207/fd0e021d/attachment.sig>