On Saturday, August 27, 2016, Jan Vesely <<a href="mailto:jan.vesely@rutgers.edu">jan.vesely@rutgers.edu</a>> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">On Sat, 2016-08-27 at 12:05 -0400, Rhys Kidd wrote:<br>
> As reported by Clang, TGSI_OPCODE_DFMA (defined magic number 118) is<br>
> currently initialized twice for Cayman and Evergreen.<br>
> When Jan Vesely added double precision FMA opcode it did make sense<br>
> to locate it immediately after TGSI_OPCODE_DMAD, although this is<br>
> out of order.<br>
><br>
> This change cleans up the prior magic number definition and ensures<br>
> any later reordering of this struct will not create problems.<br>
><br>
> Prior change was:<br>
><br>
> commit 015e2e0fce3eea7884f8df275c2fad<wbr>c35143a324<br>
> Author: Jan Vesely <<a href="javascript:;" onclick="_e(event, 'cvml', 'jan.vesely@rutgers.edu')">jan.vesely@rutgers.edu</a>><br>
> Date: Sat Jul 2 16:14:54 2016 -0400<br>
><br>
> r600g: Add double precision FMA ops<br>
><br>
> Bugzilla: <a href="https://bugs.freedesktop.org/show_bug.cgi?id=96782" target="_blank">https://bugs.freedesktop.org/<wbr>show_bug.cgi?id=96782</a><br>
> Fixes: 54c4d525da7c7fc1e103d7a3e6db01<wbr>5abb132d5d ("r600g: Enable<br>
> FMA on chips that support it")<br>
><br>
> Signed-off-by: Jan Vesely <<a href="javascript:;" onclick="_e(event, 'cvml', 'jan.vesely@rutgers.edu')">jan.vesely@rutgers.edu</a>><br>
> Tested-by: James Harvey <<a href="javascript:;" onclick="_e(event, 'cvml', 'lothmordor@gmail.com')">lothmordor@gmail.com</a>><br>
> Signed-off-by: Marek Olšák <<a href="javascript:;" onclick="_e(event, 'cvml', 'marek.olsak@amd.com')">marek.olsak@amd.com</a>><br>
><br>
> Signed-off-by: Rhys Kidd <<a href="javascript:;" onclick="_e(event, 'cvml', 'rhyskidd@gmail.com')">rhyskidd@gmail.com</a>><br>
<br>
So much for using magic numbers.<br>
Reviewed-by: Jan Vesely <<a href="javascript:;" onclick="_e(event, 'cvml', 'jan.vesely@rutgers.edu')">jan.vesely@rutgers.edu</a>><br>
<br>
thanks,<br>
Jan</blockquote><div><br></div><div>Thanks for reviewing Jan.</div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<br>
> ---<br>
> src/gallium/drivers/r600/<wbr>r600_shader.c | 4 ++--<br>
> 1 file changed, 2 insertions(+), 2 deletions(-)<br>
><br>
> diff --git a/src/gallium/drivers/r600/<wbr>r600_shader.c<br>
> b/src/gallium/drivers/r600/<wbr>r600_shader.c<br>
> index 64aacca..a39301f 100644<br>
> --- a/src/gallium/drivers/r600/<wbr>r600_shader.c<br>
> +++ b/src/gallium/drivers/r600/<wbr>r600_shader.c<br>
> @@ -9210,7 +9210,7 @@ static const struct<br>
> r600_shader_tgsi_instruction eg_shader_tgsi_instruction[] =<br>
> [TGSI_OPCODE_BREAKC] = { ALU_OP0_NOP,<br>
> tgsi_unsupported},<br>
> [TGSI_OPCODE_KILL_IF] = { ALU_OP2_KILLGT,<br>
> tgsi_kill}, /* conditional kill */<br>
> [TGSI_OPCODE_END] = { ALU_OP0_NOP, tgsi_end}, /* aka<br>
> HALT */<br>
> - [118] = { ALU_OP0_NOP,<br>
> tgsi_unsupported},<br>
> + /* Refer below for TGSI_OPCODE_DFMA */<br>
> [TGSI_OPCODE_F2I] = { ALU_OP1_FLT_TO_INT, tgsi_f2i},<br>
> [TGSI_OPCODE_IDIV] = { ALU_OP0_NOP, tgsi_idiv},<br>
> [TGSI_OPCODE_IMAX] = { ALU_OP2_MAX_INT, tgsi_op2},<br>
> @@ -9433,7 +9433,7 @@ static const struct<br>
> r600_shader_tgsi_instruction cm_shader_tgsi_instruction[] =<br>
> [TGSI_OPCODE_BREAKC] = { ALU_OP0_NOP,<br>
> tgsi_unsupported},<br>
> [TGSI_OPCODE_KILL_IF] = { ALU_OP2_KILLGT,<br>
> tgsi_kill}, /* conditional kill */<br>
> [TGSI_OPCODE_END] = { ALU_OP0_NOP, tgsi_end}, /* aka<br>
> HALT */<br>
> - [118] = { ALU_OP0_NOP,<br>
> tgsi_unsupported},<br>
> + /* Refer below for TGSI_OPCODE_DFMA */<br>
> [TGSI_OPCODE_F2I] = { ALU_OP1_FLT_TO_INT, tgsi_op2},<br>
> [TGSI_OPCODE_IDIV] = { ALU_OP0_NOP, tgsi_idiv},<br>
> [TGSI_OPCODE_IMAX] = { ALU_OP2_MAX_INT, tgsi_op2},<br>
</blockquote>