From 3b5c8ec1a9b31d99dd2a0192ea7d46b19c36dde8 Mon Sep 17 00:00:00 2001 From: Jason Ekstrand Date: Fri, 11 Nov 2022 15:02:07 -0600 Subject: [PATCH] r600/nir: Fix u64vec2 immediate lowering There were a couple of issues here: 1. We should be using nir_const_value_for_uint instead of setting the union fields directly to ensure the rest of the union is zeroed. 2. It was always filling out the first two components of val even if the incoming constant had 2 64-bit components. Fixes: 165fb5117bf7 ("r600/sfn: add lowering passes to get 64 bit ops lowered to 32 bit vec2") Reviewed-by: Alyssa Rosenzweig Part-of: (cherry picked from commit f3f1c28f8e6d40823e3d12415a8d0ea622f9fa20) --- .pick_status.json | 2 +- src/gallium/drivers/r600/sfn/sfn_nir_lower_64bit.cpp | 8 ++++---- 2 files changed, 5 insertions(+), 5 deletions(-) diff --git a/.pick_status.json b/.pick_status.json index 81e516629d6..cb024654a3b 100644 --- a/.pick_status.json +++ b/.pick_status.json @@ -4054,7 +4054,7 @@ "description": "r600/nir: Fix u64vec2 immediate lowering", "nominated": true, "nomination_type": 1, - "resolution": 0, + "resolution": 1, "main_sha": null, "because_sha": "165fb5117bf70402e66d34538d4085e060f57fea" }, diff --git a/src/gallium/drivers/r600/sfn/sfn_nir_lower_64bit.cpp b/src/gallium/drivers/r600/sfn/sfn_nir_lower_64bit.cpp index 9828802ba2c..09f294ca481 100644 --- a/src/gallium/drivers/r600/sfn/sfn_nir_lower_64bit.cpp +++ b/src/gallium/drivers/r600/sfn/sfn_nir_lower_64bit.cpp @@ -929,12 +929,12 @@ Lower64BitToVec2::lower(nir_instr *instr) } case nir_instr_type_load_const: { auto lc = nir_instr_as_load_const(instr); - assert(lc->def.num_components < 3); - nir_const_value val[4] = {{0}}; + assert(lc->def.num_components <= 2); + nir_const_value val[4]; for (uint i = 0; i < lc->def.num_components; ++i) { uint64_t v = lc->value[i].u64; - val[0].u32 = v & 0xffffffff; - val[1].u32 = (v >> 32) & 0xffffffff; + val[i * 2 + 0] = nir_const_value_for_uint(v & 0xffffffff, 32); + val[i * 2 + 1] = nir_const_value_for_uint(v >> 32, 32); } return nir_build_imm(b, 2 * lc->def.num_components, 32, val);