From 2d6f82a294ad1ab1eab0020cf65df5ecc9591272 Mon Sep 17 00:00:00 2001 From: =?utf8?q?Samuel=20Iglesias=20Gons=C3=A1lvez?= Date: Thu, 26 May 2016 07:56:37 +0200 Subject: [PATCH] i965/fs: fix offset when loading double vector input varyings MIME-Version: 1.0 Content-Type: text/plain; charset=utf8 Content-Transfer-Encoding: 8bit When we are not packing a double input varying, we might need to read its data in a non-aligned to 64-bit offset, so we read the wrong data. This is happening when using explicit locations in varyings because Mesa disables packing varying for that case. const_index is in 32-bit size units but offset() is multiplying it by destination type size units. When operating with double input varyings, const_index value could be not aligned to 64 bits. To fix it, we load the double vector as if it was a float based vector with twice the number of components. Signed-off-by: Samuel Iglesias Gonsálvez Reviewed-by: Timothy Arceri Cc: "12.0" --- src/mesa/drivers/dri/i965/brw_fs_nir.cpp | 22 +++++++++++++++++++++- 1 file changed, 21 insertions(+), 1 deletion(-) diff --git a/src/mesa/drivers/dri/i965/brw_fs_nir.cpp b/src/mesa/drivers/dri/i965/brw_fs_nir.cpp index a4e5559a369..5b196af636c 100644 --- a/src/mesa/drivers/dri/i965/brw_fs_nir.cpp +++ b/src/mesa/drivers/dri/i965/brw_fs_nir.cpp @@ -3670,9 +3670,21 @@ fs_visitor::nir_emit_intrinsic(const fs_builder &bld, nir_intrinsic_instr *instr case nir_intrinsic_load_input: { fs_reg src; + unsigned num_components = instr->num_components; + enum brw_reg_type type = dest.type; + if (stage == MESA_SHADER_VERTEX) { src = fs_reg(ATTR, instr->const_index[0], dest.type); } else { + assert(type_sz(type) >= 4); + if (type == BRW_REGISTER_TYPE_DF) { + /* const_index is in 32-bit type size units that could not be aligned + * with DF. We need to read the double vector as if it was a float + * vector of twice the number of components to fetch the right data. + */ + dest = retype(dest, BRW_REGISTER_TYPE_F); + num_components *= 2; + } src = offset(retype(nir_inputs, dest.type), bld, instr->const_index[0]); } @@ -3681,10 +3693,18 @@ fs_visitor::nir_emit_intrinsic(const fs_builder &bld, nir_intrinsic_instr *instr assert(const_offset && "Indirect input loads not allowed"); src = offset(src, bld, const_offset->u32[0]); - for (unsigned j = 0; j < instr->num_components; j++) { + for (unsigned j = 0; j < num_components; j++) { bld.MOV(offset(dest, bld, j), offset(src, bld, j)); } + if (type == BRW_REGISTER_TYPE_DF) { + /* Once the double vector is read, set again its original register + * type to continue with normal execution. + */ + src = retype(src, type); + dest = retype(dest, type); + } + if (type_sz(src.type) == 8) { shuffle_32bit_load_result_to_64bit_data(bld, dest, -- 2.30.2