diff --git a/src/gallium/drivers/d3d12/d3d12_compiler.cpp b/src/gallium/drivers/d3d12/d3d12_compiler.cpp
index 14c307e4723..3fe15b06cf6 100644
--- a/src/gallium/drivers/d3d12/d3d12_compiler.cpp
+++ b/src/gallium/drivers/d3d12/d3d12_compiler.cpp
@@ -737,7 +737,7 @@ validate_geometry_shader_variant(struct d3d12_selection_context *sel_ctx)
    if (sel_ctx->fill_mode_lowered != PIPE_POLYGON_MODE_FILL) {
       key.fill_mode = sel_ctx->fill_mode_lowered;
       key.cull_mode = sel_ctx->cull_mode_lowered;
-      key.has_front_face = BITSET_TEST(fs->initial->info.system_values_read, SYSTEM_VALUE_FRONT_FACE);
+      key.has_front_face = (fs->initial->info.inputs_read & VARYING_BIT_FACE) != 0;
       if (key.cull_mode != PIPE_FACE_NONE || key.has_front_face)
          key.front_ccw = ctx->gfx_pipeline_state.rast->base.front_ccw ^ (ctx->flip_y < 0);
       key.edge_flag_fix = needs_edge_flag_fix(ctx->initial_api_prim);
diff --git a/src/gallium/drivers/d3d12/d3d12_screen.cpp b/src/gallium/drivers/d3d12/d3d12_screen.cpp
index 0ef4d06450f..c57ac12196c 100644
--- a/src/gallium/drivers/d3d12/d3d12_screen.cpp
+++ b/src/gallium/drivers/d3d12/d3d12_screen.cpp
@@ -231,9 +231,6 @@ d3d12_get_param_default(struct pipe_screen *pscreen, enum pipe_cap param)
    case PIPE_CAP_FS_COORD_ORIGIN_UPPER_LEFT:
       return 1;
 
-   case PIPE_CAP_FS_FACE_IS_INTEGER_SYSVAL:
-      return 1;
-
    case PIPE_CAP_ACCELERATED:
       return screen->vendor_id != HW_VENDOR_MICROSOFT;
 
diff --git a/src/microsoft/compiler/dxil_nir.c b/src/microsoft/compiler/dxil_nir.c
index 49359965692..a3ad53f69a0 100644
--- a/src/microsoft/compiler/dxil_nir.c
+++ b/src/microsoft/compiler/dxil_nir.c
@@ -1470,9 +1470,6 @@ lower_sysval_to_load_input_impl(nir_builder *b, nir_intrinsic_instr *intr,
 {
    gl_system_value sysval = SYSTEM_VALUE_MAX;
    switch (intr->intrinsic) {
-   case nir_intrinsic_load_front_face:
-      sysval = SYSTEM_VALUE_FRONT_FACE;
-      break;
    case nir_intrinsic_load_instance_id:
       sysval = SYSTEM_VALUE_INSTANCE_ID;
       break;
@@ -1487,22 +1484,13 @@ lower_sysval_to_load_input_impl(nir_builder *b, nir_intrinsic_instr *intr,
    nir_variable *var = sysval_vars[sysval];
    assert(var);
 
-   const nir_alu_type dest_type = (sysval == SYSTEM_VALUE_FRONT_FACE)
-      ? nir_type_uint32 : nir_get_nir_type_for_glsl_type(var->type);
-   const unsigned bit_size = (sysval == SYSTEM_VALUE_FRONT_FACE)
-      ? 32 : intr->def.bit_size;
+   const nir_alu_type dest_type = nir_get_nir_type_for_glsl_type(var->type);
+   const unsigned bit_size = intr->def.bit_size;
 
    b->cursor = nir_before_instr(&intr->instr);
    nir_def *result = nir_load_input(b, intr->def.num_components, bit_size, nir_imm_int(b, 0),
       .base = var->data.driver_location, .dest_type = dest_type);
 
-   /* The nir_type_uint32 is really a nir_type_bool32, but that type is very
-    * inconvenient at this point during compilation.  Convert to
-    * nir_type_bool1 by comparing with zero.
-    */
-   if (sysval == SYSTEM_VALUE_FRONT_FACE)
-      result = nir_ine_imm(b, result, 0);
-
    nir_def_rewrite_uses(&intr->def, result);
    return true;
 }
@@ -2268,36 +2256,17 @@ dxil_nir_lower_unsupported_subgroup_scan(nir_shader *s)
    return ret;
 }
 
-static bool
-lower_load_face(nir_builder *b, nir_intrinsic_instr *intr, void *data)
-{
-   if (intr->intrinsic != nir_intrinsic_load_front_face)
-      return false;
-
-   b->cursor = nir_before_instr(&intr->instr);
-
-   nir_variable *var = data;
-   nir_def *load = nir_ine_imm(b, nir_load_var(b, var), 0);
-
-   nir_def_rewrite_uses(&intr->def, load);
-   nir_instr_remove(&intr->instr);
-   return true;
-}
-
 bool
 dxil_nir_forward_front_face(nir_shader *nir)
 {
    assert(nir->info.stage == MESA_SHADER_FRAGMENT);
 
-   nir_variable *var = nir_variable_create(nir, nir_var_shader_in,
-                                           glsl_uint_type(),
-                                           "gl_FrontFacing");
-   var->data.location = VARYING_SLOT_VAR12;
-   var->data.interpolation = INTERP_MODE_FLAT;
-
-   return nir_shader_intrinsics_pass(nir, lower_load_face,
-                                       nir_metadata_block_index | nir_metadata_dominance,
-                                       var);
+   nir_variable *var = nir_find_variable_with_location(nir, nir_var_shader_in, VARYING_SLOT_FACE);
+   if (var) {
+      var->data.location = VARYING_SLOT_VAR12;
+      return true;
+   }
+   return false;
 }
 
 static bool
diff --git a/src/microsoft/compiler/dxil_signature.c b/src/microsoft/compiler/dxil_signature.c
index 02501075554..e62c30225a3 100644
--- a/src/microsoft/compiler/dxil_signature.c
+++ b/src/microsoft/compiler/dxil_signature.c
@@ -201,9 +201,6 @@ get_semantic_sv_name(nir_variable *var, struct semantic_info *info, gl_shader_st
    case SYSTEM_VALUE_VERTEX_ID_ZERO_BASE:
       info->kind = DXIL_SEM_VERTEX_ID;
       break;
-   case SYSTEM_VALUE_FRONT_FACE:
-      info->kind = DXIL_SEM_IS_FRONT_FACE;
-      break;
    case SYSTEM_VALUE_INSTANCE_ID:
       info->kind = DXIL_SEM_INSTANCE_ID;
       break;
diff --git a/src/microsoft/compiler/nir_to_dxil.c b/src/microsoft/compiler/nir_to_dxil.c
index cb17af02469..7f2892528f5 100644
--- a/src/microsoft/compiler/nir_to_dxil.c
+++ b/src/microsoft/compiler/nir_to_dxil.c
@@ -589,7 +589,6 @@ struct ntd_context {
    const struct dxil_value **scratchvars;
    const struct dxil_value **consts;
 
-   nir_variable *ps_front_face;
    nir_variable *system_value[SYSTEM_VALUE_MAX];
 
    nir_function *tess_ctrl_patch_constant_func;
diff --git a/src/microsoft/spirv_to_dxil/dxil_spirv_nir.c b/src/microsoft/spirv_to_dxil/dxil_spirv_nir.c
index 45a00f5eac4..cfbb5cc1ea8 100644
--- a/src/microsoft/spirv_to_dxil/dxil_spirv_nir.c
+++ b/src/microsoft/spirv_to_dxil/dxil_spirv_nir.c
@@ -885,6 +885,7 @@ dxil_spirv_nir_passes(nir_shader *nir,
    const struct nir_lower_sysvals_to_varyings_options sysvals_to_varyings = {
       .frag_coord = true,
       .point_coord = true,
+      .front_face = true,
    };
    NIR_PASS_V(nir, nir_lower_sysvals_to_varyings, &sysvals_to_varyings);