gallivm/nir: fix subgroup invocation read.

Again if you get passed an invoc but the exec mask has the
active lane somewhere other than at 0, then if we have an
invoc we should find the active lane and extract the value
from invoc rather than using the idx.

This fixes a bunch of VK 1.2 subgroup tests once 1.2 is enabled:
dEQP-VK.subgroups.ballot_broadcast.compute.subgroupbroadcast_nonconst*

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Acked-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12953>

diff --git a/src/gallium/auxiliary/gallivm/lp_bld_nir_soa.c b/src/gallium/auxiliary/gallivm/lp_bld_nir_soa.c
index dde6831..9b72a9e 100644 (file)
--- a/src/gallium/auxiliary/gallivm/lp_bld_nir_soa.c
+++ b/src/gallium/auxiliary/gallivm/lp_bld_nir_soa.c
@@ -2220,28 +2220,27 @@ static void emit_read_invocation(struct lp_build_nir_context *bld_base,
    LLVMBuilderRef builder = gallivm->builder;
    LLVMValueRef idx;
    struct lp_build_context *uint_bld = get_int_bld(bld_base, true, bit_size);
-   if (invoc) {
-      idx = invoc;
-      idx = LLVMBuildExtractElement(gallivm->builder, idx, lp_build_const_int32(gallivm, 0), "");
-   } else {
-      /* have to find the first active invocation */
-      LLVMValueRef exec_mask = mask_vec(bld_base);
-      struct lp_build_loop_state loop_state;
-      LLVMValueRef res_store = lp_build_alloca(gallivm, bld_base->int_bld.elem_type, "");
-      LLVMValueRef outer_cond = LLVMBuildICmp(builder, LLVMIntNE, exec_mask, bld_base->uint_bld.zero, "");
-      lp_build_loop_begin(&loop_state, gallivm, lp_build_const_int32(gallivm, bld_base->uint_bld.type.length));
 
-      LLVMValueRef if_cond = LLVMBuildExtractElement(gallivm->builder, outer_cond, loop_state.counter, "");
-      struct lp_build_if_state ifthen;
+   /* have to find the first active invocation */
+   LLVMValueRef exec_mask = mask_vec(bld_base);
+   struct lp_build_loop_state loop_state;
+   LLVMValueRef res_store = lp_build_alloca(gallivm, bld_base->int_bld.elem_type, "");
+   LLVMValueRef outer_cond = LLVMBuildICmp(builder, LLVMIntNE, exec_mask, bld_base->uint_bld.zero, "");
+   lp_build_loop_begin(&loop_state, gallivm, lp_build_const_int32(gallivm, bld_base->uint_bld.type.length));
 
-      lp_build_if(&ifthen, gallivm, if_cond);
-      LLVMBuildStore(builder, loop_state.counter, res_store);
-      lp_build_endif(&ifthen);
+   LLVMValueRef if_cond = LLVMBuildExtractElement(gallivm->builder, outer_cond, loop_state.counter, "");
+   struct lp_build_if_state ifthen;
 
-      lp_build_loop_end_cond(&loop_state, lp_build_const_int32(gallivm, -1),
-                             lp_build_const_int32(gallivm, -1), LLVMIntEQ);
-      idx = LLVMBuildLoad(builder, res_store, "");
-   }
+   lp_build_if(&ifthen, gallivm, if_cond);
+   LLVMValueRef store_val = loop_state.counter;
+   if (invoc)
+      store_val = LLVMBuildExtractElement(gallivm->builder, invoc, loop_state.counter, "");
+   LLVMBuildStore(builder, store_val, res_store);
+   lp_build_endif(&ifthen);
+
+   lp_build_loop_end_cond(&loop_state, lp_build_const_int32(gallivm, -1),
+                          lp_build_const_int32(gallivm, -1), LLVMIntEQ);
+   idx = LLVMBuildLoad(builder, res_store, "");
 
    LLVMValueRef value = LLVMBuildExtractElement(gallivm->builder,
                                                 src, idx, "");

diff --git a/src/gallium/frontends/lavapipe/lvp_device.c b/src/gallium/frontends/lavapipe/lvp_device.c
index 745d448..e098596 100644 (file)
--- a/src/gallium/frontends/lavapipe/lvp_device.c
+++ b/src/gallium/frontends/lavapipe/lvp_device.c
@@ -521,7 +521,7 @@ lvp_get_physical_device_features_1_2(struct lvp_physical_device *pdevice,
    f->vulkanMemoryModelAvailabilityVisibilityChains = false;
    f->shaderOutputViewportIndex = true;
    f->shaderOutputLayer = true;
-   f->subgroupBroadcastDynamicId = false;
+   f->subgroupBroadcastDynamicId = true;
 }
 
 VKAPI_ATTR void VKAPI_CALL lvp_GetPhysicalDeviceFeatures2(