gallium/tgsi: start adding hw atomics (v3.2)

This adds support for a hw atomic counters to TGSI.

A new register file for storing atomic counters is added,
along with a new atomic counter semantic, along with docs
for both.

v2: drop semantic, move hw counter to backend,
Ilia pointed out SSO would have busted my plan, and he
was right.
v3: drop BUFFER decls. (Marek)
v3.1: minor fixups for whitespace, set ureg error
if we overflow the hw atomic limits. (nha)
v3.2: fix some docs inconsistencies (Ilia)

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Tested-By: Gert Wollny <gw.fossdev@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

src/gallium/auxiliary/tgsi/tgsi_strings.c

@@ -58,6 +58,7 @@ static const char *tgsi_file_names[] =
    "BUFFER",
    "MEMORY",
    "CONSTBUF",
+   "HWATOMIC",
 };
 
 const char *tgsi_semantic_names[TGSI_SEMANTIC_COUNT] =

src/gallium/auxiliary/tgsi/tgsi_ureg.c

@@ -80,6 +80,7 @@ struct ureg_tokens {
 #define UREG_MAX_SYSTEM_VALUE PIPE_MAX_ATTRIBS
 #define UREG_MAX_OUTPUT (4 * PIPE_MAX_SHADER_OUTPUTS)
 #define UREG_MAX_CONSTANT_RANGE 32
+#define UREG_MAX_HW_ATOMIC_RANGE 32
 #define UREG_MAX_IMMEDIATE 4096
 #define UREG_MAX_ADDR 3
 #define UREG_MAX_ARRAY_TEMPS 256
@@ -92,6 +93,15 @@ struct const_decl {
    unsigned nr_constant_ranges;
 };
 
+struct hw_atomic_decl {
+   struct {
+      unsigned first;
+      unsigned last;
+      unsigned array_id;
+   } hw_atomic_range[UREG_MAX_HW_ATOMIC_RANGE];
+   unsigned nr_hw_atomic_ranges;
+};
+
 #define DOMAIN_DECL 0
 #define DOMAIN_INSN 1
 
@@ -182,6 +192,8 @@ struct ureg_program
 
    struct const_decl const_decls[PIPE_MAX_CONSTANT_BUFFERS];
 
+   struct hw_atomic_decl hw_atomic_decls[PIPE_MAX_HW_ATOMIC_BUFFERS];
+
    unsigned properties[TGSI_PROPERTY_COUNT];
 
    unsigned nr_addrs;
@@ -583,6 +595,30 @@ out:
    return ureg_src_dimension(src, 0);
 }
 
+
+/* Returns a new hw atomic register.  Keep track of which have been
+ * referred to so that we can emit decls later.
+ */
+void
+ureg_DECL_hw_atomic(struct ureg_program *ureg,
+                    unsigned first,
+                    unsigned last,
+                    unsigned buffer_id,
+                    unsigned array_id)
+{
+   struct hw_atomic_decl *decl = &ureg->hw_atomic_decls[buffer_id];
+
+   if (decl->nr_hw_atomic_ranges < UREG_MAX_HW_ATOMIC_RANGE) {
+      uint i = decl->nr_hw_atomic_ranges++;
+
+      decl->hw_atomic_range[i].first = first;
+      decl->hw_atomic_range[i].last = last;
+      decl->hw_atomic_range[i].array_id = array_id;
+   } else {
+      set_bad(ureg);
+   }
+}
+
 static struct ureg_dst alloc_temporary( struct ureg_program *ureg,
                                         boolean local )
 {
@@ -1501,6 +1537,35 @@ emit_decl_semantic(struct ureg_program *ureg,
    }
 }
 
+static void
+emit_decl_atomic_2d(struct ureg_program *ureg,
+                    unsigned first,
+                    unsigned last,
+                    unsigned index2D,
+                    unsigned array_id)
+{
+   union tgsi_any_token *out = get_tokens(ureg, DOMAIN_DECL, array_id ? 4 : 3);
+
+   out[0].value = 0;
+   out[0].decl.Type = TGSI_TOKEN_TYPE_DECLARATION;
+   out[0].decl.NrTokens = 3;
+   out[0].decl.File = TGSI_FILE_HW_ATOMIC;
+   out[0].decl.UsageMask = TGSI_WRITEMASK_XYZW;
+   out[0].decl.Dimension = 1;
+   out[0].decl.Array = array_id != 0;
+
+   out[1].value = 0;
+   out[1].decl_range.First = first;
+   out[1].decl_range.Last = last;
+
+   out[2].value = 0;
+   out[2].decl_dim.Index2D = index2D;
+
+   if (array_id) {
+      out[3].value = 0;
+      out[3].array.ArrayID = array_id;
+   }
+}
 
 static void
 emit_decl_fs(struct ureg_program *ureg,
@@ -1908,6 +1973,22 @@ static void emit_decls( struct ureg_program *ureg )
       }
    }
 
+   for (i = 0; i < PIPE_MAX_HW_ATOMIC_BUFFERS; i++) {
+      struct hw_atomic_decl *decl = &ureg->hw_atomic_decls[i];
+
+      if (decl->nr_hw_atomic_ranges) {
+         uint j;
+
+         for (j = 0; j < decl->nr_hw_atomic_ranges; j++) {
+            emit_decl_atomic_2d(ureg,
+                                decl->hw_atomic_range[j].first,
+                                decl->hw_atomic_range[j].last,
+                                i,
+                                decl->hw_atomic_range[j].array_id);
+         }
+      }
+   }
+
    if (ureg->nr_temps) {
       unsigned array = 0;
       for (i = 0; i < ureg->nr_temps;) {

src/gallium/auxiliary/tgsi/tgsi_ureg.h

@@ -316,6 +316,13 @@ struct ureg_src
 ureg_DECL_constant( struct ureg_program *,
                     unsigned index );
 
+void
+ureg_DECL_hw_atomic(struct ureg_program *ureg,
+                    unsigned first,
+                    unsigned last,
+                    unsigned buffer_id,
+                    unsigned array_id);
+
 struct ureg_dst
 ureg_DECL_temporary( struct ureg_program * );
 

src/gallium/docs/source/tgsi.rst

@@ -2638,9 +2638,11 @@ logical operations.  In this context atomicity means that another
 concurrent memory access operation that affects the same memory
 location is guaranteed to be performed strictly before or after the
 entire execution of the atomic operation. The resource may be a BUFFER,
-IMAGE, or MEMORY.  In the case of an image, the offset works the same as for
-``LOAD`` and ``STORE``, specified above. These atomic operations may
-only be used with 32-bit integer image formats.
+IMAGE, HWATOMIC, or MEMORY.  In the case of an image, the offset works
+the same as for ``LOAD`` and ``STORE``, specified above. For atomic
+counters, the offset is an immediate index to the base hw atomic
+counter for this operation.
+These atomic operations may only be used with 32-bit integer image formats.
 
 .. opcode:: ATOMUADD - Atomic integer addition
 
@@ -3517,6 +3519,31 @@ accessing a misaligned address is undefined.
 Usage of the STORE opcode is only allowed if the WR (writable) flag
 is set.
 
+Hardware Atomic Register File
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+Hardware atomics are declared as a 2D array with an optional array id.
+
+The first member of the dimension is the buffer resource the atomic
+is located in.
+The second member is a range into the buffer resource, either for
+one or multiple counters. If this is an array, the declaration will have
+an unique array id.
+
+Each counter is 4 bytes in size, and index and ranges are in counters not bytes.
+DCL HWATOMIC[0][0]
+DCL HWATOMIC[0][1]
+
+This declares two atomics, one at the start of the buffer and one in the
+second 4 bytes.
+
+DCL HWATOMIC[0][0]
+DCL HWATOMIC[1][0]
+DCL HWATOMIC[1][1..3], ARRAY(1)
+
+This declares 5 atomics, one in buffer 0 at 0,
+one in buffer 1 at 0, and an array of 3 atomics in
+the buffer 1, starting at 1.
 
 Properties
 ^^^^^^^^^^^^^^^^^^^^^^^^

src/gallium/include/pipe/p_shader_tokens.h

@@ -75,6 +75,7 @@ enum tgsi_file_type {
    TGSI_FILE_BUFFER,
    TGSI_FILE_MEMORY,
    TGSI_FILE_CONSTBUF,
+   TGSI_FILE_HW_ATOMIC,
    TGSI_FILE_COUNT,      /**< how many TGSI_FILE_ types */
 };
 

src/gallium/include/pipe/p_state.h

@@ -75,6 +75,7 @@ extern "C" {
 #define PIPE_MAX_CLIP_OR_CULL_DISTANCE_ELEMENT_COUNT 2
 #define PIPE_MAX_WINDOW_RECTANGLES 8
 
+#define PIPE_MAX_HW_ATOMIC_BUFFERS 32
 
 struct pipe_reference
 {