src/gallium/drivers/radeon/r600_query.c

/*
 * Copyright 2010 Jerome Glisse <glisse@freedesktop.org>
 * Copyright 2014 Marek Olšák <marek.olsak@amd.com>
 *
 * Permission is hereby granted, free of charge, to any person obtaining a
 * copy of this software and associated documentation files (the "Software"),
 * to deal in the Software without restriction, including without limitation
 * on the rights to use, copy, modify, merge, publish, distribute, sub
 * license, and/or sell copies of the Software, and to permit persons to whom
 * the Software is furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice (including the next
 * paragraph) shall be included in all copies or substantial portions of the
 * Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
 * THE AUTHOR(S) AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
 * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
 * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
 * USE OR OTHER DEALINGS IN THE SOFTWARE.
 */

#include "r600_query.h"
#include "r600_cs.h"
#include "util/u_memory.h"
#include "util/u_upload_mgr.h"

#include "tgsi/tgsi_text.h"

struct r600_hw_query_params {
	unsigned start_offset;
	unsigned end_offset;
	unsigned fence_offset;
	unsigned pair_stride;
	unsigned pair_count;
};

/* Queries without buffer handling or suspend/resume. */
struct r600_query_sw {
	struct r600_query b;

	uint64_t begin_result;
	uint64_t end_result;
	/* Fence for GPU_FINISHED. */
	struct pipe_fence_handle *fence;
};

static void r600_query_sw_destroy(struct r600_common_context *rctx,
				  struct r600_query *rquery)
{
	struct pipe_screen *screen = rctx->b.screen;
	struct r600_query_sw *query = (struct r600_query_sw *)rquery;

	screen->fence_reference(screen, &query->fence, NULL);
	FREE(query);
}

static enum radeon_value_id winsys_id_from_type(unsigned type)
{
	switch (type) {
	case R600_QUERY_REQUESTED_VRAM: return RADEON_REQUESTED_VRAM_MEMORY;
	case R600_QUERY_REQUESTED_GTT: return RADEON_REQUESTED_GTT_MEMORY;
	case R600_QUERY_MAPPED_VRAM: return RADEON_MAPPED_VRAM;
	case R600_QUERY_MAPPED_GTT: return RADEON_MAPPED_GTT;
	case R600_QUERY_BUFFER_WAIT_TIME: return RADEON_BUFFER_WAIT_TIME_NS;
	case R600_QUERY_NUM_GFX_IBS: return RADEON_NUM_GFX_IBS;
	case R600_QUERY_NUM_SDMA_IBS: return RADEON_NUM_SDMA_IBS;
	case R600_QUERY_NUM_BYTES_MOVED: return RADEON_NUM_BYTES_MOVED;
	case R600_QUERY_NUM_EVICTIONS: return RADEON_NUM_EVICTIONS;
	case R600_QUERY_VRAM_USAGE: return RADEON_VRAM_USAGE;
	case R600_QUERY_GTT_USAGE: return RADEON_GTT_USAGE;
	case R600_QUERY_GPU_TEMPERATURE: return RADEON_GPU_TEMPERATURE;
	case R600_QUERY_CURRENT_GPU_SCLK: return RADEON_CURRENT_SCLK;
	case R600_QUERY_CURRENT_GPU_MCLK: return RADEON_CURRENT_MCLK;
	default: unreachable("query type does not correspond to winsys id");
	}
}

static bool r600_query_sw_begin(struct r600_common_context *rctx,
				struct r600_query *rquery)
{
	struct r600_query_sw *query = (struct r600_query_sw *)rquery;

	switch(query->b.type) {
	case PIPE_QUERY_TIMESTAMP_DISJOINT:
	case PIPE_QUERY_GPU_FINISHED:
		break;
	case R600_QUERY_DRAW_CALLS:
		query->begin_result = rctx->num_draw_calls;
		break;
	case R600_QUERY_SPILL_DRAW_CALLS:
		query->begin_result = rctx->num_spill_draw_calls;
		break;
	case R600_QUERY_COMPUTE_CALLS:
		query->begin_result = rctx->num_compute_calls;
		break;
	case R600_QUERY_SPILL_COMPUTE_CALLS:
		query->begin_result = rctx->num_spill_compute_calls;
		break;
	case R600_QUERY_DMA_CALLS:
		query->begin_result = rctx->num_dma_calls;
		break;
	case R600_QUERY_CP_DMA_CALLS:
		query->begin_result = rctx->num_cp_dma_calls;
		break;
	case R600_QUERY_NUM_VS_FLUSHES:
		query->begin_result = rctx->num_vs_flushes;
		break;
	case R600_QUERY_NUM_PS_FLUSHES:
		query->begin_result = rctx->num_ps_flushes;
		break;
	case R600_QUERY_NUM_CS_FLUSHES:
		query->begin_result = rctx->num_cs_flushes;
		break;
	case R600_QUERY_NUM_FB_CACHE_FLUSHES:
		query->begin_result = rctx->num_fb_cache_flushes;
		break;
	case R600_QUERY_NUM_L2_INVALIDATES:
		query->begin_result = rctx->num_L2_invalidates;
		break;
	case R600_QUERY_NUM_L2_WRITEBACKS:
		query->begin_result = rctx->num_L2_writebacks;
		break;
	case R600_QUERY_REQUESTED_VRAM:
	case R600_QUERY_REQUESTED_GTT:
	case R600_QUERY_MAPPED_VRAM:
	case R600_QUERY_MAPPED_GTT:
	case R600_QUERY_VRAM_USAGE:
	case R600_QUERY_GTT_USAGE:
	case R600_QUERY_GPU_TEMPERATURE:
	case R600_QUERY_CURRENT_GPU_SCLK:
	case R600_QUERY_CURRENT_GPU_MCLK:
	case R600_QUERY_BACK_BUFFER_PS_DRAW_RATIO:
		query->begin_result = 0;
		break;
	case R600_QUERY_BUFFER_WAIT_TIME:
	case R600_QUERY_NUM_GFX_IBS:
	case R600_QUERY_NUM_SDMA_IBS:
	case R600_QUERY_NUM_BYTES_MOVED:
	case R600_QUERY_NUM_EVICTIONS: {
		enum radeon_value_id ws_id = winsys_id_from_type(query->b.type);
		query->begin_result = rctx->ws->query_value(rctx->ws, ws_id);
		break;
	}
	case R600_QUERY_GPU_LOAD:
		query->begin_result = r600_begin_counter_gui(rctx->screen);
		break;
	case R600_QUERY_GPU_SHADERS_BUSY:
		query->begin_result = r600_begin_counter_spi(rctx->screen);
		break;
	case R600_QUERY_NUM_COMPILATIONS:
		query->begin_result = p_atomic_read(&rctx->screen->num_compilations);
		break;
	case R600_QUERY_NUM_SHADERS_CREATED:
		query->begin_result = p_atomic_read(&rctx->screen->num_shaders_created);
		break;
	case R600_QUERY_NUM_SHADER_CACHE_HITS:
		query->begin_result =
			p_atomic_read(&rctx->screen->num_shader_cache_hits);
		break;
	case R600_QUERY_GPIN_ASIC_ID:
	case R600_QUERY_GPIN_NUM_SIMD:
	case R600_QUERY_GPIN_NUM_RB:
	case R600_QUERY_GPIN_NUM_SPI:
	case R600_QUERY_GPIN_NUM_SE:
		break;
	default:
		unreachable("r600_query_sw_begin: bad query type");
	}

	return true;
}

static bool r600_query_sw_end(struct r600_common_context *rctx,
			      struct r600_query *rquery)
{
	struct r600_query_sw *query = (struct r600_query_sw *)rquery;

	switch(query->b.type) {
	case PIPE_QUERY_TIMESTAMP_DISJOINT:
		break;
	case PIPE_QUERY_GPU_FINISHED:
		rctx->b.flush(&rctx->b, &query->fence, PIPE_FLUSH_DEFERRED);
		break;
	case R600_QUERY_DRAW_CALLS:
		query->end_result = rctx->num_draw_calls;
		break;
	case R600_QUERY_SPILL_DRAW_CALLS:
		query->end_result = rctx->num_spill_draw_calls;
		break;
	case R600_QUERY_COMPUTE_CALLS:
		query->end_result = rctx->num_compute_calls;
		break;
	case R600_QUERY_SPILL_COMPUTE_CALLS:
		query->end_result = rctx->num_spill_compute_calls;
		break;
	case R600_QUERY_DMA_CALLS:
		query->end_result = rctx->num_dma_calls;
		break;
	case R600_QUERY_CP_DMA_CALLS:
		query->end_result = rctx->num_cp_dma_calls;
		break;
	case R600_QUERY_NUM_VS_FLUSHES:
		query->end_result = rctx->num_vs_flushes;
		break;
	case R600_QUERY_NUM_PS_FLUSHES:
		query->end_result = rctx->num_ps_flushes;
		break;
	case R600_QUERY_NUM_CS_FLUSHES:
		query->end_result = rctx->num_cs_flushes;
		break;
	case R600_QUERY_NUM_FB_CACHE_FLUSHES:
		query->end_result = rctx->num_fb_cache_flushes;
		break;
	case R600_QUERY_NUM_L2_INVALIDATES:
		query->end_result = rctx->num_L2_invalidates;
		break;
	case R600_QUERY_NUM_L2_WRITEBACKS:
		query->end_result = rctx->num_L2_writebacks;
		break;
	case R600_QUERY_REQUESTED_VRAM:
	case R600_QUERY_REQUESTED_GTT:
	case R600_QUERY_MAPPED_VRAM:
	case R600_QUERY_MAPPED_GTT:
	case R600_QUERY_VRAM_USAGE:
	case R600_QUERY_GTT_USAGE:
	case R600_QUERY_GPU_TEMPERATURE:
	case R600_QUERY_CURRENT_GPU_SCLK:
	case R600_QUERY_CURRENT_GPU_MCLK:
	case R600_QUERY_BUFFER_WAIT_TIME:
	case R600_QUERY_NUM_GFX_IBS:
	case R600_QUERY_NUM_SDMA_IBS:
	case R600_QUERY_NUM_BYTES_MOVED:
	case R600_QUERY_NUM_EVICTIONS: {
		enum radeon_value_id ws_id = winsys_id_from_type(query->b.type);
		query->end_result = rctx->ws->query_value(rctx->ws, ws_id);
		break;
	}
	case R600_QUERY_GPU_LOAD:
		query->end_result = r600_end_counter_gui(rctx->screen,
							 query->begin_result);
		query->begin_result = 0;
		break;
	case R600_QUERY_GPU_SHADERS_BUSY:
		query->end_result = r600_end_counter_spi(rctx->screen,
							 query->begin_result);
		query->begin_result = 0;
		break;
	case R600_QUERY_NUM_COMPILATIONS:
		query->end_result = p_atomic_read(&rctx->screen->num_compilations);
		break;
	case R600_QUERY_NUM_SHADERS_CREATED:
		query->end_result = p_atomic_read(&rctx->screen->num_shaders_created);
		break;
	case R600_QUERY_BACK_BUFFER_PS_DRAW_RATIO:
		query->end_result = rctx->last_tex_ps_draw_ratio;
		break;
	case R600_QUERY_NUM_SHADER_CACHE_HITS:
		query->end_result =
			p_atomic_read(&rctx->screen->num_shader_cache_hits);
		break;
	case R600_QUERY_GPIN_ASIC_ID:
	case R600_QUERY_GPIN_NUM_SIMD:
	case R600_QUERY_GPIN_NUM_RB:
	case R600_QUERY_GPIN_NUM_SPI:
	case R600_QUERY_GPIN_NUM_SE:
		break;
	default:
		unreachable("r600_query_sw_end: bad query type");
	}

	return true;
}

static bool r600_query_sw_get_result(struct r600_common_context *rctx,
				     struct r600_query *rquery,
				     bool wait,
				     union pipe_query_result *result)
{
	struct r600_query_sw *query = (struct r600_query_sw *)rquery;

	switch (query->b.type) {
	case PIPE_QUERY_TIMESTAMP_DISJOINT:
		/* Convert from cycles per millisecond to cycles per second (Hz). */
		result->timestamp_disjoint.frequency =
			(uint64_t)rctx->screen->info.clock_crystal_freq * 1000;
		result->timestamp_disjoint.disjoint = false;
		return true;
	case PIPE_QUERY_GPU_FINISHED: {
		struct pipe_screen *screen = rctx->b.screen;
		result->b = screen->fence_finish(screen, &rctx->b, query->fence,
						 wait ? PIPE_TIMEOUT_INFINITE : 0);
		return result->b;
	}

	case R600_QUERY_GPIN_ASIC_ID:
		result->u32 = 0;
		return true;
	case R600_QUERY_GPIN_NUM_SIMD:
		result->u32 = rctx->screen->info.num_good_compute_units;
		return true;
	case R600_QUERY_GPIN_NUM_RB:
		result->u32 = rctx->screen->info.num_render_backends;
		return true;
	case R600_QUERY_GPIN_NUM_SPI:
		result->u32 = 1; /* all supported chips have one SPI per SE */
		return true;
	case R600_QUERY_GPIN_NUM_SE:
		result->u32 = rctx->screen->info.max_se;
		return true;
	}

	result->u64 = query->end_result - query->begin_result;

	switch (query->b.type) {
	case R600_QUERY_BUFFER_WAIT_TIME:
	case R600_QUERY_GPU_TEMPERATURE:
		result->u64 /= 1000;
		break;
	case R600_QUERY_CURRENT_GPU_SCLK:
	case R600_QUERY_CURRENT_GPU_MCLK:
		result->u64 *= 1000000;
		break;
	}

	return true;
}


static struct r600_query_ops sw_query_ops = {
	.destroy = r600_query_sw_destroy,
	.begin = r600_query_sw_begin,
	.end = r600_query_sw_end,
	.get_result = r600_query_sw_get_result,
	.get_result_resource = NULL
};

static struct pipe_query *r600_query_sw_create(struct pipe_context *ctx,
					       unsigned query_type)
{
	struct r600_query_sw *query;

	query = CALLOC_STRUCT(r600_query_sw);
	if (!query)
		return NULL;

	query->b.type = query_type;
	query->b.ops = &sw_query_ops;

	return (struct pipe_query *)query;
}

void r600_query_hw_destroy(struct r600_common_context *rctx,
			   struct r600_query *rquery)
{
	struct r600_query_hw *query = (struct r600_query_hw *)rquery;
	struct r600_query_buffer *prev = query->buffer.previous;

	/* Release all query buffers. */
	while (prev) {
		struct r600_query_buffer *qbuf = prev;
		prev = prev->previous;
		r600_resource_reference(&qbuf->buf, NULL);
		FREE(qbuf);
	}

	r600_resource_reference(&query->buffer.buf, NULL);
	FREE(rquery);
}

static struct r600_resource *r600_new_query_buffer(struct r600_common_context *ctx,
						   struct r600_query_hw *query)
{
	unsigned buf_size = MAX2(query->result_size,
				 ctx->screen->info.min_alloc_size);

	/* Queries are normally read by the CPU after
	 * being written by the gpu, hence staging is probably a good
	 * usage pattern.
	 */
	struct r600_resource *buf = (struct r600_resource*)
		pipe_buffer_create(ctx->b.screen, 0,
				   PIPE_USAGE_STAGING, buf_size);
	if (!buf)
		return NULL;

	if (!query->ops->prepare_buffer(ctx, query, buf)) {
		r600_resource_reference(&buf, NULL);
		return NULL;
	}

	return buf;
}

static bool r600_query_hw_prepare_buffer(struct r600_common_context *ctx,
					 struct r600_query_hw *query,
					 struct r600_resource *buffer)
{
	/* Callers ensure that the buffer is currently unused by the GPU. */
	uint32_t *results = ctx->ws->buffer_map(buffer->buf, NULL,
						PIPE_TRANSFER_WRITE |
						PIPE_TRANSFER_UNSYNCHRONIZED);
	if (!results)
		return false;

	memset(results, 0, buffer->b.b.width0);

	if (query->b.type == PIPE_QUERY_OCCLUSION_COUNTER ||
	    query->b.type == PIPE_QUERY_OCCLUSION_PREDICATE) {
		unsigned num_results;
		unsigned i, j;

		/* Set top bits for unused backends. */
		num_results = buffer->b.b.width0 / query->result_size;
		for (j = 0; j < num_results; j++) {
			for (i = 0; i < ctx->max_db; i++) {
				if (!(ctx->backend_mask & (1<<i))) {
					results[(i * 4)+1] = 0x80000000;
					results[(i * 4)+3] = 0x80000000;
				}
			}
			results += 4 * ctx->max_db;
		}
	}

	return true;
}

static void r600_query_hw_get_result_resource(struct r600_common_context *rctx,
                                              struct r600_query *rquery,
                                              bool wait,
                                              enum pipe_query_value_type result_type,
                                              int index,
                                              struct pipe_resource *resource,
                                              unsigned offset);

static struct r600_query_ops query_hw_ops = {
	.destroy = r600_query_hw_destroy,
	.begin = r600_query_hw_begin,
	.end = r600_query_hw_end,
	.get_result = r600_query_hw_get_result,
	.get_result_resource = r600_query_hw_get_result_resource,
};

static void r600_query_hw_do_emit_start(struct r600_common_context *ctx,
					struct r600_query_hw *query,
					struct r600_resource *buffer,
					uint64_t va);
static void r600_query_hw_do_emit_stop(struct r600_common_context *ctx,
				       struct r600_query_hw *query,
				       struct r600_resource *buffer,
				       uint64_t va);
static void r600_query_hw_add_result(struct r600_common_context *ctx,
				     struct r600_query_hw *, void *buffer,
				     union pipe_query_result *result);
static void r600_query_hw_clear_result(struct r600_query_hw *,
				       union pipe_query_result *);

static struct r600_query_hw_ops query_hw_default_hw_ops = {
	.prepare_buffer = r600_query_hw_prepare_buffer,
	.emit_start = r600_query_hw_do_emit_start,
	.emit_stop = r600_query_hw_do_emit_stop,
	.clear_result = r600_query_hw_clear_result,
	.add_result = r600_query_hw_add_result,
};

bool r600_query_hw_init(struct r600_common_context *rctx,
			struct r600_query_hw *query)
{
	query->buffer.buf = r600_new_query_buffer(rctx, query);
	if (!query->buffer.buf)
		return false;

	return true;
}

static struct pipe_query *r600_query_hw_create(struct r600_common_context *rctx,
					       unsigned query_type,
					       unsigned index)
{
	struct r600_query_hw *query = CALLOC_STRUCT(r600_query_hw);
	if (!query)
		return NULL;

	query->b.type = query_type;
	query->b.ops = &query_hw_ops;
	query->ops = &query_hw_default_hw_ops;

	switch (query_type) {
	case PIPE_QUERY_OCCLUSION_COUNTER:
	case PIPE_QUERY_OCCLUSION_PREDICATE:
		query->result_size = 16 * rctx->max_db;
		query->result_size += 16; /* for the fence + alignment */
		query->num_cs_dw_begin = 6;
		query->num_cs_dw_end = 6 + r600_gfx_write_fence_dwords(rctx->screen);
		break;
	case PIPE_QUERY_TIME_ELAPSED:
		query->result_size = 24;
		query->num_cs_dw_begin = 8;
		query->num_cs_dw_end = 8 + r600_gfx_write_fence_dwords(rctx->screen);
		break;
	case PIPE_QUERY_TIMESTAMP:
		query->result_size = 16;
		query->num_cs_dw_end = 8 + r600_gfx_write_fence_dwords(rctx->screen);
		query->flags = R600_QUERY_HW_FLAG_NO_START;
		break;
	case PIPE_QUERY_PRIMITIVES_EMITTED:
	case PIPE_QUERY_PRIMITIVES_GENERATED:
	case PIPE_QUERY_SO_STATISTICS:
	case PIPE_QUERY_SO_OVERFLOW_PREDICATE:
		/* NumPrimitivesWritten, PrimitiveStorageNeeded. */
		query->result_size = 32;
		query->num_cs_dw_begin = 6;
		query->num_cs_dw_end = 6;
		query->stream = index;
		break;
	case PIPE_QUERY_PIPELINE_STATISTICS:
		/* 11 values on EG, 8 on R600. */
		query->result_size = (rctx->chip_class >= EVERGREEN ? 11 : 8) * 16;
		query->result_size += 8; /* for the fence + alignment */
		query->num_cs_dw_begin = 6;
		query->num_cs_dw_end = 6 + r600_gfx_write_fence_dwords(rctx->screen);
		break;
	default:
		assert(0);
		FREE(query);
		return NULL;
	}

	if (!r600_query_hw_init(rctx, query)) {
		FREE(query);
		return NULL;
	}

	return (struct pipe_query *)query;
}

static void r600_update_occlusion_query_state(struct r600_common_context *rctx,
					      unsigned type, int diff)
{
	if (type == PIPE_QUERY_OCCLUSION_COUNTER ||
	    type == PIPE_QUERY_OCCLUSION_PREDICATE) {
		bool old_enable = rctx->num_occlusion_queries != 0;
		bool old_perfect_enable =
			rctx->num_perfect_occlusion_queries != 0;
		bool enable, perfect_enable;

		rctx->num_occlusion_queries += diff;
		assert(rctx->num_occlusion_queries >= 0);

		if (type == PIPE_QUERY_OCCLUSION_COUNTER) {
			rctx->num_perfect_occlusion_queries += diff;
			assert(rctx->num_perfect_occlusion_queries >= 0);
		}

		enable = rctx->num_occlusion_queries != 0;
		perfect_enable = rctx->num_perfect_occlusion_queries != 0;

		if (enable != old_enable || perfect_enable != old_perfect_enable) {
			rctx->set_occlusion_query_state(&rctx->b, enable);
		}
	}
}

static unsigned event_type_for_stream(struct r600_query_hw *query)
{
	switch (query->stream) {
	default:
	case 0: return EVENT_TYPE_SAMPLE_STREAMOUTSTATS;
	case 1: return EVENT_TYPE_SAMPLE_STREAMOUTSTATS1;
	case 2: return EVENT_TYPE_SAMPLE_STREAMOUTSTATS2;
	case 3: return EVENT_TYPE_SAMPLE_STREAMOUTSTATS3;
	}
}

static void r600_query_hw_do_emit_start(struct r600_common_context *ctx,
					struct r600_query_hw *query,
					struct r600_resource *buffer,
					uint64_t va)
{
	struct radeon_winsys_cs *cs = ctx->gfx.cs;

	switch (query->b.type) {
	case PIPE_QUERY_OCCLUSION_COUNTER:
	case PIPE_QUERY_OCCLUSION_PREDICATE:
		radeon_emit(cs, PKT3(PKT3_EVENT_WRITE, 2, 0));
		radeon_emit(cs, EVENT_TYPE(EVENT_TYPE_ZPASS_DONE) | EVENT_INDEX(1));
		radeon_emit(cs, va);
		radeon_emit(cs, (va >> 32) & 0xFFFF);
		break;
	case PIPE_QUERY_PRIMITIVES_EMITTED:
	case PIPE_QUERY_PRIMITIVES_GENERATED:
	case PIPE_QUERY_SO_STATISTICS:
	case PIPE_QUERY_SO_OVERFLOW_PREDICATE:
		radeon_emit(cs, PKT3(PKT3_EVENT_WRITE, 2, 0));
		radeon_emit(cs, EVENT_TYPE(event_type_for_stream(query)) | EVENT_INDEX(3));
		radeon_emit(cs, va);
		radeon_emit(cs, (va >> 32) & 0xFFFF);
		break;
	case PIPE_QUERY_TIME_ELAPSED:
		r600_gfx_write_event_eop(ctx, EVENT_TYPE_BOTTOM_OF_PIPE_TS,
					 0, 3, NULL, va, 0, 0);
		break;
	case PIPE_QUERY_PIPELINE_STATISTICS:
		radeon_emit(cs, PKT3(PKT3_EVENT_WRITE, 2, 0));
		radeon_emit(cs, EVENT_TYPE(EVENT_TYPE_SAMPLE_PIPELINESTAT) | EVENT_INDEX(2));
		radeon_emit(cs, va);
		radeon_emit(cs, (va >> 32) & 0xFFFF);
		break;
	default:
		assert(0);
	}
	r600_emit_reloc(ctx, &ctx->gfx, query->buffer.buf, RADEON_USAGE_WRITE,
			RADEON_PRIO_QUERY);
}

static void r600_query_hw_emit_start(struct r600_common_context *ctx,
				     struct r600_query_hw *query)
{
	uint64_t va;

	if (!query->buffer.buf)
		return; // previous buffer allocation failure

	r600_update_occlusion_query_state(ctx, query->b.type, 1);
	r600_update_prims_generated_query_state(ctx, query->b.type, 1);

	ctx->need_gfx_cs_space(&ctx->b, query->num_cs_dw_begin + query->num_cs_dw_end,
			       true);

	/* Get a new query buffer if needed. */
	if (query->buffer.results_end + query->result_size > query->buffer.buf->b.b.width0) {
		struct r600_query_buffer *qbuf = MALLOC_STRUCT(r600_query_buffer);
		*qbuf = query->buffer;
		query->buffer.results_end = 0;
		query->buffer.previous = qbuf;
		query->buffer.buf = r600_new_query_buffer(ctx, query);
		if (!query->buffer.buf)
			return;
	}

	/* emit begin query */
	va = query->buffer.buf->gpu_address + query->buffer.results_end;

	query->ops->emit_start(ctx, query, query->buffer.buf, va);

	ctx->num_cs_dw_queries_suspend += query->num_cs_dw_end;
}

static void r600_query_hw_do_emit_stop(struct r600_common_context *ctx,
				       struct r600_query_hw *query,
				       struct r600_resource *buffer,
				       uint64_t va)
{
	struct radeon_winsys_cs *cs = ctx->gfx.cs;
	uint64_t fence_va = 0;

	switch (query->b.type) {
	case PIPE_QUERY_OCCLUSION_COUNTER:
	case PIPE_QUERY_OCCLUSION_PREDICATE:
		va += 8;
		radeon_emit(cs, PKT3(PKT3_EVENT_WRITE, 2, 0));
		radeon_emit(cs, EVENT_TYPE(EVENT_TYPE_ZPASS_DONE) | EVENT_INDEX(1));
		radeon_emit(cs, va);
		radeon_emit(cs, (va >> 32) & 0xFFFF);

		fence_va = va + ctx->max_db * 16 - 8;
		break;
	case PIPE_QUERY_PRIMITIVES_EMITTED:
	case PIPE_QUERY_PRIMITIVES_GENERATED:
	case PIPE_QUERY_SO_STATISTICS:
	case PIPE_QUERY_SO_OVERFLOW_PREDICATE:
		va += query->result_size/2;
		radeon_emit(cs, PKT3(PKT3_EVENT_WRITE, 2, 0));
		radeon_emit(cs, EVENT_TYPE(event_type_for_stream(query)) | EVENT_INDEX(3));
		radeon_emit(cs, va);
		radeon_emit(cs, (va >> 32) & 0xFFFF);
		break;
	case PIPE_QUERY_TIME_ELAPSED:
		va += 8;
		/* fall through */
	case PIPE_QUERY_TIMESTAMP:
		r600_gfx_write_event_eop(ctx, EVENT_TYPE_BOTTOM_OF_PIPE_TS,
					 0, 3, NULL, va, 0, 0);
		fence_va = va + 8;
		break;
	case PIPE_QUERY_PIPELINE_STATISTICS: {
		unsigned sample_size = (query->result_size - 8) / 2;

		va += sample_size;
		radeon_emit(cs, PKT3(PKT3_EVENT_WRITE, 2, 0));
		radeon_emit(cs, EVENT_TYPE(EVENT_TYPE_SAMPLE_PIPELINESTAT) | EVENT_INDEX(2));
		radeon_emit(cs, va);
		radeon_emit(cs, (va >> 32) & 0xFFFF);

		fence_va = va + sample_size;
		break;
	}
	default:
		assert(0);
	}
	r600_emit_reloc(ctx, &ctx->gfx, query->buffer.buf, RADEON_USAGE_WRITE,
			RADEON_PRIO_QUERY);

	if (fence_va)
		r600_gfx_write_event_eop(ctx, EVENT_TYPE_BOTTOM_OF_PIPE_TS, 0, 1,
					 query->buffer.buf, fence_va, 0, 0x80000000);
}

static void r600_query_hw_emit_stop(struct r600_common_context *ctx,
				    struct r600_query_hw *query)
{
	uint64_t va;

	if (!query->buffer.buf)
		return; // previous buffer allocation failure

	/* The queries which need begin already called this in begin_query. */
	if (query->flags & R600_QUERY_HW_FLAG_NO_START) {
		ctx->need_gfx_cs_space(&ctx->b, query->num_cs_dw_end, false);
	}

	/* emit end query */
	va = query->buffer.buf->gpu_address + query->buffer.results_end;

	query->ops->emit_stop(ctx, query, query->buffer.buf, va);

	query->buffer.results_end += query->result_size;

	if (!(query->flags & R600_QUERY_HW_FLAG_NO_START))
		ctx->num_cs_dw_queries_suspend -= query->num_cs_dw_end;

	r600_update_occlusion_query_state(ctx, query->b.type, -1);
	r600_update_prims_generated_query_state(ctx, query->b.type, -1);
}

static void r600_emit_query_predication(struct r600_common_context *ctx,
					struct r600_atom *atom)
{
	struct radeon_winsys_cs *cs = ctx->gfx.cs;
	struct r600_query_hw *query = (struct r600_query_hw *)ctx->render_cond;
	struct r600_query_buffer *qbuf;
	uint32_t op;
	bool flag_wait;

	if (!query)
		return;

	flag_wait = ctx->render_cond_mode == PIPE_RENDER_COND_WAIT ||
		    ctx->render_cond_mode == PIPE_RENDER_COND_BY_REGION_WAIT;

	switch (query->b.type) {
	case PIPE_QUERY_OCCLUSION_COUNTER:
	case PIPE_QUERY_OCCLUSION_PREDICATE:
		op = PRED_OP(PREDICATION_OP_ZPASS);
		break;
	case PIPE_QUERY_PRIMITIVES_EMITTED:
	case PIPE_QUERY_PRIMITIVES_GENERATED:
	case PIPE_QUERY_SO_STATISTICS:
	case PIPE_QUERY_SO_OVERFLOW_PREDICATE:
		op = PRED_OP(PREDICATION_OP_PRIMCOUNT);
		break;
	default:
		assert(0);
		return;
	}

	/* if true then invert, see GL_ARB_conditional_render_inverted */
	if (ctx->render_cond_invert)
		op |= PREDICATION_DRAW_NOT_VISIBLE; /* Draw if not visable/overflow */
	else
		op |= PREDICATION_DRAW_VISIBLE; /* Draw if visable/overflow */

	op |= flag_wait ? PREDICATION_HINT_WAIT : PREDICATION_HINT_NOWAIT_DRAW;
	
	/* emit predicate packets for all data blocks */
	for (qbuf = &query->buffer; qbuf; qbuf = qbuf->previous) {
		unsigned results_base = 0;
		uint64_t va = qbuf->buf->gpu_address;

		while (results_base < qbuf->results_end) {
			radeon_emit(cs, PKT3(PKT3_SET_PREDICATION, 1, 0));
			radeon_emit(cs, va + results_base);
			radeon_emit(cs, op | (((va + results_base) >> 32) & 0xFF));
			r600_emit_reloc(ctx, &ctx->gfx, qbuf->buf, RADEON_USAGE_READ,
					RADEON_PRIO_QUERY);
			results_base += query->result_size;

			/* set CONTINUE bit for all packets except the first */
			op |= PREDICATION_CONTINUE;
		}
	}
}

static struct pipe_query *r600_create_query(struct pipe_context *ctx, unsigned query_type, unsigned index)
{
	struct r600_common_context *rctx = (struct r600_common_context *)ctx;

	if (query_type == PIPE_QUERY_TIMESTAMP_DISJOINT ||
	    query_type == PIPE_QUERY_GPU_FINISHED ||
	    query_type >= PIPE_QUERY_DRIVER_SPECIFIC)
		return r600_query_sw_create(ctx, query_type);

	return r600_query_hw_create(rctx, query_type, index);
}

static void r600_destroy_query(struct pipe_context *ctx, struct pipe_query *query)
{
	struct r600_common_context *rctx = (struct r600_common_context *)ctx;
	struct r600_query *rquery = (struct r600_query *)query;

	rquery->ops->destroy(rctx, rquery);
}

static boolean r600_begin_query(struct pipe_context *ctx,
                                struct pipe_query *query)
{
	struct r600_common_context *rctx = (struct r600_common_context *)ctx;
	struct r600_query *rquery = (struct r600_query *)query;

	return rquery->ops->begin(rctx, rquery);
}

void r600_query_hw_reset_buffers(struct r600_common_context *rctx,
				 struct r600_query_hw *query)
{
	struct r600_query_buffer *prev = query->buffer.previous;

	/* Discard the old query buffers. */
	while (prev) {
		struct r600_query_buffer *qbuf = prev;
		prev = prev->previous;
		r600_resource_reference(&qbuf->buf, NULL);
		FREE(qbuf);
	}

	query->buffer.results_end = 0;
	query->buffer.previous = NULL;

	/* Obtain a new buffer if the current one can't be mapped without a stall. */
	if (r600_rings_is_buffer_referenced(rctx, query->buffer.buf->buf, RADEON_USAGE_READWRITE) ||
	    !rctx->ws->buffer_wait(query->buffer.buf->buf, 0, RADEON_USAGE_READWRITE)) {
		r600_resource_reference(&query->buffer.buf, NULL);
		query->buffer.buf = r600_new_query_buffer(rctx, query);
	} else {
		if (!query->ops->prepare_buffer(rctx, query, query->buffer.buf))
			r600_resource_reference(&query->buffer.buf, NULL);
	}
}

bool r600_query_hw_begin(struct r600_common_context *rctx,
			 struct r600_query *rquery)
{
	struct r600_query_hw *query = (struct r600_query_hw *)rquery;

	if (query->flags & R600_QUERY_HW_FLAG_NO_START) {
		assert(0);
		return false;
	}

	if (!(query->flags & R600_QUERY_HW_FLAG_BEGIN_RESUMES))
		r600_query_hw_reset_buffers(rctx, query);

	r600_query_hw_emit_start(rctx, query);
	if (!query->buffer.buf)
		return false;

	LIST_ADDTAIL(&query->list, &rctx->active_queries);
	return true;
}

static bool r600_end_query(struct pipe_context *ctx, struct pipe_query *query)
{
	struct r600_common_context *rctx = (struct r600_common_context *)ctx;
	struct r600_query *rquery = (struct r600_query *)query;

	return rquery->ops->end(rctx, rquery);
}

bool r600_query_hw_end(struct r600_common_context *rctx,
		       struct r600_query *rquery)
{
	struct r600_query_hw *query = (struct r600_query_hw *)rquery;

	if (query->flags & R600_QUERY_HW_FLAG_NO_START)
		r600_query_hw_reset_buffers(rctx, query);

	r600_query_hw_emit_stop(rctx, query);

	if (!(query->flags & R600_QUERY_HW_FLAG_NO_START))
		LIST_DELINIT(&query->list);

	if (!query->buffer.buf)
		return false;

	return true;
}

static void r600_get_hw_query_params(struct r600_common_context *rctx,
				     struct r600_query_hw *rquery, int index,
				     struct r600_hw_query_params *params)
{
	params->pair_stride = 0;
	params->pair_count = 1;

	switch (rquery->b.type) {
	case PIPE_QUERY_OCCLUSION_COUNTER:
	case PIPE_QUERY_OCCLUSION_PREDICATE:
		params->start_offset = 0;
		params->end_offset = 8;
		params->fence_offset = rctx->max_db * 16;
		params->pair_stride = 16;
		params->pair_count = rctx->max_db;
		break;
	case PIPE_QUERY_TIME_ELAPSED:
		params->start_offset = 0;
		params->end_offset = 8;
		params->fence_offset = 16;
		break;
	case PIPE_QUERY_TIMESTAMP:
		params->start_offset = 0;
		params->end_offset = 0;
		params->fence_offset = 8;
		break;
	case PIPE_QUERY_PRIMITIVES_EMITTED:
		params->start_offset = 8;
		params->end_offset = 24;
		params->fence_offset = params->end_offset + 4;
		break;
	case PIPE_QUERY_PRIMITIVES_GENERATED:
		params->start_offset = 0;
		params->end_offset = 16;
		params->fence_offset = params->end_offset + 4;
		break;
	case PIPE_QUERY_SO_STATISTICS:
		params->start_offset = 8 - index * 8;
		params->end_offset = 24 - index * 8;
		params->fence_offset = params->end_offset + 4;
		break;
	case PIPE_QUERY_PIPELINE_STATISTICS:
	{
		/* Offsets apply to EG+ */
		static const unsigned offsets[] = {56, 48, 24, 32, 40, 16, 8, 0, 64, 72, 80};
		params->start_offset = offsets[index];
		params->end_offset = 88 + offsets[index];
		params->fence_offset = 2 * 88;
		break;
	}
	default:
		unreachable("r600_get_hw_query_params unsupported");
	}
}

static unsigned r600_query_read_result(void *map, unsigned start_index, unsigned end_index,
				       bool test_status_bit)
{
	uint32_t *current_result = (uint32_t*)map;
	uint64_t start, end;

	start = (uint64_t)current_result[start_index] |
		(uint64_t)current_result[start_index+1] << 32;
	end = (uint64_t)current_result[end_index] |
	      (uint64_t)current_result[end_index+1] << 32;

	if (!test_status_bit ||
	    ((start & 0x8000000000000000UL) && (end & 0x8000000000000000UL))) {
		return end - start;
	}
	return 0;
}

static void r600_query_hw_add_result(struct r600_common_context *ctx,
				     struct r600_query_hw *query,
				     void *buffer,
				     union pipe_query_result *result)
{
	switch (query->b.type) {
	case PIPE_QUERY_OCCLUSION_COUNTER: {
		for (unsigned i = 0; i < ctx->max_db; ++i) {
			unsigned results_base = i * 16;
			result->u64 +=
				r600_query_read_result(buffer + results_base, 0, 2, true);
		}
		break;
	}
	case PIPE_QUERY_OCCLUSION_PREDICATE: {
		for (unsigned i = 0; i < ctx->max_db; ++i) {
			unsigned results_base = i * 16;
			result->b = result->b ||
				r600_query_read_result(buffer + results_base, 0, 2, true) != 0;
		}
		break;
	}
	case PIPE_QUERY_TIME_ELAPSED:
		result->u64 += r600_query_read_result(buffer, 0, 2, false);
		break;
	case PIPE_QUERY_TIMESTAMP:
		result->u64 = *(uint64_t*)buffer;
		break;
	case PIPE_QUERY_PRIMITIVES_EMITTED:
		/* SAMPLE_STREAMOUTSTATS stores this structure:
		 * {
		 *    u64 NumPrimitivesWritten;
		 *    u64 PrimitiveStorageNeeded;
		 * }
		 * We only need NumPrimitivesWritten here. */
		result->u64 += r600_query_read_result(buffer, 2, 6, true);
		break;
	case PIPE_QUERY_PRIMITIVES_GENERATED:
		/* Here we read PrimitiveStorageNeeded. */
		result->u64 += r600_query_read_result(buffer, 0, 4, true);
		break;
	case PIPE_QUERY_SO_STATISTICS:
		result->so_statistics.num_primitives_written +=
			r600_query_read_result(buffer, 2, 6, true);
		result->so_statistics.primitives_storage_needed +=
			r600_query_read_result(buffer, 0, 4, true);
		break;
	case PIPE_QUERY_SO_OVERFLOW_PREDICATE:
		result->b = result->b ||
			r600_query_read_result(buffer, 2, 6, true) !=
			r600_query_read_result(buffer, 0, 4, true);
		break;
	case PIPE_QUERY_PIPELINE_STATISTICS:
		if (ctx->chip_class >= EVERGREEN) {
			result->pipeline_statistics.ps_invocations +=
				r600_query_read_result(buffer, 0, 22, false);
			result->pipeline_statistics.c_primitives +=
				r600_query_read_result(buffer, 2, 24, false);
			result->pipeline_statistics.c_invocations +=
				r600_query_read_result(buffer, 4, 26, false);
			result->pipeline_statistics.vs_invocations +=
				r600_query_read_result(buffer, 6, 28, false);
			result->pipeline_statistics.gs_invocations +=
				r600_query_read_result(buffer, 8, 30, false);
			result->pipeline_statistics.gs_primitives +=
				r600_query_read_result(buffer, 10, 32, false);
			result->pipeline_statistics.ia_primitives +=
				r600_query_read_result(buffer, 12, 34, false);
			result->pipeline_statistics.ia_vertices +=
				r600_query_read_result(buffer, 14, 36, false);
			result->pipeline_statistics.hs_invocations +=
				r600_query_read_result(buffer, 16, 38, false);
			result->pipeline_statistics.ds_invocations +=
				r600_query_read_result(buffer, 18, 40, false);
			result->pipeline_statistics.cs_invocations +=
				r600_query_read_result(buffer, 20, 42, false);
		} else {
			result->pipeline_statistics.ps_invocations +=
				r600_query_read_result(buffer, 0, 16, false);
			result->pipeline_statistics.c_primitives +=
				r600_query_read_result(buffer, 2, 18, false);
			result->pipeline_statistics.c_invocations +=
				r600_query_read_result(buffer, 4, 20, false);
			result->pipeline_statistics.vs_invocations +=
				r600_query_read_result(buffer, 6, 22, false);
			result->pipeline_statistics.gs_invocations +=
				r600_query_read_result(buffer, 8, 24, false);
			result->pipeline_statistics.gs_primitives +=
				r600_query_read_result(buffer, 10, 26, false);
			result->pipeline_statistics.ia_primitives +=
				r600_query_read_result(buffer, 12, 28, false);
			result->pipeline_statistics.ia_vertices +=
				r600_query_read_result(buffer, 14, 30, false);
		}
#if 0 /* for testing */
		printf("Pipeline stats: IA verts=%llu, IA prims=%llu, VS=%llu, HS=%llu, "
		       "DS=%llu, GS=%llu, GS prims=%llu, Clipper=%llu, "
		       "Clipper prims=%llu, PS=%llu, CS=%llu\n",
		       result->pipeline_statistics.ia_vertices,
		       result->pipeline_statistics.ia_primitives,
		       result->pipeline_statistics.vs_invocations,
		       result->pipeline_statistics.hs_invocations,
		       result->pipeline_statistics.ds_invocations,
		       result->pipeline_statistics.gs_invocations,
		       result->pipeline_statistics.gs_primitives,
		       result->pipeline_statistics.c_invocations,
		       result->pipeline_statistics.c_primitives,
		       result->pipeline_statistics.ps_invocations,
		       result->pipeline_statistics.cs_invocations);
#endif
		break;
	default:
		assert(0);
	}
}

static boolean r600_get_query_result(struct pipe_context *ctx,
				     struct pipe_query *query, boolean wait,
				     union pipe_query_result *result)
{
	struct r600_common_context *rctx = (struct r600_common_context *)ctx;
	struct r600_query *rquery = (struct r600_query *)query;

	return rquery->ops->get_result(rctx, rquery, wait, result);
}

static void r600_get_query_result_resource(struct pipe_context *ctx,
                                           struct pipe_query *query,
                                           boolean wait,
                                           enum pipe_query_value_type result_type,
                                           int index,
                                           struct pipe_resource *resource,
                                           unsigned offset)
{
	struct r600_common_context *rctx = (struct r600_common_context *)ctx;
	struct r600_query *rquery = (struct r600_query *)query;

	rquery->ops->get_result_resource(rctx, rquery, wait, result_type, index,
	                                 resource, offset);
}

static void r600_query_hw_clear_result(struct r600_query_hw *query,
				       union pipe_query_result *result)
{
	util_query_clear_result(result, query->b.type);
}

bool r600_query_hw_get_result(struct r600_common_context *rctx,
			      struct r600_query *rquery,
			      bool wait, union pipe_query_result *result)
{
	struct r600_query_hw *query = (struct r600_query_hw *)rquery;
	struct r600_query_buffer *qbuf;

	query->ops->clear_result(query, result);

	for (qbuf = &query->buffer; qbuf; qbuf = qbuf->previous) {
		unsigned results_base = 0;
		void *map;

		map = r600_buffer_map_sync_with_rings(rctx, qbuf->buf,
						      PIPE_TRANSFER_READ |
						      (wait ? 0 : PIPE_TRANSFER_DONTBLOCK));
		if (!map)
			return false;

		while (results_base != qbuf->results_end) {
			query->ops->add_result(rctx, query, map + results_base,
					       result);
			results_base += query->result_size;
		}
	}

	/* Convert the time to expected units. */
	if (rquery->type == PIPE_QUERY_TIME_ELAPSED ||
	    rquery->type == PIPE_QUERY_TIMESTAMP) {
		result->u64 = (1000000 * result->u64) / rctx->screen->info.clock_crystal_freq;
	}
	return true;
}

/* Create the compute shader that is used to collect the results.
 *
 * One compute grid with a single thread is launched for every query result
 * buffer. The thread (optionally) reads a previous summary buffer, then
 * accumulates data from the query result buffer, and writes the result either
 * to a summary buffer to be consumed by the next grid invocation or to the
 * user-supplied buffer.
 *
 * Data layout:
 *
 * CONST
 *  0.x = end_offset
 *  0.y = result_stride
 *  0.z = result_count
 *  0.w = bit field:
 *          1: read previously accumulated values
 *          2: write accumulated values for chaining
 *          4: write result available
 *          8: convert result to boolean (0/1)
 *         16: only read one dword and use that as result
 *         32: apply timestamp conversion
 *         64: store full 64 bits result
 *        128: store signed 32 bits result
 *  1.x = fence_offset
 *  1.y = pair_stride
 *  1.z = pair_count
 *
 * BUFFER[0] = query result buffer
 * BUFFER[1] = previous summary buffer
 * BUFFER[2] = next summary buffer or user-supplied buffer
 */
static void r600_create_query_result_shader(struct r600_common_context *rctx)
{
	/* TEMP[0].xy = accumulated result so far
	 * TEMP[0].z = result not available
	 *
	 * TEMP[1].x = current result index
	 * TEMP[1].y = current pair index
	 */
	static const char text_tmpl[] =
		"COMP\n"
		"PROPERTY CS_FIXED_BLOCK_WIDTH 1\n"
		"PROPERTY CS_FIXED_BLOCK_HEIGHT 1\n"
		"PROPERTY CS_FIXED_BLOCK_DEPTH 1\n"
		"DCL BUFFER[0]\n"
		"DCL BUFFER[1]\n"
		"DCL BUFFER[2]\n"
		"DCL CONST[0..1]\n"
		"DCL TEMP[0..5]\n"
		"IMM[0] UINT32 {0, 31, 2147483647, 4294967295}\n"
		"IMM[1] UINT32 {1, 2, 4, 8}\n"
		"IMM[2] UINT32 {16, 32, 64, 128}\n"
		"IMM[3] UINT32 {1000000, 0, %u, 0}\n" /* for timestamp conversion */

		"AND TEMP[5], CONST[0].wwww, IMM[2].xxxx\n"
		"UIF TEMP[5]\n"
			/* Check result availability. */
			"LOAD TEMP[1].x, BUFFER[0], CONST[1].xxxx\n"
			"ISHR TEMP[0].z, TEMP[1].xxxx, IMM[0].yyyy\n"
			"MOV TEMP[1], TEMP[0].zzzz\n"
			"NOT TEMP[0].z, TEMP[0].zzzz\n"

			/* Load result if available. */
			"UIF TEMP[1]\n"
				"LOAD TEMP[0].xy, BUFFER[0], IMM[0].xxxx\n"
			"ENDIF\n"
		"ELSE\n"
			/* Load previously accumulated result if requested. */
			"MOV TEMP[0], IMM[0].xxxx\n"
			"AND TEMP[4], CONST[0].wwww, IMM[1].xxxx\n"
			"UIF TEMP[4]\n"
				"LOAD TEMP[0].xyz, BUFFER[1], IMM[0].xxxx\n"
			"ENDIF\n"

			"MOV TEMP[1].x, IMM[0].xxxx\n"
			"BGNLOOP\n"
				/* Break if accumulated result so far is not available. */
				"UIF TEMP[0].zzzz\n"
					"BRK\n"
				"ENDIF\n"

				/* Break if result_index >= result_count. */
				"USGE TEMP[5], TEMP[1].xxxx, CONST[0].zzzz\n"
				"UIF TEMP[5]\n"
					"BRK\n"
				"ENDIF\n"

				/* Load fence and check result availability */
				"UMAD TEMP[5].x, TEMP[1].xxxx, CONST[0].yyyy, CONST[1].xxxx\n"
				"LOAD TEMP[5].x, BUFFER[0], TEMP[5].xxxx\n"
				"ISHR TEMP[0].z, TEMP[5].xxxx, IMM[0].yyyy\n"
				"NOT TEMP[0].z, TEMP[0].zzzz\n"
				"UIF TEMP[0].zzzz\n"
					"BRK\n"
				"ENDIF\n"

				"MOV TEMP[1].y, IMM[0].xxxx\n"
				"BGNLOOP\n"
					/* Load start and end. */
					"UMUL TEMP[5].x, TEMP[1].xxxx, CONST[0].yyyy\n"
					"UMAD TEMP[5].x, TEMP[1].yyyy, CONST[1].yyyy, TEMP[5].xxxx\n"
					"LOAD TEMP[2].xy, BUFFER[0], TEMP[5].xxxx\n"

					"UADD TEMP[5].x, TEMP[5].xxxx, CONST[0].xxxx\n"
					"LOAD TEMP[3].xy, BUFFER[0], TEMP[5].xxxx\n"

					"U64ADD TEMP[3].xy, TEMP[3], -TEMP[2]\n"
					"U64ADD TEMP[0].xy, TEMP[0], TEMP[3]\n"

					/* Increment pair index */
					"UADD TEMP[1].y, TEMP[1].yyyy, IMM[1].xxxx\n"
					"USGE TEMP[5], TEMP[1].yyyy, CONST[1].zzzz\n"
					"UIF TEMP[5]\n"
						"BRK\n"
					"ENDIF\n"
				"ENDLOOP\n"

				/* Increment result index */
				"UADD TEMP[1].x, TEMP[1].xxxx, IMM[1].xxxx\n"
			"ENDLOOP\n"
		"ENDIF\n"

		"AND TEMP[4], CONST[0].wwww, IMM[1].yyyy\n"
		"UIF TEMP[4]\n"
			/* Store accumulated data for chaining. */
			"STORE BUFFER[2].xyz, IMM[0].xxxx, TEMP[0]\n"
		"ELSE\n"
			"AND TEMP[4], CONST[0].wwww, IMM[1].zzzz\n"
			"UIF TEMP[4]\n"
				/* Store result availability. */
				"NOT TEMP[0].z, TEMP[0]\n"
				"AND TEMP[0].z, TEMP[0].zzzz, IMM[1].xxxx\n"
				"STORE BUFFER[2].x, IMM[0].xxxx, TEMP[0].zzzz\n"

				"AND TEMP[4], CONST[0].wwww, IMM[2].zzzz\n"
				"UIF TEMP[4]\n"
					"STORE BUFFER[2].y, IMM[0].xxxx, IMM[0].xxxx\n"
				"ENDIF\n"
			"ELSE\n"
				/* Store result if it is available. */
				"NOT TEMP[4], TEMP[0].zzzz\n"
				"UIF TEMP[4]\n"
					/* Apply timestamp conversion */
					"AND TEMP[4], CONST[0].wwww, IMM[2].yyyy\n"
					"UIF TEMP[4]\n"
						"U64MUL TEMP[0].xy, TEMP[0], IMM[3].xyxy\n"
						"U64DIV TEMP[0].xy, TEMP[0], IMM[3].zwzw\n"
					"ENDIF\n"

					/* Convert to boolean */
					"AND TEMP[4], CONST[0].wwww, IMM[1].wwww\n"
					"UIF TEMP[4]\n"
						"U64SNE TEMP[0].x, TEMP[0].xyxy, IMM[0].xxxx\n"
						"AND TEMP[0].x, TEMP[0].xxxx, IMM[1].xxxx\n"
						"MOV TEMP[0].y, IMM[0].xxxx\n"
					"ENDIF\n"

					"AND TEMP[4], CONST[0].wwww, IMM[2].zzzz\n"
					"UIF TEMP[4]\n"
						"STORE BUFFER[2].xy, IMM[0].xxxx, TEMP[0].xyxy\n"
					"ELSE\n"
						/* Clamping */
						"UIF TEMP[0].yyyy\n"
							"MOV TEMP[0].x, IMM[0].wwww\n"
						"ENDIF\n"

						"AND TEMP[4], CONST[0].wwww, IMM[2].wwww\n"
						"UIF TEMP[4]\n"
							"UMIN TEMP[0].x, TEMP[0].xxxx, IMM[0].zzzz\n"
						"ENDIF\n"

						"STORE BUFFER[2].x, IMM[0].xxxx, TEMP[0].xxxx\n"
					"ENDIF\n"
				"ENDIF\n"
			"ENDIF\n"
		"ENDIF\n"

		"END\n";

	char text[sizeof(text_tmpl) + 32];
	struct tgsi_token tokens[1024];
	struct pipe_compute_state state = {};

	/* Hard code the frequency into the shader so that the backend can
	 * use the full range of optimizations for divide-by-constant.
	 */
	snprintf(text, sizeof(text), text_tmpl,
		 rctx->screen->info.clock_crystal_freq);

	if (!tgsi_text_translate(text, tokens, ARRAY_SIZE(tokens))) {
		assert(false);
		return;
	}

	state.ir_type = PIPE_SHADER_IR_TGSI;
	state.prog = tokens;

	rctx->query_result_shader = rctx->b.create_compute_state(&rctx->b, &state);
}

static void r600_restore_qbo_state(struct r600_common_context *rctx,
				   struct r600_qbo_state *st)
{
	rctx->b.bind_compute_state(&rctx->b, st->saved_compute);

	rctx->b.set_constant_buffer(&rctx->b, PIPE_SHADER_COMPUTE, 0, &st->saved_const0);
	pipe_resource_reference(&st->saved_const0.buffer, NULL);

	rctx->b.set_shader_buffers(&rctx->b, PIPE_SHADER_COMPUTE, 0, 3, st->saved_ssbo);
	for (unsigned i = 0; i < 3; ++i)
		pipe_resource_reference(&st->saved_ssbo[i].buffer, NULL);
}

static void r600_query_hw_get_result_resource(struct r600_common_context *rctx,
                                              struct r600_query *rquery,
                                              bool wait,
                                              enum pipe_query_value_type result_type,
                                              int index,
                                              struct pipe_resource *resource,
                                              unsigned offset)
{
	struct r600_query_hw *query = (struct r600_query_hw *)rquery;
	struct r600_query_buffer *qbuf;
	struct r600_query_buffer *qbuf_prev;
	struct pipe_resource *tmp_buffer = NULL;
	unsigned tmp_buffer_offset = 0;
	struct r600_qbo_state saved_state = {};
	struct pipe_grid_info grid = {};
	struct pipe_constant_buffer constant_buffer = {};
	struct pipe_shader_buffer ssbo[3];
	struct r600_hw_query_params params;
	struct {
		uint32_t end_offset;
		uint32_t result_stride;
		uint32_t result_count;
		uint32_t config;
		uint32_t fence_offset;
		uint32_t pair_stride;
		uint32_t pair_count;
	} consts;

	if (!rctx->query_result_shader) {
		r600_create_query_result_shader(rctx);
		if (!rctx->query_result_shader)
			return;
	}

	if (query->buffer.previous) {
		u_suballocator_alloc(rctx->allocator_zeroed_memory, 16, 16,
				     &tmp_buffer_offset, &tmp_buffer);
		if (!tmp_buffer)
			return;
	}

	rctx->save_qbo_state(&rctx->b, &saved_state);

	r600_get_hw_query_params(rctx, query, index >= 0 ? index : 0, &params);
	consts.end_offset = params.end_offset - params.start_offset;
	consts.fence_offset = params.fence_offset - params.start_offset;
	consts.result_stride = query->result_size;
	consts.pair_stride = params.pair_stride;
	consts.pair_count = params.pair_count;

	constant_buffer.buffer_size = sizeof(consts);
	constant_buffer.user_buffer = &consts;

	ssbo[1].buffer = tmp_buffer;
	ssbo[1].buffer_offset = tmp_buffer_offset;
	ssbo[1].buffer_size = 16;

	ssbo[2] = ssbo[1];

	rctx->b.bind_compute_state(&rctx->b, rctx->query_result_shader);

	grid.block[0] = 1;
	grid.block[1] = 1;
	grid.block[2] = 1;
	grid.grid[0] = 1;
	grid.grid[1] = 1;
	grid.grid[2] = 1;

	consts.config = 0;
	if (index < 0)
		consts.config |= 4;
	if (query->b.type == PIPE_QUERY_OCCLUSION_PREDICATE ||
	    query->b.type == PIPE_QUERY_SO_OVERFLOW_PREDICATE)
		consts.config |= 8;
	else if (query->b.type == PIPE_QUERY_TIMESTAMP ||
		 query->b.type == PIPE_QUERY_TIME_ELAPSED)
		consts.config |= 32;

	switch (result_type) {
	case PIPE_QUERY_TYPE_U64:
	case PIPE_QUERY_TYPE_I64:
		consts.config |= 64;
		break;
	case PIPE_QUERY_TYPE_I32:
		consts.config |= 128;
		break;
	case PIPE_QUERY_TYPE_U32:
		break;
	}

	rctx->flags |= rctx->screen->barrier_flags.cp_to_L2;

	for (qbuf = &query->buffer; qbuf; qbuf = qbuf_prev) {
		if (query->b.type != PIPE_QUERY_TIMESTAMP) {
			qbuf_prev = qbuf->previous;
			consts.result_count = qbuf->results_end / query->result_size;
			consts.config &= ~3;
			if (qbuf != &query->buffer)
				consts.config |= 1;
			if (qbuf->previous)
				consts.config |= 2;
		} else {
			/* Only read the last timestamp. */
			qbuf_prev = NULL;
			consts.result_count = 0;
			consts.config |= 16;
			params.start_offset += qbuf->results_end - query->result_size;
		}

		rctx->b.set_constant_buffer(&rctx->b, PIPE_SHADER_COMPUTE, 0, &constant_buffer);

		ssbo[0].buffer = &qbuf->buf->b.b;
		ssbo[0].buffer_offset = params.start_offset;
		ssbo[0].buffer_size = qbuf->results_end - params.start_offset;

		if (!qbuf->previous) {
			ssbo[2].buffer = resource;
			ssbo[2].buffer_offset = offset;
			ssbo[2].buffer_size = 8;

			((struct r600_resource *)resource)->TC_L2_dirty = true;
		}

		rctx->b.set_shader_buffers(&rctx->b, PIPE_SHADER_COMPUTE, 0, 3, ssbo);

		if (wait && qbuf == &query->buffer) {
			uint64_t va;

			/* Wait for result availability. Wait only for readiness
			 * of the last entry, since the fence writes should be
			 * serialized in the CP.
			 */
			va = qbuf->buf->gpu_address + qbuf->results_end - query->result_size;
			va += params.fence_offset;

			r600_gfx_wait_fence(rctx, va, 0x80000000, 0x80000000);
		}

		rctx->b.launch_grid(&rctx->b, &grid);
		rctx->flags |= rctx->screen->barrier_flags.compute_to_L2;
	}

	r600_restore_qbo_state(rctx, &saved_state);
	pipe_resource_reference(&tmp_buffer, NULL);
}

static void r600_render_condition(struct pipe_context *ctx,
				  struct pipe_query *query,
				  boolean condition,
				  uint mode)
{
	struct r600_common_context *rctx = (struct r600_common_context *)ctx;
	struct r600_query_hw *rquery = (struct r600_query_hw *)query;
	struct r600_query_buffer *qbuf;
	struct r600_atom *atom = &rctx->render_cond_atom;

	rctx->render_cond = query;
	rctx->render_cond_invert = condition;
	rctx->render_cond_mode = mode;

	/* Compute the size of SET_PREDICATION packets. */
	atom->num_dw = 0;
	if (query) {
		for (qbuf = &rquery->buffer; qbuf; qbuf = qbuf->previous)
			atom->num_dw += (qbuf->results_end / rquery->result_size) * 5;
	}

	rctx->set_atom_dirty(rctx, atom, query != NULL);
}

void r600_suspend_queries(struct r600_common_context *ctx)
{
	struct r600_query_hw *query;

	LIST_FOR_EACH_ENTRY(query, &ctx->active_queries, list) {
		r600_query_hw_emit_stop(ctx, query);
	}
	assert(ctx->num_cs_dw_queries_suspend == 0);
}

static unsigned r600_queries_num_cs_dw_for_resuming(struct r600_common_context *ctx,
						    struct list_head *query_list)
{
	struct r600_query_hw *query;
	unsigned num_dw = 0;

	LIST_FOR_EACH_ENTRY(query, query_list, list) {
		/* begin + end */
		num_dw += query->num_cs_dw_begin + query->num_cs_dw_end;

		/* Workaround for the fact that
		 * num_cs_dw_nontimer_queries_suspend is incremented for every
		 * resumed query, which raises the bar in need_cs_space for
		 * queries about to be resumed.
		 */
		num_dw += query->num_cs_dw_end;
	}
	/* primitives generated query */
	num_dw += ctx->streamout.enable_atom.num_dw;
	/* guess for ZPASS enable or PERFECT_ZPASS_COUNT enable updates */
	num_dw += 13;

	return num_dw;
}

void r600_resume_queries(struct r600_common_context *ctx)
{
	struct r600_query_hw *query;
	unsigned num_cs_dw = r600_queries_num_cs_dw_for_resuming(ctx, &ctx->active_queries);

	assert(ctx->num_cs_dw_queries_suspend == 0);

	/* Check CS space here. Resuming must not be interrupted by flushes. */
	ctx->need_gfx_cs_space(&ctx->b, num_cs_dw, true);

	LIST_FOR_EACH_ENTRY(query, &ctx->active_queries, list) {
		r600_query_hw_emit_start(ctx, query);
	}
}

/* Get backends mask */
void r600_query_init_backend_mask(struct r600_common_context *ctx)
{
	struct radeon_winsys_cs *cs = ctx->gfx.cs;
	struct r600_resource *buffer;
	uint32_t *results;
	unsigned num_backends = ctx->screen->info.num_render_backends;
	unsigned i, mask = 0;

	/* if backend_map query is supported by the kernel */
	if (ctx->screen->info.r600_gb_backend_map_valid) {
		unsigned num_tile_pipes = ctx->screen->info.num_tile_pipes;
		unsigned backend_map = ctx->screen->info.r600_gb_backend_map;
		unsigned item_width, item_mask;

		if (ctx->chip_class >= EVERGREEN) {
			item_width = 4;
			item_mask = 0x7;
		} else {
			item_width = 2;
			item_mask = 0x3;
		}

		while (num_tile_pipes--) {
			i = backend_map & item_mask;
			mask |= (1<<i);
			backend_map >>= item_width;
		}
		if (mask != 0) {
			ctx->backend_mask = mask;
			return;
		}
	}

	/* otherwise backup path for older kernels */

	/* create buffer for event data */
	buffer = (struct r600_resource*)
		pipe_buffer_create(ctx->b.screen, 0,
				   PIPE_USAGE_STAGING, ctx->max_db*16);
	if (!buffer)
		goto err;

	/* initialize buffer with zeroes */
	results = r600_buffer_map_sync_with_rings(ctx, buffer, PIPE_TRANSFER_WRITE);
	if (results) {
		memset(results, 0, ctx->max_db * 4 * 4);

		/* emit EVENT_WRITE for ZPASS_DONE */
		radeon_emit(cs, PKT3(PKT3_EVENT_WRITE, 2, 0));
		radeon_emit(cs, EVENT_TYPE(EVENT_TYPE_ZPASS_DONE) | EVENT_INDEX(1));
		radeon_emit(cs, buffer->gpu_address);
		radeon_emit(cs, buffer->gpu_address >> 32);

		r600_emit_reloc(ctx, &ctx->gfx, buffer,
                                RADEON_USAGE_WRITE, RADEON_PRIO_QUERY);

		/* analyze results */
		results = r600_buffer_map_sync_with_rings(ctx, buffer, PIPE_TRANSFER_READ);
		if (results) {
			for(i = 0; i < ctx->max_db; i++) {
				/* at least highest bit will be set if backend is used */
				if (results[i*4 + 1])
					mask |= (1<<i);
			}
		}
	}

	r600_resource_reference(&buffer, NULL);

	if (mask != 0) {
		ctx->backend_mask = mask;
		return;
	}

err:
	/* fallback to old method - set num_backends lower bits to 1 */
	ctx->backend_mask = (~((uint32_t)0))>>(32-num_backends);
	return;
}

#define XFULL(name_, query_type_, type_, result_type_, group_id_) \
	{ \
		.name = name_, \
		.query_type = R600_QUERY_##query_type_, \
		.type = PIPE_DRIVER_QUERY_TYPE_##type_, \
		.result_type = PIPE_DRIVER_QUERY_RESULT_TYPE_##result_type_, \
		.group_id = group_id_ \
	}

#define X(name_, query_type_, type_, result_type_) \
	XFULL(name_, query_type_, type_, result_type_, ~(unsigned)0)

#define XG(group_, name_, query_type_, type_, result_type_) \
	XFULL(name_, query_type_, type_, result_type_, R600_QUERY_GROUP_##group_)

static struct pipe_driver_query_info r600_driver_query_list[] = {
	X("num-compilations",		NUM_COMPILATIONS,	UINT64, CUMULATIVE),
	X("num-shaders-created",	NUM_SHADERS_CREATED,	UINT64, CUMULATIVE),
	X("num-shader-cache-hits",	NUM_SHADER_CACHE_HITS,	UINT64, CUMULATIVE),
	X("draw-calls",			DRAW_CALLS,		UINT64, AVERAGE),
	X("spill-draw-calls",		SPILL_DRAW_CALLS,	UINT64, AVERAGE),
	X("compute-calls",		COMPUTE_CALLS,		UINT64, AVERAGE),
	X("spill-compute-calls",	SPILL_COMPUTE_CALLS,	UINT64, AVERAGE),
	X("dma-calls",			DMA_CALLS,		UINT64, AVERAGE),
	X("cp-dma-calls",		CP_DMA_CALLS,		UINT64, AVERAGE),
	X("num-vs-flushes",		NUM_VS_FLUSHES,		UINT64, AVERAGE),
	X("num-ps-flushes",		NUM_PS_FLUSHES,		UINT64, AVERAGE),
	X("num-cs-flushes",		NUM_CS_FLUSHES,		UINT64, AVERAGE),
	X("num-fb-cache-flushes",	NUM_FB_CACHE_FLUSHES,	UINT64, AVERAGE),
	X("num-L2-invalidates",		NUM_L2_INVALIDATES,	UINT64, AVERAGE),
	X("num-L2-writebacks",		NUM_L2_WRITEBACKS,	UINT64, AVERAGE),
	X("requested-VRAM",		REQUESTED_VRAM,		BYTES, AVERAGE),
	X("requested-GTT",		REQUESTED_GTT,		BYTES, AVERAGE),
	X("mapped-VRAM",		MAPPED_VRAM,		BYTES, AVERAGE),
	X("mapped-GTT",			MAPPED_GTT,		BYTES, AVERAGE),
	X("buffer-wait-time",		BUFFER_WAIT_TIME,	MICROSECONDS, CUMULATIVE),
	X("num-GFX-IBs",		NUM_GFX_IBS,		UINT64, AVERAGE),
	X("num-SDMA-IBs",		NUM_SDMA_IBS,		UINT64, AVERAGE),
	X("num-bytes-moved",		NUM_BYTES_MOVED,	BYTES, CUMULATIVE),
	X("num-evictions",		NUM_EVICTIONS,		UINT64, CUMULATIVE),
	X("VRAM-usage",			VRAM_USAGE,		BYTES, AVERAGE),
	X("GTT-usage",			GTT_USAGE,		BYTES, AVERAGE),
	X("back-buffer-ps-draw-ratio",	BACK_BUFFER_PS_DRAW_RATIO, UINT64, AVERAGE),

	/* GPIN queries are for the benefit of old versions of GPUPerfStudio,
	 * which use it as a fallback path to detect the GPU type.
	 *
	 * Note: The names of these queries are significant for GPUPerfStudio
	 * (and possibly their order as well). */
	XG(GPIN, "GPIN_000",		GPIN_ASIC_ID,		UINT, AVERAGE),
	XG(GPIN, "GPIN_001",		GPIN_NUM_SIMD,		UINT, AVERAGE),
	XG(GPIN, "GPIN_002",		GPIN_NUM_RB,		UINT, AVERAGE),
	XG(GPIN, "GPIN_003",		GPIN_NUM_SPI,		UINT, AVERAGE),
	XG(GPIN, "GPIN_004",		GPIN_NUM_SE,		UINT, AVERAGE),

	/* The following queries must be at the end of the list because their
	 * availability is adjusted dynamically based on the DRM version. */
	X("GPU-load",			GPU_LOAD,		UINT64, AVERAGE),
	X("GPU-shaders-busy",		GPU_SHADERS_BUSY,	UINT64, AVERAGE),
	X("temperature",		GPU_TEMPERATURE,	UINT64, AVERAGE),
	X("shader-clock",		CURRENT_GPU_SCLK,	HZ, AVERAGE),
	X("memory-clock",		CURRENT_GPU_MCLK,	HZ, AVERAGE),
};

#undef X
#undef XG
#undef XFULL

static unsigned r600_get_num_queries(struct r600_common_screen *rscreen)
{
	if (rscreen->info.drm_major == 2 && rscreen->info.drm_minor >= 42)
		return ARRAY_SIZE(r600_driver_query_list);
	else if (rscreen->info.drm_major == 3)
		return ARRAY_SIZE(r600_driver_query_list) - 3;
	else
		return ARRAY_SIZE(r600_driver_query_list) - 5;
}

static int r600_get_driver_query_info(struct pipe_screen *screen,
				      unsigned index,
				      struct pipe_driver_query_info *info)
{
	struct r600_common_screen *rscreen = (struct r600_common_screen*)screen;
	unsigned num_queries = r600_get_num_queries(rscreen);

	if (!info) {
		unsigned num_perfcounters =
			r600_get_perfcounter_info(rscreen, 0, NULL);

		return num_queries + num_perfcounters;
	}

	if (index >= num_queries)
		return r600_get_perfcounter_info(rscreen, index - num_queries, info);

	*info = r600_driver_query_list[index];

	switch (info->query_type) {
	case R600_QUERY_REQUESTED_VRAM:
	case R600_QUERY_VRAM_USAGE:
	case R600_QUERY_MAPPED_VRAM:
		info->max_value.u64 = rscreen->info.vram_size;
		break;
	case R600_QUERY_REQUESTED_GTT:
	case R600_QUERY_GTT_USAGE:
	case R600_QUERY_MAPPED_GTT:
		info->max_value.u64 = rscreen->info.gart_size;
		break;
	case R600_QUERY_GPU_TEMPERATURE:
		info->max_value.u64 = 125;
		break;
	}

	if (info->group_id != ~(unsigned)0 && rscreen->perfcounters)
		info->group_id += rscreen->perfcounters->num_groups;

	return 1;
}

/* Note: Unfortunately, GPUPerfStudio hardcodes the order of hardware
 * performance counter groups, so be careful when changing this and related
 * functions.
 */
static int r600_get_driver_query_group_info(struct pipe_screen *screen,
					    unsigned index,
					    struct pipe_driver_query_group_info *info)
{
	struct r600_common_screen *rscreen = (struct r600_common_screen *)screen;
	unsigned num_pc_groups = 0;

	if (rscreen->perfcounters)
		num_pc_groups = rscreen->perfcounters->num_groups;

	if (!info)
		return num_pc_groups + R600_NUM_SW_QUERY_GROUPS;

	if (index < num_pc_groups)
		return r600_get_perfcounter_group_info(rscreen, index, info);

	index -= num_pc_groups;
	if (index >= R600_NUM_SW_QUERY_GROUPS)
		return 0;

	info->name = "GPIN";
	info->max_active_queries = 5;
	info->num_queries = 5;
	return 1;
}

void r600_query_init(struct r600_common_context *rctx)
{
	rctx->b.create_query = r600_create_query;
	rctx->b.create_batch_query = r600_create_batch_query;
	rctx->b.destroy_query = r600_destroy_query;
	rctx->b.begin_query = r600_begin_query;
	rctx->b.end_query = r600_end_query;
	rctx->b.get_query_result = r600_get_query_result;
	rctx->b.get_query_result_resource = r600_get_query_result_resource;
	rctx->render_cond_atom.emit = r600_emit_query_predication;

	if (((struct r600_common_screen*)rctx->b.screen)->info.num_render_backends > 0)
	    rctx->b.render_condition = r600_render_condition;

	LIST_INITHEAD(&rctx->active_queries);
}

void r600_init_screen_query_functions(struct r600_common_screen *rscreen)
{
	rscreen->b.get_driver_query_info = r600_get_driver_query_info;
	rscreen->b.get_driver_query_group_info = r600_get_driver_query_group_info;
}