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_cs.h"
#include "util/u_memory.h"


struct r600_query_buffer {
	/* The buffer where query results are stored. */
	struct r600_resource			*buf;
	/* Offset of the next free result after current query data */
	unsigned				results_end;
	/* If a query buffer is full, a new buffer is created and the old one
	 * is put in here. When we calculate the result, we sum up the samples
	 * from all buffers. */
	struct r600_query_buffer		*previous;
};

struct r600_query {
	/* The query buffer and how many results are in it. */
	struct r600_query_buffer		buffer;
	/* The type of query */
	unsigned				type;
	/* Size of the result in memory for both begin_query and end_query,
	 * this can be one or two numbers, or it could even be a size of a structure. */
	unsigned				result_size;
	/* The number of dwords for begin_query or end_query. */
	unsigned				num_cs_dw;
	/* linked list of queries */
	struct list_head			list;
	/* for custom non-GPU queries */
	uint64_t begin_result;
	uint64_t end_result;
};


static bool r600_is_timer_query(unsigned type)
{
	return type == PIPE_QUERY_TIME_ELAPSED ||
	       type == PIPE_QUERY_TIMESTAMP;
}

static bool r600_query_needs_begin(unsigned type)
{
	return type != PIPE_QUERY_GPU_FINISHED &&
	       type != PIPE_QUERY_TIMESTAMP;
}

static struct r600_resource *r600_new_query_buffer(struct r600_common_context *ctx, unsigned type)
{
	unsigned j, i, num_results, buf_size = 4096;
	uint32_t *results;

	/* Non-GPU queries. */
	switch (type) {
	case PIPE_QUERY_TIMESTAMP_DISJOINT:
	case R600_QUERY_DRAW_CALLS:
	case R600_QUERY_REQUESTED_VRAM:
	case R600_QUERY_REQUESTED_GTT:
	case R600_QUERY_BUFFER_WAIT_TIME:
	case R600_QUERY_NUM_CS_FLUSHES:
	case R600_QUERY_NUM_BYTES_MOVED:
	case R600_QUERY_VRAM_USAGE:
	case R600_QUERY_GTT_USAGE:
		return NULL;
	}

	/* 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, PIPE_BIND_CUSTOM,
				   PIPE_USAGE_STAGING, buf_size);

	switch (type) {
	case PIPE_QUERY_OCCLUSION_COUNTER:
	case PIPE_QUERY_OCCLUSION_PREDICATE:
		results = r600_buffer_map_sync_with_rings(ctx, buf, PIPE_TRANSFER_WRITE);
		memset(results, 0, buf_size);

		/* Set top bits for unused backends. */
		num_results = buf_size / (16 * ctx->max_db);
		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;
		}
		ctx->ws->buffer_unmap(buf->cs_buf);
		break;
	case PIPE_QUERY_GPU_FINISHED:
	case PIPE_QUERY_TIME_ELAPSED:
	case PIPE_QUERY_TIMESTAMP:
		break;
	case PIPE_QUERY_PRIMITIVES_EMITTED:
	case PIPE_QUERY_PRIMITIVES_GENERATED:
	case PIPE_QUERY_SO_STATISTICS:
	case PIPE_QUERY_SO_OVERFLOW_PREDICATE:
	case PIPE_QUERY_PIPELINE_STATISTICS:
		results = r600_buffer_map_sync_with_rings(ctx, buf, PIPE_TRANSFER_WRITE);
		memset(results, 0, buf_size);
		ctx->ws->buffer_unmap(buf->cs_buf);
		break;
	default:
		assert(0);
	}
	return buf;
}

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 enable;

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

		enable = rctx->num_occlusion_queries != 0;

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

static void r600_emit_query_begin(struct r600_common_context *ctx, struct r600_query *query)
{
	struct radeon_winsys_cs *cs = ctx->rings.gfx.cs;
	uint64_t va;

	r600_update_occlusion_query_state(ctx, query->type, 1);
	r600_update_prims_generated_query_state(ctx, query->type, 1);
	ctx->need_gfx_cs_space(&ctx->b, query->num_cs_dw * 2, 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.buf = r600_new_query_buffer(ctx, query->type);
		query->buffer.results_end = 0;
		query->buffer.previous = qbuf;
	}

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

	switch (query->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 >> 32UL) & 0xFF);
		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_SAMPLE_STREAMOUTSTATS) | EVENT_INDEX(3));
		radeon_emit(cs, va);
		radeon_emit(cs, (va >> 32UL) & 0xFF);
		break;
	case PIPE_QUERY_TIME_ELAPSED:
		radeon_emit(cs, PKT3(PKT3_EVENT_WRITE_EOP, 4, 0));
		radeon_emit(cs, EVENT_TYPE(EVENT_TYPE_CACHE_FLUSH_AND_INV_TS_EVENT) | EVENT_INDEX(5));
		radeon_emit(cs, va);
		radeon_emit(cs, (3 << 29) | ((va >> 32UL) & 0xFF));
		radeon_emit(cs, 0);
		radeon_emit(cs, 0);
		break;
	case PIPE_QUERY_PIPELINE_STATISTICS:
		if (!ctx->num_pipelinestat_queries) {
			radeon_emit(cs, PKT3(PKT3_EVENT_WRITE, 0, 0));
			radeon_emit(cs, EVENT_TYPE(EVENT_TYPE_PIPELINESTAT_START) | EVENT_INDEX(0));
		}
		ctx->num_pipelinestat_queries++;
		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 >> 32UL) & 0xFF);
		break;
	default:
		assert(0);
	}
	r600_emit_reloc(ctx, &ctx->rings.gfx, query->buffer.buf, RADEON_USAGE_WRITE,
			RADEON_PRIO_MIN);

	if (!r600_is_timer_query(query->type)) {
		ctx->num_cs_dw_nontimer_queries_suspend += query->num_cs_dw;
	}
}

static void r600_emit_query_end(struct r600_common_context *ctx, struct r600_query *query)
{
	struct radeon_winsys_cs *cs = ctx->rings.gfx.cs;
	uint64_t va;

	/* The queries which need begin already called this in begin_query. */
	if (!r600_query_needs_begin(query->type)) {
		ctx->need_gfx_cs_space(&ctx->b, query->num_cs_dw, FALSE);
	}

	va = query->buffer.buf->gpu_address;

	/* emit end query */
	switch (query->type) {
	case PIPE_QUERY_OCCLUSION_COUNTER:
	case PIPE_QUERY_OCCLUSION_PREDICATE:
		va += query->buffer.results_end + 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 >> 32UL) & 0xFF);
		break;
	case PIPE_QUERY_PRIMITIVES_EMITTED:
	case PIPE_QUERY_PRIMITIVES_GENERATED:
	case PIPE_QUERY_SO_STATISTICS:
	case PIPE_QUERY_SO_OVERFLOW_PREDICATE:
		va += query->buffer.results_end + query->result_size/2;
		radeon_emit(cs, PKT3(PKT3_EVENT_WRITE, 2, 0));
		radeon_emit(cs, EVENT_TYPE(EVENT_TYPE_SAMPLE_STREAMOUTSTATS) | EVENT_INDEX(3));
		radeon_emit(cs, va);
		radeon_emit(cs, (va >> 32UL) & 0xFF);
		break;
	case PIPE_QUERY_TIME_ELAPSED:
		va += query->buffer.results_end + query->result_size/2;
		/* fall through */
	case PIPE_QUERY_TIMESTAMP:
		radeon_emit(cs, PKT3(PKT3_EVENT_WRITE_EOP, 4, 0));
		radeon_emit(cs, EVENT_TYPE(EVENT_TYPE_CACHE_FLUSH_AND_INV_TS_EVENT) | EVENT_INDEX(5));
		radeon_emit(cs, va);
		radeon_emit(cs, (3 << 29) | ((va >> 32UL) & 0xFF));
		radeon_emit(cs, 0);
		radeon_emit(cs, 0);
		break;
	case PIPE_QUERY_PIPELINE_STATISTICS:
		assert(ctx->num_pipelinestat_queries > 0);
		ctx->num_pipelinestat_queries--;
		if (!ctx->num_pipelinestat_queries) {
			radeon_emit(cs, PKT3(PKT3_EVENT_WRITE, 0, 0));
			radeon_emit(cs, EVENT_TYPE(EVENT_TYPE_PIPELINESTAT_STOP) | EVENT_INDEX(0));
		}
		va += query->buffer.results_end + query->result_size/2;
		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 >> 32UL) & 0xFF);
		break;
        case PIPE_QUERY_GPU_FINISHED:
		break;
	default:
		assert(0);
	}
	r600_emit_reloc(ctx, &ctx->rings.gfx, query->buffer.buf, RADEON_USAGE_WRITE,
			RADEON_PRIO_MIN);

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

	if (r600_query_needs_begin(query->type)) {
		if (!r600_is_timer_query(query->type)) {
			ctx->num_cs_dw_nontimer_queries_suspend -= query->num_cs_dw;
		}
	}

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

static void r600_emit_query_predication(struct r600_common_context *ctx, struct r600_query *query,
					int operation, bool flag_wait)
{
	struct radeon_winsys_cs *cs = ctx->rings.gfx.cs;

	if (operation == PREDICATION_OP_CLEAR) {
		ctx->need_gfx_cs_space(&ctx->b, 3, FALSE);

		radeon_emit(cs, PKT3(PKT3_SET_PREDICATION, 1, 0));
		radeon_emit(cs, 0);
		radeon_emit(cs, PRED_OP(PREDICATION_OP_CLEAR));
	} else {
		struct r600_query_buffer *qbuf;
		unsigned count;
		uint32_t op;

		/* Find how many results there are. */
		count = 0;
		for (qbuf = &query->buffer; qbuf; qbuf = qbuf->previous) {
			count += qbuf->results_end / query->result_size;
		}

		ctx->need_gfx_cs_space(&ctx->b, 5 * count, TRUE);

		op = PRED_OP(operation) | PREDICATION_DRAW_VISIBLE |
				(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) & 0xFFFFFFFFUL);
				radeon_emit(cs, op | (((va + results_base) >> 32UL) & 0xFF));
				r600_emit_reloc(ctx, &ctx->rings.gfx, qbuf->buf, RADEON_USAGE_READ,
						RADEON_PRIO_MIN);
				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;
	struct r600_query *query;
	bool skip_allocation = false;

	query = CALLOC_STRUCT(r600_query);
	if (query == NULL)
		return NULL;

	query->type = query_type;

	switch (query_type) {
	case PIPE_QUERY_OCCLUSION_COUNTER:
	case PIPE_QUERY_OCCLUSION_PREDICATE:
		query->result_size = 16 * rctx->max_db;
		query->num_cs_dw = 6;
		break;
	case PIPE_QUERY_GPU_FINISHED:
		query->num_cs_dw = 2;
		break;
	case PIPE_QUERY_TIME_ELAPSED:
		query->result_size = 16;
		query->num_cs_dw = 8;
		break;
	case PIPE_QUERY_TIMESTAMP:
		query->result_size = 8;
		query->num_cs_dw = 8;
		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 = 6;
		break;
	case PIPE_QUERY_PIPELINE_STATISTICS:
		/* 11 values on EG, 8 on R600. */
		query->result_size = (rctx->chip_class >= EVERGREEN ? 11 : 8) * 16;
		query->num_cs_dw = 8;
		break;
	/* Non-GPU queries. */
	case PIPE_QUERY_TIMESTAMP_DISJOINT:
	case R600_QUERY_DRAW_CALLS:
	case R600_QUERY_REQUESTED_VRAM:
	case R600_QUERY_REQUESTED_GTT:
	case R600_QUERY_BUFFER_WAIT_TIME:
	case R600_QUERY_NUM_CS_FLUSHES:
	case R600_QUERY_NUM_BYTES_MOVED:
	case R600_QUERY_VRAM_USAGE:
	case R600_QUERY_GTT_USAGE:
		skip_allocation = true;
		break;
	default:
		assert(0);
		FREE(query);
		return NULL;
	}

	if (!skip_allocation) {
		query->buffer.buf = r600_new_query_buffer(rctx, query_type);
		if (!query->buffer.buf) {
			FREE(query);
			return NULL;
		}
	}
	return (struct pipe_query*)query;
}

static void r600_destroy_query(struct pipe_context *ctx, struct pipe_query *query)
{
	struct r600_query *rquery = (struct r600_query*)query;
	struct r600_query_buffer *prev = rquery->buffer.previous;

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

	pipe_resource_reference((struct pipe_resource**)&rquery->buffer.buf, NULL);
	FREE(query);
}

static void 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;
	struct r600_query_buffer *prev = rquery->buffer.previous;

	if (!r600_query_needs_begin(rquery->type)) {
		assert(0);
		return;
	}

	/* Non-GPU queries. */
	switch (rquery->type) {
	case PIPE_QUERY_TIMESTAMP_DISJOINT:
		return;
	case R600_QUERY_DRAW_CALLS:
		rquery->begin_result = rctx->num_draw_calls;
		return;
	case R600_QUERY_REQUESTED_VRAM:
	case R600_QUERY_REQUESTED_GTT:
	case R600_QUERY_VRAM_USAGE:
	case R600_QUERY_GTT_USAGE:
		rquery->begin_result = 0;
		return;
	case R600_QUERY_BUFFER_WAIT_TIME:
		rquery->begin_result = rctx->ws->query_value(rctx->ws, RADEON_BUFFER_WAIT_TIME_NS);
		return;
	case R600_QUERY_NUM_CS_FLUSHES:
		rquery->begin_result = rctx->ws->query_value(rctx->ws, RADEON_NUM_CS_FLUSHES);
		return;
	case R600_QUERY_NUM_BYTES_MOVED:
		rquery->begin_result = rctx->ws->query_value(rctx->ws, RADEON_NUM_BYTES_MOVED);
		return;
	}

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

	/* Obtain a new buffer if the current one can't be mapped without a stall. */
	if (r600_rings_is_buffer_referenced(rctx, rquery->buffer.buf->cs_buf, RADEON_USAGE_READWRITE) ||
	    rctx->ws->buffer_is_busy(rquery->buffer.buf->buf, RADEON_USAGE_READWRITE)) {
		pipe_resource_reference((struct pipe_resource**)&rquery->buffer.buf, NULL);
		rquery->buffer.buf = r600_new_query_buffer(rctx, rquery->type);
	}

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

	r600_emit_query_begin(rctx, rquery);

	if (!r600_is_timer_query(rquery->type)) {
		LIST_ADDTAIL(&rquery->list, &rctx->active_nontimer_queries);
	}
}

static void 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;

	/* Non-GPU queries. */
	switch (rquery->type) {
	case PIPE_QUERY_TIMESTAMP_DISJOINT:
		return;
	case R600_QUERY_DRAW_CALLS:
		rquery->end_result = rctx->num_draw_calls;
		return;
	case R600_QUERY_REQUESTED_VRAM:
		rquery->end_result = rctx->ws->query_value(rctx->ws, RADEON_REQUESTED_VRAM_MEMORY);
		return;
	case R600_QUERY_REQUESTED_GTT:
		rquery->end_result = rctx->ws->query_value(rctx->ws, RADEON_REQUESTED_GTT_MEMORY);
		return;
	case R600_QUERY_BUFFER_WAIT_TIME:
		rquery->end_result = rctx->ws->query_value(rctx->ws, RADEON_BUFFER_WAIT_TIME_NS);
		return;
	case R600_QUERY_NUM_CS_FLUSHES:
		rquery->end_result = rctx->ws->query_value(rctx->ws, RADEON_NUM_CS_FLUSHES);
		return;
	case R600_QUERY_NUM_BYTES_MOVED:
		rquery->end_result = rctx->ws->query_value(rctx->ws, RADEON_NUM_BYTES_MOVED);
		return;
	case R600_QUERY_VRAM_USAGE:
		rquery->end_result = rctx->ws->query_value(rctx->ws, RADEON_VRAM_USAGE);
		return;
	case R600_QUERY_GTT_USAGE:
		rquery->end_result = rctx->ws->query_value(rctx->ws, RADEON_GTT_USAGE);
		return;
	}

	r600_emit_query_end(rctx, rquery);

	if (r600_query_needs_begin(rquery->type) && !r600_is_timer_query(rquery->type)) {
		LIST_DELINIT(&rquery->list);
	}
}

static unsigned r600_query_read_result(char *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 boolean r600_get_query_buffer_result(struct r600_common_context *ctx,
					    struct r600_query *query,
					    struct r600_query_buffer *qbuf,
					    boolean wait,
					    union pipe_query_result *result)
{
	unsigned results_base = 0;
	char *map;

	/* Non-GPU queries. */
	switch (query->type) {
	case PIPE_QUERY_TIMESTAMP_DISJOINT:
		/* Convert from cycles per millisecond to cycles per second (Hz). */
		result->timestamp_disjoint.frequency =
			(uint64_t)ctx->screen->info.r600_clock_crystal_freq * 1000;
		result->timestamp_disjoint.disjoint = FALSE;
		return TRUE;
	case R600_QUERY_DRAW_CALLS:
	case R600_QUERY_REQUESTED_VRAM:
	case R600_QUERY_REQUESTED_GTT:
	case R600_QUERY_BUFFER_WAIT_TIME:
	case R600_QUERY_NUM_CS_FLUSHES:
	case R600_QUERY_NUM_BYTES_MOVED:
	case R600_QUERY_VRAM_USAGE:
	case R600_QUERY_GTT_USAGE:
		result->u64 = query->end_result - query->begin_result;
		return TRUE;
	}

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

	/* count all results across all data blocks */
	switch (query->type) {
	case PIPE_QUERY_OCCLUSION_COUNTER:
		while (results_base != qbuf->results_end) {
			result->u64 +=
				r600_query_read_result(map + results_base, 0, 2, true);
			results_base += 16;
		}
		break;
	case PIPE_QUERY_OCCLUSION_PREDICATE:
		while (results_base != qbuf->results_end) {
			result->b = result->b ||
				r600_query_read_result(map + results_base, 0, 2, true) != 0;
			results_base += 16;
		}
		break;
	case PIPE_QUERY_GPU_FINISHED:
		result->b = TRUE;
		break;
	case PIPE_QUERY_TIME_ELAPSED:
		while (results_base != qbuf->results_end) {
			result->u64 +=
				r600_query_read_result(map + results_base, 0, 2, false);
			results_base += query->result_size;
		}
		break;
	case PIPE_QUERY_TIMESTAMP:
	{
		uint32_t *current_result = (uint32_t*)map;
		result->u64 = (uint64_t)current_result[0] |
			      (uint64_t)current_result[1] << 32;
		break;
	}
	case PIPE_QUERY_PRIMITIVES_EMITTED:
		/* SAMPLE_STREAMOUTSTATS stores this structure:
		 * {
		 *    u64 NumPrimitivesWritten;
		 *    u64 PrimitiveStorageNeeded;
		 * }
		 * We only need NumPrimitivesWritten here. */
		while (results_base != qbuf->results_end) {
			result->u64 +=
				r600_query_read_result(map + results_base, 2, 6, true);
			results_base += query->result_size;
		}
		break;
	case PIPE_QUERY_PRIMITIVES_GENERATED:
		/* Here we read PrimitiveStorageNeeded. */
		while (results_base != qbuf->results_end) {
			result->u64 +=
				r600_query_read_result(map + results_base, 0, 4, true);
			results_base += query->result_size;
		}
		break;
	case PIPE_QUERY_SO_STATISTICS:
		while (results_base != qbuf->results_end) {
			result->so_statistics.num_primitives_written +=
				r600_query_read_result(map + results_base, 2, 6, true);
			result->so_statistics.primitives_storage_needed +=
				r600_query_read_result(map + results_base, 0, 4, true);
			results_base += query->result_size;
		}
		break;
	case PIPE_QUERY_SO_OVERFLOW_PREDICATE:
		while (results_base != qbuf->results_end) {
			result->b = result->b ||
				r600_query_read_result(map + results_base, 2, 6, true) !=
				r600_query_read_result(map + results_base, 0, 4, true);
			results_base += query->result_size;
		}
		break;
	case PIPE_QUERY_PIPELINE_STATISTICS:
		if (ctx->chip_class >= EVERGREEN) {
			while (results_base != qbuf->results_end) {
				result->pipeline_statistics.ps_invocations +=
					r600_query_read_result(map + results_base, 0, 22, false);
				result->pipeline_statistics.c_primitives +=
					r600_query_read_result(map + results_base, 2, 24, false);
				result->pipeline_statistics.c_invocations +=
					r600_query_read_result(map + results_base, 4, 26, false);
				result->pipeline_statistics.vs_invocations +=
					r600_query_read_result(map + results_base, 6, 28, false);
				result->pipeline_statistics.gs_invocations +=
					r600_query_read_result(map + results_base, 8, 30, false);
				result->pipeline_statistics.gs_primitives +=
					r600_query_read_result(map + results_base, 10, 32, false);
				result->pipeline_statistics.ia_primitives +=
					r600_query_read_result(map + results_base, 12, 34, false);
				result->pipeline_statistics.ia_vertices +=
					r600_query_read_result(map + results_base, 14, 36, false);
				result->pipeline_statistics.hs_invocations +=
					r600_query_read_result(map + results_base, 16, 38, false);
				result->pipeline_statistics.ds_invocations +=
					r600_query_read_result(map + results_base, 18, 40, false);
				result->pipeline_statistics.cs_invocations +=
					r600_query_read_result(map + results_base, 20, 42, false);
				results_base += query->result_size;
			}
		} else {
			while (results_base != qbuf->results_end) {
				result->pipeline_statistics.ps_invocations +=
					r600_query_read_result(map + results_base, 0, 16, false);
				result->pipeline_statistics.c_primitives +=
					r600_query_read_result(map + results_base, 2, 18, false);
				result->pipeline_statistics.c_invocations +=
					r600_query_read_result(map + results_base, 4, 20, false);
				result->pipeline_statistics.vs_invocations +=
					r600_query_read_result(map + results_base, 6, 22, false);
				result->pipeline_statistics.gs_invocations +=
					r600_query_read_result(map + results_base, 8, 24, false);
				result->pipeline_statistics.gs_primitives +=
					r600_query_read_result(map + results_base, 10, 26, false);
				result->pipeline_statistics.ia_primitives +=
					r600_query_read_result(map + results_base, 12, 28, false);
				result->pipeline_statistics.ia_vertices +=
					r600_query_read_result(map + results_base, 14, 30, false);
				results_base += query->result_size;
			}
		}
#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);
	}

	ctx->ws->buffer_unmap(qbuf->buf->cs_buf);
	return TRUE;
}

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;
	struct r600_query_buffer *qbuf;

	util_query_clear_result(result, rquery->type);

	for (qbuf = &rquery->buffer; qbuf; qbuf = qbuf->previous) {
		if (!r600_get_query_buffer_result(rctx, rquery, qbuf, wait, result)) {
			return FALSE;
		}
	}

	/* 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.r600_clock_crystal_freq;
	}
	return TRUE;
}

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 *rquery = (struct r600_query *)query;
	bool wait_flag = false;

	rctx->current_render_cond = query;
	rctx->current_render_cond_cond = condition;
	rctx->current_render_cond_mode = mode;

	if (query == NULL) {
		if (rctx->predicate_drawing) {
			rctx->predicate_drawing = false;
			r600_emit_query_predication(rctx, NULL, PREDICATION_OP_CLEAR, false);
		}
		return;
	}

	if (mode == PIPE_RENDER_COND_WAIT ||
	    mode == PIPE_RENDER_COND_BY_REGION_WAIT) {
		wait_flag = true;
	}

	rctx->predicate_drawing = true;

	switch (rquery->type) {
	case PIPE_QUERY_OCCLUSION_COUNTER:
	case PIPE_QUERY_OCCLUSION_PREDICATE:
		r600_emit_query_predication(rctx, rquery, PREDICATION_OP_ZPASS, wait_flag);
		break;
	case PIPE_QUERY_PRIMITIVES_EMITTED:
	case PIPE_QUERY_PRIMITIVES_GENERATED:
	case PIPE_QUERY_SO_STATISTICS:
	case PIPE_QUERY_SO_OVERFLOW_PREDICATE:
		r600_emit_query_predication(rctx, rquery, PREDICATION_OP_PRIMCOUNT, wait_flag);
		break;
	default:
		assert(0);
	}
}

void r600_suspend_nontimer_queries(struct r600_common_context *ctx)
{
	struct r600_query *query;

	LIST_FOR_EACH_ENTRY(query, &ctx->active_nontimer_queries, list) {
		r600_emit_query_end(ctx, query);
	}
	assert(ctx->num_cs_dw_nontimer_queries_suspend == 0);
}

static unsigned r600_queries_num_cs_dw_for_resuming(struct r600_common_context *ctx)
{
	struct r600_query *query;
	unsigned num_dw = 0;

	LIST_FOR_EACH_ENTRY(query, &ctx->active_nontimer_queries, list) {
		/* begin + end */
		num_dw += query->num_cs_dw * 2;

		/* 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;
	}
	/* 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_nontimer_queries(struct r600_common_context *ctx)
{
	struct r600_query *query;

	assert(ctx->num_cs_dw_nontimer_queries_suspend == 0);

	/* Check CS space here. Resuming must not be interrupted by flushes. */
	ctx->need_gfx_cs_space(&ctx->b,
			       r600_queries_num_cs_dw_for_resuming(ctx), TRUE);

	LIST_FOR_EACH_ENTRY(query, &ctx->active_nontimer_queries, list) {
		r600_emit_query_begin(ctx, query);
	}
}

/* Get backends mask */
void r600_query_init_backend_mask(struct r600_common_context *ctx)
{
	struct radeon_winsys_cs *cs = ctx->rings.gfx.cs;
	struct r600_resource *buffer;
	uint32_t *results;
	unsigned num_backends = ctx->screen->info.r600_num_backends;
	unsigned i, mask = 0;

	/* if backend_map query is supported by the kernel */
	if (ctx->screen->info.r600_backend_map_valid) {
		unsigned num_tile_pipes = ctx->screen->info.r600_num_tile_pipes;
		unsigned backend_map = ctx->screen->info.r600_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, PIPE_BIND_CUSTOM,
				   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);
		ctx->ws->buffer_unmap(buffer->cs_buf);

		/* 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->rings.gfx, buffer, RADEON_USAGE_WRITE, RADEON_PRIO_MIN);

		/* 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);
			}
			ctx->ws->buffer_unmap(buffer->cs_buf);
		}
	}

	pipe_resource_reference((struct pipe_resource**)&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;
}

void r600_query_init(struct r600_common_context *rctx)
{
	rctx->b.create_query = r600_create_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;

	if (((struct r600_common_screen*)rctx->b.screen)->info.r600_num_backends > 0)
	    rctx->b.render_condition = r600_render_condition;

	LIST_INITHEAD(&rctx->active_nontimer_queries);
}