WebSVN – Kolibri OS – Diff – /drivers/video/drm/radeon/r300.c


/*
 * Copyright 2008 Advanced Micro Devices, Inc.
 * Copyright 2008 Red Hat Inc.
 * Copyright 2009 Jerome Glisse.
 *
 * 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
 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
 * 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 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 NONINFRINGEMENT.  IN NO EVENT SHALL
 * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) 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.
 *
 * Authors: Dave Airlie
 *          Alex Deucher
 *          Jerome Glisse
 */
#include 
#include "drmP.h"
#include "drm.h"
#include "radeon_reg.h"
#include "radeon.h"
#include "radeon_drm.h"
 
#include "r300d.h"
#include "rv350d.h"
#include "r300_reg_safe.h"
 
/* This files gather functions specifics to: r300,r350,rv350,rv370,rv380
 *
 * GPU Errata:
 * - HOST_PATH_CNTL: r300 family seems to dislike write to HOST_PATH_CNTL
 *   using MMIO to flush host path read cache, this lead to HARDLOCKUP.
 *   However, scheduling such write to the ring seems harmless, i suspect
 *   the CP read collide with the flush somehow, or maybe the MC, hard to
 *   tell. (Jerome Glisse)
 */
 
/*
 * rv370,rv380 PCIE GART
 */
static int rv370_debugfs_pcie_gart_info_init(struct radeon_device *rdev);
 
void rv370_pcie_gart_tlb_flush(struct radeon_device *rdev)
{
	uint32_t tmp;
	int i;
 
	/* Workaround HW bug do flush 2 times */
	for (i = 0; i < 2; i++) {
		tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_CNTL);
		WREG32_PCIE(RADEON_PCIE_TX_GART_CNTL, tmp | RADEON_PCIE_TX_GART_INVALIDATE_TLB);
		(void)RREG32_PCIE(RADEON_PCIE_TX_GART_CNTL);
		WREG32_PCIE(RADEON_PCIE_TX_GART_CNTL, tmp);
	}
		mb();
}
 
int rv370_pcie_gart_set_page(struct radeon_device *rdev, int i, uint64_t addr)
{
	void __iomem *ptr = (void *)rdev->gart.table.vram.ptr;
 
	if (i < 0 || i > rdev->gart.num_gpu_pages) {
		return -EINVAL;
	}
	addr = (lower_32_bits(addr) >> 8) |
	       ((upper_32_bits(addr) & 0xff) << 24) |
	       0xc;
	/* on x86 we want this to be CPU endian, on powerpc
	 * on powerpc without HW swappers, it'll get swapped on way
	 * into VRAM - so no need for cpu_to_le32 on VRAM tables */
	writel(addr, ((void __iomem *)ptr) + (i * 4));
	return 0;
}
 
int rv370_pcie_gart_init(struct radeon_device *rdev)
{
	int r;
 
	if (rdev->gart.table.vram.robj) {
		WARN(1, "RV370 PCIE GART already initialized.\n");
		return 0;
	}
	/* Initialize common gart structure */
	r = radeon_gart_init(rdev);
	if (r)
		return r;
	r = rv370_debugfs_pcie_gart_info_init(rdev);
	if (r)
		DRM_ERROR("Failed to register debugfs file for PCIE gart !\n");
	rdev->gart.table_size = rdev->gart.num_gpu_pages * 4;
	rdev->asic->gart_tlb_flush = &rv370_pcie_gart_tlb_flush;
	rdev->asic->gart_set_page = &rv370_pcie_gart_set_page;
	return radeon_gart_table_vram_alloc(rdev);
}
 
int rv370_pcie_gart_enable(struct radeon_device *rdev)
{
	uint32_t table_addr;
	uint32_t tmp;
	int r;
 
	if (rdev->gart.table.vram.robj == NULL) {
		dev_err(rdev->dev, "No VRAM object for PCIE GART.\n");
		return -EINVAL;
	}
	r = radeon_gart_table_vram_pin(rdev);
	if (r)
		return r;
	/* discard memory request outside of configured range */
	tmp = RADEON_PCIE_TX_GART_UNMAPPED_ACCESS_DISCARD;
	WREG32_PCIE(RADEON_PCIE_TX_GART_CNTL, tmp);
	WREG32_PCIE(RADEON_PCIE_TX_GART_START_LO, rdev->mc.gtt_location);
	tmp = rdev->mc.gtt_location + rdev->mc.gtt_size - RADEON_GPU_PAGE_SIZE;
	WREG32_PCIE(RADEON_PCIE_TX_GART_END_LO, tmp);
	WREG32_PCIE(RADEON_PCIE_TX_GART_START_HI, 0);
	WREG32_PCIE(RADEON_PCIE_TX_GART_END_HI, 0);
	table_addr = rdev->gart.table_addr;
	WREG32_PCIE(RADEON_PCIE_TX_GART_BASE, table_addr);
	/* FIXME: setup default page */
	WREG32_PCIE(RADEON_PCIE_TX_DISCARD_RD_ADDR_LO, rdev->mc.vram_location);
	WREG32_PCIE(RADEON_PCIE_TX_DISCARD_RD_ADDR_HI, 0);
	/* Clear error */
	WREG32_PCIE(0x18, 0);
	tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_CNTL);
	tmp |= RADEON_PCIE_TX_GART_EN;
	tmp |= RADEON_PCIE_TX_GART_UNMAPPED_ACCESS_DISCARD;
	WREG32_PCIE(RADEON_PCIE_TX_GART_CNTL, tmp);
	rv370_pcie_gart_tlb_flush(rdev);
	DRM_INFO("PCIE GART of %uM enabled (table at 0x%08X).\n",
		 (unsigned)(rdev->mc.gtt_size >> 20), table_addr);
	rdev->gart.ready = true;
	return 0;
}
 
void rv370_pcie_gart_disable(struct radeon_device *rdev)
{
	u32 tmp;
	int r;
 
	tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_CNTL);
	tmp |= RADEON_PCIE_TX_GART_UNMAPPED_ACCESS_DISCARD;
	WREG32_PCIE(RADEON_PCIE_TX_GART_CNTL, tmp & ~RADEON_PCIE_TX_GART_EN);
	if (rdev->gart.table.vram.robj) {
		r = radeon_bo_reserve(rdev->gart.table.vram.robj, false);
		if (likely(r == 0)) {
			radeon_bo_kunmap(rdev->gart.table.vram.robj);
			radeon_bo_unpin(rdev->gart.table.vram.robj);
			radeon_bo_unreserve(rdev->gart.table.vram.robj);
		}
	}
}
 
void rv370_pcie_gart_fini(struct radeon_device *rdev)
{
			rv370_pcie_gart_disable(rdev);
	radeon_gart_table_vram_free(rdev);
	radeon_gart_fini(rdev);
}
 
void r300_fence_ring_emit(struct radeon_device *rdev,
			  struct radeon_fence *fence)
{
	/* Who ever call radeon_fence_emit should call ring_lock and ask
	 * for enough space (today caller are ib schedule and buffer move) */
	/* Write SC register so SC & US assert idle */
	radeon_ring_write(rdev, PACKET0(0x43E0, 0));
	radeon_ring_write(rdev, 0);
	radeon_ring_write(rdev, PACKET0(0x43E4, 0));
	radeon_ring_write(rdev, 0);
	/* Flush 3D cache */
	radeon_ring_write(rdev, PACKET0(0x4E4C, 0));
	radeon_ring_write(rdev, (2 << 0));
	radeon_ring_write(rdev, PACKET0(0x4F18, 0));
	radeon_ring_write(rdev, (1 << 0));
	/* Wait until IDLE & CLEAN */
	radeon_ring_write(rdev, PACKET0(0x1720, 0));
	radeon_ring_write(rdev, (1 << 17) | (1 << 16)  | (1 << 9));
	radeon_ring_write(rdev, PACKET0(RADEON_HOST_PATH_CNTL, 0));
	radeon_ring_write(rdev, rdev->config.r300.hdp_cntl |
				RADEON_HDP_READ_BUFFER_INVALIDATE);
	radeon_ring_write(rdev, PACKET0(RADEON_HOST_PATH_CNTL, 0));
	radeon_ring_write(rdev, rdev->config.r300.hdp_cntl);
	/* Emit fence sequence & fire IRQ */
	radeon_ring_write(rdev, PACKET0(rdev->fence_drv.scratch_reg, 0));
	radeon_ring_write(rdev, fence->seq);
	radeon_ring_write(rdev, PACKET0(RADEON_GEN_INT_STATUS, 0));
	radeon_ring_write(rdev, RADEON_SW_INT_FIRE);
}
 
 
#if 0
 
 
int r300_copy_dma(struct radeon_device *rdev,
		  uint64_t src_offset,
		  uint64_t dst_offset,
		  unsigned num_pages,
		  struct radeon_fence *fence)
{
	uint32_t size;
	uint32_t cur_size;
	int i, num_loops;
	int r = 0;
 
	/* radeon pitch is /64 */
	size = num_pages << PAGE_SHIFT;
	num_loops = DIV_ROUND_UP(size, 0x1FFFFF);
	r = radeon_ring_lock(rdev, num_loops * 4 + 64);
	if (r) {
		DRM_ERROR("radeon: moving bo (%d).\n", r);
		return r;
	}
	/* Must wait for 2D idle & clean before DMA or hangs might happen */
	radeon_ring_write(rdev, PACKET0(RADEON_WAIT_UNTIL, 0 ));
	radeon_ring_write(rdev, (1 << 16));
	for (i = 0; i < num_loops; i++) {
		cur_size = size;
		if (cur_size > 0x1FFFFF) {
			cur_size = 0x1FFFFF;
		}
		size -= cur_size;
		radeon_ring_write(rdev, PACKET0(0x720, 2));
		radeon_ring_write(rdev, src_offset);
		radeon_ring_write(rdev, dst_offset);
		radeon_ring_write(rdev, cur_size | (1 << 31) | (1 << 30));
		src_offset += cur_size;
		dst_offset += cur_size;
	}
	radeon_ring_write(rdev, PACKET0(RADEON_WAIT_UNTIL, 0));
	radeon_ring_write(rdev, RADEON_WAIT_DMA_GUI_IDLE);
	if (fence) {
		r = radeon_fence_emit(rdev, fence);
	}
	radeon_ring_unlock_commit(rdev);
	return r;
}
 
#endif
 
void r300_ring_start(struct radeon_device *rdev)
{
	unsigned gb_tile_config;
	int r;
 
	/* Sub pixel 1/12 so we can have 4K rendering according to doc */
	gb_tile_config = (R300_ENABLE_TILING | R300_TILE_SIZE_16);
	switch(rdev->num_gb_pipes) {
	case 2:
		gb_tile_config |= R300_PIPE_COUNT_R300;
		break;
	case 3:
		gb_tile_config |= R300_PIPE_COUNT_R420_3P;
		break;
	case 4:
		gb_tile_config |= R300_PIPE_COUNT_R420;
		break;
	case 1:
	default:
		gb_tile_config |= R300_PIPE_COUNT_RV350;
		break;
	}
 
	r = radeon_ring_lock(rdev, 64);
	if (r) {
		return;
	}
	radeon_ring_write(rdev, PACKET0(RADEON_ISYNC_CNTL, 0));
	radeon_ring_write(rdev,
			  RADEON_ISYNC_ANY2D_IDLE3D |
			  RADEON_ISYNC_ANY3D_IDLE2D |
			  RADEON_ISYNC_WAIT_IDLEGUI |
			  RADEON_ISYNC_CPSCRATCH_IDLEGUI);
	radeon_ring_write(rdev, PACKET0(R300_GB_TILE_CONFIG, 0));
	radeon_ring_write(rdev, gb_tile_config);
	radeon_ring_write(rdev, PACKET0(RADEON_WAIT_UNTIL, 0));
	radeon_ring_write(rdev,
			  RADEON_WAIT_2D_IDLECLEAN |
			  RADEON_WAIT_3D_IDLECLEAN);
	radeon_ring_write(rdev, PACKET0(0x170C, 0));
	radeon_ring_write(rdev, 1 << 31);
	radeon_ring_write(rdev, PACKET0(R300_GB_SELECT, 0));
	radeon_ring_write(rdev, 0);
	radeon_ring_write(rdev, PACKET0(R300_GB_ENABLE, 0));
	radeon_ring_write(rdev, 0);
	radeon_ring_write(rdev, PACKET0(R300_RB3D_DSTCACHE_CTLSTAT, 0));
	radeon_ring_write(rdev, R300_RB3D_DC_FLUSH | R300_RB3D_DC_FREE);
	radeon_ring_write(rdev, PACKET0(R300_RB3D_ZCACHE_CTLSTAT, 0));
	radeon_ring_write(rdev, R300_ZC_FLUSH | R300_ZC_FREE);
	radeon_ring_write(rdev, PACKET0(RADEON_WAIT_UNTIL, 0));
	radeon_ring_write(rdev,
			  RADEON_WAIT_2D_IDLECLEAN |
			  RADEON_WAIT_3D_IDLECLEAN);
	radeon_ring_write(rdev, PACKET0(R300_GB_AA_CONFIG, 0));
	radeon_ring_write(rdev, 0);
	radeon_ring_write(rdev, PACKET0(R300_RB3D_DSTCACHE_CTLSTAT, 0));
	radeon_ring_write(rdev, R300_RB3D_DC_FLUSH | R300_RB3D_DC_FREE);
	radeon_ring_write(rdev, PACKET0(R300_RB3D_ZCACHE_CTLSTAT, 0));
	radeon_ring_write(rdev, R300_ZC_FLUSH | R300_ZC_FREE);
	radeon_ring_write(rdev, PACKET0(R300_GB_MSPOS0, 0));
	radeon_ring_write(rdev,
			  ((6 << R300_MS_X0_SHIFT) |
			   (6 << R300_MS_Y0_SHIFT) |
			   (6 << R300_MS_X1_SHIFT) |
			   (6 << R300_MS_Y1_SHIFT) |
			   (6 << R300_MS_X2_SHIFT) |
			   (6 << R300_MS_Y2_SHIFT) |
			   (6 << R300_MSBD0_Y_SHIFT) |
			   (6 << R300_MSBD0_X_SHIFT)));
	radeon_ring_write(rdev, PACKET0(R300_GB_MSPOS1, 0));
	radeon_ring_write(rdev,
			  ((6 << R300_MS_X3_SHIFT) |
			   (6 << R300_MS_Y3_SHIFT) |
			   (6 << R300_MS_X4_SHIFT) |
			   (6 << R300_MS_Y4_SHIFT) |
			   (6 << R300_MS_X5_SHIFT) |
			   (6 << R300_MS_Y5_SHIFT) |
			   (6 << R300_MSBD1_SHIFT)));
	radeon_ring_write(rdev, PACKET0(R300_GA_ENHANCE, 0));
	radeon_ring_write(rdev, R300_GA_DEADLOCK_CNTL | R300_GA_FASTSYNC_CNTL);
	radeon_ring_write(rdev, PACKET0(R300_GA_POLY_MODE, 0));
	radeon_ring_write(rdev,
			  R300_FRONT_PTYPE_TRIANGE | R300_BACK_PTYPE_TRIANGE);
	radeon_ring_write(rdev, PACKET0(R300_GA_ROUND_MODE, 0));
	radeon_ring_write(rdev,
			  R300_GEOMETRY_ROUND_NEAREST |
			  R300_COLOR_ROUND_NEAREST);
	radeon_ring_unlock_commit(rdev);
}
 
void r300_errata(struct radeon_device *rdev)
{
	rdev->pll_errata = 0;
 
	if (rdev->family == CHIP_R300 &&
	    (RREG32(RADEON_CONFIG_CNTL) & RADEON_CFG_ATI_REV_ID_MASK) == RADEON_CFG_ATI_REV_A11) {
		rdev->pll_errata |= CHIP_ERRATA_R300_CG;
	}
}
 
int r300_mc_wait_for_idle(struct radeon_device *rdev)
{
	unsigned i;
	uint32_t tmp;
 
	for (i = 0; i < rdev->usec_timeout; i++) {
		/* read MC_STATUS */
		tmp = RREG32(0x0150);
		if (tmp & (1 << 4)) {
			return 0;
		}
		DRM_UDELAY(1);
	}
	return -1;
}
 
void r300_gpu_init(struct radeon_device *rdev)
{
	uint32_t gb_tile_config, tmp;
 
	r100_hdp_reset(rdev);
	/* FIXME: rv380 one pipes ? */
	if ((rdev->family == CHIP_R300) || (rdev->family == CHIP_R350)) {
		/* r300,r350 */
		rdev->num_gb_pipes = 2;
	} else {
		/* rv350,rv370,rv380 */
		rdev->num_gb_pipes = 1;
	}
	rdev->num_z_pipes = 1;
	gb_tile_config = (R300_ENABLE_TILING | R300_TILE_SIZE_16);
	switch (rdev->num_gb_pipes) {
	case 2:
		gb_tile_config |= R300_PIPE_COUNT_R300;
		break;
	case 3:
		gb_tile_config |= R300_PIPE_COUNT_R420_3P;
		break;
	case 4:
		gb_tile_config |= R300_PIPE_COUNT_R420;
		break;
	default:
	case 1:
		gb_tile_config |= R300_PIPE_COUNT_RV350;
		break;
	}
	WREG32(R300_GB_TILE_CONFIG, gb_tile_config);
 
	if (r100_gui_wait_for_idle(rdev)) {
		printk(KERN_WARNING "Failed to wait GUI idle while "
		       "programming pipes. Bad things might happen.\n");
	}
 
	tmp = RREG32(0x170C);
	WREG32(0x170C, tmp | (1 << 31));
 
	WREG32(R300_RB2D_DSTCACHE_MODE,
	       R300_DC_AUTOFLUSH_ENABLE |
	       R300_DC_DC_DISABLE_IGNORE_PE);
 
	if (r100_gui_wait_for_idle(rdev)) {
		printk(KERN_WARNING "Failed to wait GUI idle while "
		       "programming pipes. Bad things might happen.\n");
	}
	if (r300_mc_wait_for_idle(rdev)) {
		printk(KERN_WARNING "Failed to wait MC idle while "
		       "programming pipes. Bad things might happen.\n");
	}
	DRM_INFO("radeon: %d quad pipes, %d Z pipes initialized.\n",
		 rdev->num_gb_pipes, rdev->num_z_pipes);
}
 
int r300_ga_reset(struct radeon_device *rdev)
{
	uint32_t tmp;
	bool reinit_cp;
	int i;
 
	reinit_cp = rdev->cp.ready;
	rdev->cp.ready = false;
	for (i = 0; i < rdev->usec_timeout; i++) {
		WREG32(RADEON_CP_CSQ_MODE, 0);
		WREG32(RADEON_CP_CSQ_CNTL, 0);
		WREG32(RADEON_RBBM_SOFT_RESET, 0x32005);
		(void)RREG32(RADEON_RBBM_SOFT_RESET);
		udelay(200);
		WREG32(RADEON_RBBM_SOFT_RESET, 0);
		/* Wait to prevent race in RBBM_STATUS */
		mdelay(1);
		tmp = RREG32(RADEON_RBBM_STATUS);
		if (tmp & ((1 << 20) | (1 << 26))) {
			DRM_ERROR("VAP & CP still busy (RBBM_STATUS=0x%08X)", tmp);
			/* GA still busy soft reset it */
			WREG32(0x429C, 0x200);
			WREG32(R300_VAP_PVS_STATE_FLUSH_REG, 0);
			WREG32(0x43E0, 0);
			WREG32(0x43E4, 0);
			WREG32(0x24AC, 0);
		}
		/* Wait to prevent race in RBBM_STATUS */
		mdelay(1);
		tmp = RREG32(RADEON_RBBM_STATUS);
		if (!(tmp & ((1 << 20) | (1 << 26)))) {
			break;
		}
	}
	for (i = 0; i < rdev->usec_timeout; i++) {
		tmp = RREG32(RADEON_RBBM_STATUS);
		if (!(tmp & ((1 << 20) | (1 << 26)))) {
			DRM_INFO("GA reset succeed (RBBM_STATUS=0x%08X)\n",
				 tmp);
			if (reinit_cp) {
				return r100_cp_init(rdev, rdev->cp.ring_size);
			}
			return 0;
		}
		DRM_UDELAY(1);
	}
	tmp = RREG32(RADEON_RBBM_STATUS);
	DRM_ERROR("Failed to reset GA ! (RBBM_STATUS=0x%08X)\n", tmp);
	return -1;
}
 
int r300_gpu_reset(struct radeon_device *rdev)
{
	uint32_t status;
 
	/* reset order likely matter */
	status = RREG32(RADEON_RBBM_STATUS);
	/* reset HDP */
	r100_hdp_reset(rdev);
	/* reset rb2d */
	if (status & ((1 << 17) | (1 << 18) | (1 << 27))) {
		r100_rb2d_reset(rdev);
	}
	/* reset GA */
	if (status & ((1 << 20) | (1 << 26))) {
		r300_ga_reset(rdev);
	}
	/* reset CP */
	status = RREG32(RADEON_RBBM_STATUS);
	if (status & (1 << 16)) {
		r100_cp_reset(rdev);
	}
	/* Check if GPU is idle */
	status = RREG32(RADEON_RBBM_STATUS);
	if (status & (1 << 31)) {
		DRM_ERROR("Failed to reset GPU (RBBM_STATUS=0x%08X)\n", status);
		return -1;
	}
	DRM_INFO("GPU reset succeed (RBBM_STATUS=0x%08X)\n", status);
	return 0;
}
 
 
/*
 * r300,r350,rv350,rv380 VRAM info
 */
void r300_vram_info(struct radeon_device *rdev)
{
	uint32_t tmp;
 
	/* DDR for all card after R300 & IGP */
	rdev->mc.vram_is_ddr = true;
 
	tmp = RREG32(RADEON_MEM_CNTL);
	tmp &= R300_MEM_NUM_CHANNELS_MASK;
	switch (tmp) {
	case 0: rdev->mc.vram_width = 64; break;
	case 1: rdev->mc.vram_width = 128; break;
	case 2: rdev->mc.vram_width = 256; break;
	default:  rdev->mc.vram_width = 128; break;
	}
 
	r100_vram_init_sizes(rdev);
}
 
void rv370_set_pcie_lanes(struct radeon_device *rdev, int lanes)
{
	uint32_t link_width_cntl, mask;
 
	if (rdev->flags & RADEON_IS_IGP)
		return;
 
	if (!(rdev->flags & RADEON_IS_PCIE))
		return;
 
	/* FIXME wait for idle */
 
	switch (lanes) {
	case 0:
		mask = RADEON_PCIE_LC_LINK_WIDTH_X0;
		break;
	case 1:
		mask = RADEON_PCIE_LC_LINK_WIDTH_X1;
		break;
	case 2:
		mask = RADEON_PCIE_LC_LINK_WIDTH_X2;
		break;
	case 4:
		mask = RADEON_PCIE_LC_LINK_WIDTH_X4;
		break;
	case 8:
		mask = RADEON_PCIE_LC_LINK_WIDTH_X8;
		break;
	case 12:
		mask = RADEON_PCIE_LC_LINK_WIDTH_X12;
		break;
	case 16:
	default:
		mask = RADEON_PCIE_LC_LINK_WIDTH_X16;
		break;
	}
 
	link_width_cntl = RREG32_PCIE(RADEON_PCIE_LC_LINK_WIDTH_CNTL);
 
	if ((link_width_cntl & RADEON_PCIE_LC_LINK_WIDTH_RD_MASK) ==
	    (mask << RADEON_PCIE_LC_LINK_WIDTH_RD_SHIFT))
		return;
 
	link_width_cntl &= ~(RADEON_PCIE_LC_LINK_WIDTH_MASK |
			     RADEON_PCIE_LC_RECONFIG_NOW |
			     RADEON_PCIE_LC_RECONFIG_LATER |
			     RADEON_PCIE_LC_SHORT_RECONFIG_EN);
	link_width_cntl |= mask;
	WREG32_PCIE(RADEON_PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
	WREG32_PCIE(RADEON_PCIE_LC_LINK_WIDTH_CNTL, (link_width_cntl |
						     RADEON_PCIE_LC_RECONFIG_NOW));
 
	/* wait for lane set to complete */
	link_width_cntl = RREG32_PCIE(RADEON_PCIE_LC_LINK_WIDTH_CNTL);
	while (link_width_cntl == 0xffffffff)
		link_width_cntl = RREG32_PCIE(RADEON_PCIE_LC_LINK_WIDTH_CNTL);
 
}
 
#if defined(CONFIG_DEBUG_FS)
static int rv370_debugfs_pcie_gart_info(struct seq_file *m, void *data)
{
	struct drm_info_node *node = (struct drm_info_node *) m->private;
	struct drm_device *dev = node->minor->dev;
	struct radeon_device *rdev = dev->dev_private;
	uint32_t tmp;
 
	tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_CNTL);
	seq_printf(m, "PCIE_TX_GART_CNTL 0x%08x\n", tmp);
	tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_BASE);
	seq_printf(m, "PCIE_TX_GART_BASE 0x%08x\n", tmp);
	tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_START_LO);
	seq_printf(m, "PCIE_TX_GART_START_LO 0x%08x\n", tmp);
	tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_START_HI);
	seq_printf(m, "PCIE_TX_GART_START_HI 0x%08x\n", tmp);
	tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_END_LO);
	seq_printf(m, "PCIE_TX_GART_END_LO 0x%08x\n", tmp);
	tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_END_HI);
	seq_printf(m, "PCIE_TX_GART_END_HI 0x%08x\n", tmp);
	tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_ERROR);
	seq_printf(m, "PCIE_TX_GART_ERROR 0x%08x\n", tmp);
	return 0;
}
 
static struct drm_info_list rv370_pcie_gart_info_list[] = {
	{"rv370_pcie_gart_info", rv370_debugfs_pcie_gart_info, 0, NULL},
};
#endif
 
static int rv370_debugfs_pcie_gart_info_init(struct radeon_device *rdev)
{
#if defined(CONFIG_DEBUG_FS)
	return radeon_debugfs_add_files(rdev, rv370_pcie_gart_info_list, 1);
#else
	return 0;
#endif
}
 
 
#if 0
 
static int r300_packet0_check(struct radeon_cs_parser *p,
		struct radeon_cs_packet *pkt,
		unsigned idx, unsigned reg)
{
	struct radeon_cs_reloc *reloc;
	struct r100_cs_track *track;
	volatile uint32_t *ib;
	uint32_t tmp, tile_flags = 0;
	unsigned i;
	int r;
	u32 idx_value;
 
	ib = p->ib->ptr;
	track = (struct r100_cs_track *)p->track;
	idx_value = radeon_get_ib_value(p, idx);
 
	switch(reg) {
	case AVIVO_D1MODE_VLINE_START_END:
	case RADEON_CRTC_GUI_TRIG_VLINE:
		r = r100_cs_packet_parse_vline(p);
		if (r) {
			DRM_ERROR("No reloc for ib[%d]=0x%04X\n",
					idx, reg);
			r100_cs_dump_packet(p, pkt);
			return r;
		}
		break;
	case RADEON_DST_PITCH_OFFSET:
	case RADEON_SRC_PITCH_OFFSET:
		r = r100_reloc_pitch_offset(p, pkt, idx, reg);
		if (r)
			return r;
		break;
	case R300_RB3D_COLOROFFSET0:
	case R300_RB3D_COLOROFFSET1:
	case R300_RB3D_COLOROFFSET2:
	case R300_RB3D_COLOROFFSET3:
		i = (reg - R300_RB3D_COLOROFFSET0) >> 2;
		r = r100_cs_packet_next_reloc(p, &reloc);
		if (r) {
			DRM_ERROR("No reloc for ib[%d]=0x%04X\n",
					idx, reg);
			r100_cs_dump_packet(p, pkt);
			return r;
		}
		track->cb[i].robj = reloc->robj;
		track->cb[i].offset = idx_value;
		ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset);
		break;
	case R300_ZB_DEPTHOFFSET:
		r = r100_cs_packet_next_reloc(p, &reloc);
		if (r) {
			DRM_ERROR("No reloc for ib[%d]=0x%04X\n",
					idx, reg);
			r100_cs_dump_packet(p, pkt);
			return r;
		}
		track->zb.robj = reloc->robj;
		track->zb.offset = idx_value;
		ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset);
		break;
	case R300_TX_OFFSET_0:
	case R300_TX_OFFSET_0+4:
	case R300_TX_OFFSET_0+8:
	case R300_TX_OFFSET_0+12:
	case R300_TX_OFFSET_0+16:
	case R300_TX_OFFSET_0+20:
	case R300_TX_OFFSET_0+24:
	case R300_TX_OFFSET_0+28:
	case R300_TX_OFFSET_0+32:
	case R300_TX_OFFSET_0+36:
	case R300_TX_OFFSET_0+40:
	case R300_TX_OFFSET_0+44:
	case R300_TX_OFFSET_0+48:
	case R300_TX_OFFSET_0+52:
	case R300_TX_OFFSET_0+56:
	case R300_TX_OFFSET_0+60:
		i = (reg - R300_TX_OFFSET_0) >> 2;
		r = r100_cs_packet_next_reloc(p, &reloc);
		if (r) {
			DRM_ERROR("No reloc for ib[%d]=0x%04X\n",
					idx, reg);
			r100_cs_dump_packet(p, pkt);
			return r;
		}
 
		if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO)
			tile_flags |= R300_TXO_MACRO_TILE;
		if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO)
			tile_flags |= R300_TXO_MICRO_TILE;
 
		tmp = idx_value + ((u32)reloc->lobj.gpu_offset);
		tmp |= tile_flags;
		ib[idx] = tmp;
		track->textures[i].robj = reloc->robj;
		break;
	/* Tracked registers */
	case 0x2084:
		/* VAP_VF_CNTL */
		track->vap_vf_cntl = idx_value;
		break;
	case 0x20B4:
		/* VAP_VTX_SIZE */
		track->vtx_size = idx_value & 0x7F;
		break;
	case 0x2134:
		/* VAP_VF_MAX_VTX_INDX */
		track->max_indx = idx_value & 0x00FFFFFFUL;
		break;
	case 0x43E4:
		/* SC_SCISSOR1 */
		track->maxy = ((idx_value >> 13) & 0x1FFF) + 1;
		if (p->rdev->family < CHIP_RV515) {
			track->maxy -= 1440;
		}
		break;
	case 0x4E00:
		/* RB3D_CCTL */
		track->num_cb = ((idx_value >> 5) & 0x3) + 1;
		break;
	case 0x4E38:
	case 0x4E3C:
	case 0x4E40:
	case 0x4E44:
		/* RB3D_COLORPITCH0 */
		/* RB3D_COLORPITCH1 */
		/* RB3D_COLORPITCH2 */
		/* RB3D_COLORPITCH3 */
		r = r100_cs_packet_next_reloc(p, &reloc);
		if (r) {
			DRM_ERROR("No reloc for ib[%d]=0x%04X\n",
				  idx, reg);
			r100_cs_dump_packet(p, pkt);
			return r;
		}
 
		if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO)
			tile_flags |= R300_COLOR_TILE_ENABLE;
		if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO)
			tile_flags |= R300_COLOR_MICROTILE_ENABLE;
 
		tmp = idx_value & ~(0x7 << 16);
		tmp |= tile_flags;
		ib[idx] = tmp;
 
		i = (reg - 0x4E38) >> 2;
		track->cb[i].pitch = idx_value & 0x3FFE;
		switch (((idx_value >> 21) & 0xF)) {
		case 9:
		case 11:
		case 12:
			track->cb[i].cpp = 1;
			break;
		case 3:
		case 4:
		case 13:
		case 15:
			track->cb[i].cpp = 2;
			break;
		case 6:
			track->cb[i].cpp = 4;
			break;
		case 10:
			track->cb[i].cpp = 8;
			break;
		case 7:
			track->cb[i].cpp = 16;
			break;
		default:
			DRM_ERROR("Invalid color buffer format (%d) !\n",
				  ((idx_value >> 21) & 0xF));
			return -EINVAL;
		}
		break;
	case 0x4F00:
		/* ZB_CNTL */
		if (idx_value & 2) {
			track->z_enabled = true;
		} else {
			track->z_enabled = false;
		}
		break;
	case 0x4F10:
		/* ZB_FORMAT */
		switch ((idx_value & 0xF)) {
		case 0:
		case 1:
			track->zb.cpp = 2;
			break;
		case 2:
			track->zb.cpp = 4;
			break;
		default:
			DRM_ERROR("Invalid z buffer format (%d) !\n",
				  (idx_value & 0xF));
			return -EINVAL;
		}
		break;
	case 0x4F24:
		/* ZB_DEPTHPITCH */
		r = r100_cs_packet_next_reloc(p, &reloc);
		if (r) {
			DRM_ERROR("No reloc for ib[%d]=0x%04X\n",
				  idx, reg);
			r100_cs_dump_packet(p, pkt);
			return r;
		}
 
		if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO)
			tile_flags |= R300_DEPTHMACROTILE_ENABLE;
		if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO)
			tile_flags |= R300_DEPTHMICROTILE_TILED;;
 
		tmp = idx_value & ~(0x7 << 16);
		tmp |= tile_flags;
		ib[idx] = tmp;
 
		track->zb.pitch = idx_value & 0x3FFC;
		break;
	case 0x4104:
		for (i = 0; i < 16; i++) {
			bool enabled;
 
			enabled = !!(idx_value & (1 << i));
			track->textures[i].enabled = enabled;
		}
		break;
	case 0x44C0:
	case 0x44C4:
	case 0x44C8:
	case 0x44CC:
	case 0x44D0:
	case 0x44D4:
	case 0x44D8:
	case 0x44DC:
	case 0x44E0:
	case 0x44E4:
	case 0x44E8:
	case 0x44EC:
	case 0x44F0:
	case 0x44F4:
	case 0x44F8:
	case 0x44FC:
		/* TX_FORMAT1_[0-15] */
		i = (reg - 0x44C0) >> 2;
		tmp = (idx_value >> 25) & 0x3;
		track->textures[i].tex_coord_type = tmp;
		switch ((idx_value & 0x1F)) {
		case R300_TX_FORMAT_X8:
		case R300_TX_FORMAT_Y4X4:
		case R300_TX_FORMAT_Z3Y3X2:
			track->textures[i].cpp = 1;
			break;
		case R300_TX_FORMAT_X16:
		case R300_TX_FORMAT_Y8X8:
		case R300_TX_FORMAT_Z5Y6X5:
		case R300_TX_FORMAT_Z6Y5X5:
		case R300_TX_FORMAT_W4Z4Y4X4:
		case R300_TX_FORMAT_W1Z5Y5X5:
		case R300_TX_FORMAT_D3DMFT_CxV8U8:
		case R300_TX_FORMAT_B8G8_B8G8:
		case R300_TX_FORMAT_G8R8_G8B8:
			track->textures[i].cpp = 2;
			break;
		case R300_TX_FORMAT_Y16X16:
		case R300_TX_FORMAT_Z11Y11X10:
		case R300_TX_FORMAT_Z10Y11X11:
		case R300_TX_FORMAT_W8Z8Y8X8:
		case R300_TX_FORMAT_W2Z10Y10X10:
		case 0x17:
		case R300_TX_FORMAT_FL_I32:
		case 0x1e:
			track->textures[i].cpp = 4;
			break;
		case R300_TX_FORMAT_W16Z16Y16X16:
		case R300_TX_FORMAT_FL_R16G16B16A16:
		case R300_TX_FORMAT_FL_I32A32:
			track->textures[i].cpp = 8;
			break;
		case R300_TX_FORMAT_FL_R32G32B32A32:
			track->textures[i].cpp = 16;
			break;
		case R300_TX_FORMAT_DXT1:
			track->textures[i].cpp = 1;
			track->textures[i].compress_format = R100_TRACK_COMP_DXT1;
			break;
		case R300_TX_FORMAT_ATI2N:
			if (p->rdev->family < CHIP_R420) {
				DRM_ERROR("Invalid texture format %u\n",
					  (idx_value & 0x1F));
				return -EINVAL;
			}
			/* The same rules apply as for DXT3/5. */
			/* Pass through. */
		case R300_TX_FORMAT_DXT3:
		case R300_TX_FORMAT_DXT5:
			track->textures[i].cpp = 1;
			track->textures[i].compress_format = R100_TRACK_COMP_DXT35;
			break;
		default:
			DRM_ERROR("Invalid texture format %u\n",
				  (idx_value & 0x1F));
			return -EINVAL;
			break;
		}
		break;
	case 0x4400:
	case 0x4404:
	case 0x4408:
	case 0x440C:
	case 0x4410:
	case 0x4414:
	case 0x4418:
	case 0x441C:
	case 0x4420:
	case 0x4424:
	case 0x4428:
	case 0x442C:
	case 0x4430:
	case 0x4434:
	case 0x4438:
	case 0x443C:
		/* TX_FILTER0_[0-15] */
		i = (reg - 0x4400) >> 2;
		tmp = idx_value & 0x7;
		if (tmp == 2 || tmp == 4 || tmp == 6) {
			track->textures[i].roundup_w = false;
		}
		tmp = (idx_value >> 3) & 0x7;
		if (tmp == 2 || tmp == 4 || tmp == 6) {
			track->textures[i].roundup_h = false;
		}
		break;
	case 0x4500:
	case 0x4504:
	case 0x4508:
	case 0x450C:
	case 0x4510:
	case 0x4514:
	case 0x4518:
	case 0x451C:
	case 0x4520:
	case 0x4524:
	case 0x4528:
	case 0x452C:
	case 0x4530:
	case 0x4534:
	case 0x4538:
	case 0x453C:
		/* TX_FORMAT2_[0-15] */
		i = (reg - 0x4500) >> 2;
		tmp = idx_value & 0x3FFF;
		track->textures[i].pitch = tmp + 1;
		if (p->rdev->family >= CHIP_RV515) {
			tmp = ((idx_value >> 15) & 1) << 11;
			track->textures[i].width_11 = tmp;
			tmp = ((idx_value >> 16) & 1) << 11;
			track->textures[i].height_11 = tmp;
 
			/* ATI1N */
			if (idx_value & (1 << 14)) {
				/* The same rules apply as for DXT1. */
				track->textures[i].compress_format =
					R100_TRACK_COMP_DXT1;
			}
		} else if (idx_value & (1 << 14)) {
			DRM_ERROR("Forbidden bit TXFORMAT_MSB\n");
			return -EINVAL;
		}
		break;
	case 0x4480:
	case 0x4484:
	case 0x4488:
	case 0x448C:
	case 0x4490:
	case 0x4494:
	case 0x4498:
	case 0x449C:
	case 0x44A0:
	case 0x44A4:
	case 0x44A8:
	case 0x44AC:
	case 0x44B0:
	case 0x44B4:
	case 0x44B8:
	case 0x44BC:
		/* TX_FORMAT0_[0-15] */
		i = (reg - 0x4480) >> 2;
		tmp = idx_value & 0x7FF;
		track->textures[i].width = tmp + 1;
		tmp = (idx_value >> 11) & 0x7FF;
		track->textures[i].height = tmp + 1;
		tmp = (idx_value >> 26) & 0xF;
		track->textures[i].num_levels = tmp;
		tmp = idx_value & (1 << 31);
		track->textures[i].use_pitch = !!tmp;
		tmp = (idx_value >> 22) & 0xF;
		track->textures[i].txdepth = tmp;
		break;
	case R300_ZB_ZPASS_ADDR:
		r = r100_cs_packet_next_reloc(p, &reloc);
		if (r) {
			DRM_ERROR("No reloc for ib[%d]=0x%04X\n",
					idx, reg);
			r100_cs_dump_packet(p, pkt);
			return r;
		}
		ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset);
		break;
	case 0x4e0c:
		/* RB3D_COLOR_CHANNEL_MASK */
		track->color_channel_mask = idx_value;
		break;
	case 0x4d1c:
		/* ZB_BW_CNTL */
		track->fastfill = !!(idx_value & (1 << 2));
		break;
	case 0x4e04:
		/* RB3D_BLENDCNTL */
		track->blend_read_enable = !!(idx_value & (1 << 2));
		break;
	case 0x4be8:
		/* valid register only on RV530 */
		if (p->rdev->family == CHIP_RV530)
			break;
		/* fallthrough do not move */
	default:
		printk(KERN_ERR "Forbidden register 0x%04X in cs at %d\n",
		       reg, idx);
		return -EINVAL;
	}
	return 0;
}
 
static int r300_packet3_check(struct radeon_cs_parser *p,
			      struct radeon_cs_packet *pkt)
{
	struct radeon_cs_reloc *reloc;
	struct r100_cs_track *track;
	volatile uint32_t *ib;
	unsigned idx;
	int r;
 
	ib = p->ib->ptr;
	idx = pkt->idx + 1;
	track = (struct r100_cs_track *)p->track;
	switch(pkt->opcode) {
	case PACKET3_3D_LOAD_VBPNTR:
		r = r100_packet3_load_vbpntr(p, pkt, idx);
		if (r)
				return r;
		break;
	case PACKET3_INDX_BUFFER:
		r = r100_cs_packet_next_reloc(p, &reloc);
		if (r) {
			DRM_ERROR("No reloc for packet3 %d\n", pkt->opcode);
			r100_cs_dump_packet(p, pkt);
			return r;
		}
		ib[idx+1] = radeon_get_ib_value(p, idx + 1) + ((u32)reloc->lobj.gpu_offset);
		r = r100_cs_track_check_pkt3_indx_buffer(p, pkt, reloc->robj);
		if (r) {
			return r;
		}
		break;
	/* Draw packet */
	case PACKET3_3D_DRAW_IMMD:
		/* Number of dwords is vtx_size * (num_vertices - 1)
		 * PRIM_WALK must be equal to 3 vertex data in embedded
		 * in cmd stream */
		if (((radeon_get_ib_value(p, idx + 1) >> 4) & 0x3) != 3) {
			DRM_ERROR("PRIM_WALK must be 3 for IMMD draw\n");
			return -EINVAL;
		}
		track->vap_vf_cntl = radeon_get_ib_value(p, idx + 1);
		track->immd_dwords = pkt->count - 1;
		r = r100_cs_track_check(p->rdev, track);
		if (r) {
			return r;
		}
		break;
	case PACKET3_3D_DRAW_IMMD_2:
		/* Number of dwords is vtx_size * (num_vertices - 1)
		 * PRIM_WALK must be equal to 3 vertex data in embedded
		 * in cmd stream */
		if (((radeon_get_ib_value(p, idx) >> 4) & 0x3) != 3) {
			DRM_ERROR("PRIM_WALK must be 3 for IMMD draw\n");
			return -EINVAL;
		}
		track->vap_vf_cntl = radeon_get_ib_value(p, idx);
		track->immd_dwords = pkt->count;
		r = r100_cs_track_check(p->rdev, track);
		if (r) {
			return r;
		}
		break;
	case PACKET3_3D_DRAW_VBUF:
		track->vap_vf_cntl = radeon_get_ib_value(p, idx + 1);
		r = r100_cs_track_check(p->rdev, track);
		if (r) {
			return r;
		}
		break;
	case PACKET3_3D_DRAW_VBUF_2:
		track->vap_vf_cntl = radeon_get_ib_value(p, idx);
		r = r100_cs_track_check(p->rdev, track);
		if (r) {
			return r;
		}
		break;
	case PACKET3_3D_DRAW_INDX:
		track->vap_vf_cntl = radeon_get_ib_value(p, idx + 1);
		r = r100_cs_track_check(p->rdev, track);
		if (r) {
			return r;
		}
		break;
	case PACKET3_3D_DRAW_INDX_2:
		track->vap_vf_cntl = radeon_get_ib_value(p, idx);
		r = r100_cs_track_check(p->rdev, track);
		if (r) {
			return r;
		}
		break;
	case PACKET3_NOP:
		break;
	default:
		DRM_ERROR("Packet3 opcode %x not supported\n", pkt->opcode);
		return -EINVAL;
	}
	return 0;
}
 
int r300_cs_parse(struct radeon_cs_parser *p)
{
	struct radeon_cs_packet pkt;
	struct r100_cs_track *track;
	int r;
 
	track = kzalloc(sizeof(*track), GFP_KERNEL);
	r100_cs_track_clear(p->rdev, track);
	p->track = track;
	do {
		r = r100_cs_packet_parse(p, &pkt, p->idx);
		if (r) {
			return r;
		}
		p->idx += pkt.count + 2;
		switch (pkt.type) {
		case PACKET_TYPE0:
			r = r100_cs_parse_packet0(p, &pkt,
						  p->rdev->config.r300.reg_safe_bm,
						  p->rdev->config.r300.reg_safe_bm_size,
						  &r300_packet0_check);
			break;
		case PACKET_TYPE2:
			break;
		case PACKET_TYPE3:
			r = r300_packet3_check(p, &pkt);
			break;
		default:
			DRM_ERROR("Unknown packet type %d !\n", pkt.type);
			return -EINVAL;
		}
		if (r) {
			return r;
		}
	} while (p->idx < p->chunks[p->chunk_ib_idx].length_dw);
	return 0;
}
#endif
 
 
void r300_set_reg_safe(struct radeon_device *rdev)
{
	rdev->config.r300.reg_safe_bm = r300_reg_safe_bm;
	rdev->config.r300.reg_safe_bm_size = ARRAY_SIZE(r300_reg_safe_bm);
}
 
void r300_mc_program(struct radeon_device *rdev)
{
	struct r100_mc_save save;
	int r;
 
	r = r100_debugfs_mc_info_init(rdev);
	if (r) {
		dev_err(rdev->dev, "Failed to create r100_mc debugfs file.\n");
	}
 
	/* Stops all mc clients */
	r100_mc_stop(rdev, &save);
	if (rdev->flags & RADEON_IS_AGP) {
		WREG32(R_00014C_MC_AGP_LOCATION,
			S_00014C_MC_AGP_START(rdev->mc.gtt_start >> 16) |
			S_00014C_MC_AGP_TOP(rdev->mc.gtt_end >> 16));
		WREG32(R_000170_AGP_BASE, lower_32_bits(rdev->mc.agp_base));
		WREG32(R_00015C_AGP_BASE_2,
			upper_32_bits(rdev->mc.agp_base) & 0xff);
	} else {
		WREG32(R_00014C_MC_AGP_LOCATION, 0x0FFFFFFF);
		WREG32(R_000170_AGP_BASE, 0);
		WREG32(R_00015C_AGP_BASE_2, 0);
	}
	/* Wait for mc idle */
	if (r300_mc_wait_for_idle(rdev))
		DRM_INFO("Failed to wait MC idle before programming MC.\n");
	/* Program MC, should be a 32bits limited address space */
	WREG32(R_000148_MC_FB_LOCATION,
		S_000148_MC_FB_START(rdev->mc.vram_start >> 16) |
		S_000148_MC_FB_TOP(rdev->mc.vram_end >> 16));
	r100_mc_resume(rdev, &save);
}
 
void r300_clock_startup(struct radeon_device *rdev)
{
	u32 tmp;
 
	if (radeon_dynclks != -1 && radeon_dynclks)
		radeon_legacy_set_clock_gating(rdev, 1);
	/* We need to force on some of the block */
	tmp = RREG32_PLL(R_00000D_SCLK_CNTL);
	tmp |= S_00000D_FORCE_CP(1) | S_00000D_FORCE_VIP(1);
	if ((rdev->family == CHIP_RV350) || (rdev->family == CHIP_RV380))
		tmp |= S_00000D_FORCE_VAP(1);
	WREG32_PLL(R_00000D_SCLK_CNTL, tmp);
}
 
static int r300_startup(struct radeon_device *rdev)
{
	int r;
 
	/* set common regs */
	r100_set_common_regs(rdev);
	/* program mc */
	r300_mc_program(rdev);
	/* Resume clock */
	r300_clock_startup(rdev);
	/* Initialize GPU configuration (# pipes, ...) */
	r300_gpu_init(rdev);
	/* Initialize GART (initialize after TTM so we can allocate
	 * memory through TTM but finalize after TTM) */
	if (rdev->flags & RADEON_IS_PCIE) {
		r = rv370_pcie_gart_enable(rdev);
		if (r)
			return r;
	}
 
	if (rdev->family == CHIP_R300 ||
	    rdev->family == CHIP_R350 ||
	    rdev->family == CHIP_RV350)
		r100_enable_bm(rdev);
 
	if (rdev->flags & RADEON_IS_PCI) {
		r = r100_pci_gart_enable(rdev);
		if (r)
			return r;
	}
	/* Enable IRQ */
//	r100_irq_set(rdev);
	rdev->config.r300.hdp_cntl = RREG32(RADEON_HOST_PATH_CNTL);
	/* 1M ring buffer */
    r = r100_cp_init(rdev, 1024 * 1024);
    if (r) {
       dev_err(rdev->dev, "failled initializing CP (%d).\n", r);
       return r;
    }
//   r = r100_wb_init(rdev);
//   if (r)
//       dev_err(rdev->dev, "failled initializing WB (%d).\n", r);
//   r = r100_ib_init(rdev);
//   if (r) {
//       dev_err(rdev->dev, "failled initializing IB (%d).\n", r);
//       return r;
//   }
	return 0;
}
 
 
 
 
 
int r300_init(struct radeon_device *rdev)
{
	int r;
 
	/* Disable VGA */
	r100_vga_render_disable(rdev);
	/* Initialize scratch registers */
	radeon_scratch_init(rdev);
	/* Initialize surface registers */
	radeon_surface_init(rdev);
	/* TODO: disable VGA need to use VGA request */
	/* BIOS*/
	if (!radeon_get_bios(rdev)) {
		if (ASIC_IS_AVIVO(rdev))
			return -EINVAL;
	}
	if (rdev->is_atom_bios) {
		dev_err(rdev->dev, "Expecting combios for RS400/RS480 GPU\n");
		return -EINVAL;
	} else {
		r = radeon_combios_init(rdev);
		if (r)
			return r;
	}
	/* Reset gpu before posting otherwise ATOM will enter infinite loop */
	if (radeon_gpu_reset(rdev)) {
		dev_warn(rdev->dev,
			"GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n",
			RREG32(R_000E40_RBBM_STATUS),
			RREG32(R_0007C0_CP_STAT));
	}
	/* check if cards are posted or not */
	if (radeon_boot_test_post_card(rdev) == false)
		return -EINVAL;
	/* Set asic errata */
	r300_errata(rdev);
	/* Initialize clocks */
	radeon_get_clock_info(rdev->ddev);
	/* Initialize power management */
	radeon_pm_init(rdev);
	/* Get vram informations */
	r300_vram_info(rdev);
	/* Initialize memory controller (also test AGP) */
	r = r420_mc_init(rdev);
    dbgprintf("mc vram location %x\n", rdev->mc.vram_location);
	if (r)
		return r;
	/* Fence driver */
//	r = radeon_fence_driver_init(rdev);
//	if (r)
//		return r;
//	r = radeon_irq_kms_init(rdev);
//	if (r)
//		return r;
	/* Memory manager */
	r = radeon_bo_init(rdev);
	if (r)
		return r;
	if (rdev->flags & RADEON_IS_PCIE) {
		r = rv370_pcie_gart_init(rdev);
		if (r)
			return r;
	}
	if (rdev->flags & RADEON_IS_PCI) {
		r = r100_pci_gart_init(rdev);
		if (r)
			return r;
	}
	r300_set_reg_safe(rdev);
	rdev->accel_working = true;
	r = r300_startup(rdev);
	if (r) {
		/* Somethings want wront with the accel init stop accel */
		dev_err(rdev->dev, "Disabling GPU acceleration\n");
 
//		r100_cp_fini(rdev);
//		r100_wb_fini(rdev);
//		r100_ib_fini(rdev);
		if (rdev->flags & RADEON_IS_PCIE)
			rv370_pcie_gart_fini(rdev);
		if (rdev->flags & RADEON_IS_PCI)
			r100_pci_gart_fini(rdev);
//		radeon_agp_fini(rdev);
		rdev->accel_working = false;
	}
	return 0;
}

Rev 1412	Rev 1414
1	/*	1	/*
2	* Copyright 2008 Advanced Micro Devices, Inc.	2	* Copyright 2008 Advanced Micro Devices, Inc.
3	* Copyright 2008 Red Hat Inc.	3	* Copyright 2008 Red Hat Inc.
4	* Copyright 2009 Jerome Glisse.	4	* Copyright 2009 Jerome Glisse.
5	*	5	*
6	* Permission is hereby granted, free of charge, to any person obtaining a	6	* Permission is hereby granted, free of charge, to any person obtaining a
7	* copy of this software and associated documentation files (the "Software"),	7	* copy of this software and associated documentation files (the "Software"),
8	* to deal in the Software without restriction, including without limitation	8	* to deal in the Software without restriction, including without limitation
9	* the rights to use, copy, modify, merge, publish, distribute, sublicense,	9	* the rights to use, copy, modify, merge, publish, distribute, sublicense,
10	* and/or sell copies of the Software, and to permit persons to whom the	10	* and/or sell copies of the Software, and to permit persons to whom the
11	* Software is furnished to do so, subject to the following conditions:	11	* Software is furnished to do so, subject to the following conditions:
12	*	12	*
13	* The above copyright notice and this permission notice shall be included in	13	* The above copyright notice and this permission notice shall be included in
14	* all copies or substantial portions of the Software.	14	* all copies or substantial portions of the Software.
15	*	15	*
16	* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR	16	* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
17	* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,	17	* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
18	* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL	18	* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
19	* THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR	19	* THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
20	* OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,	20	* OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
21	* ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR	21	* ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
22	* OTHER DEALINGS IN THE SOFTWARE.	22	* OTHER DEALINGS IN THE SOFTWARE.
23	*	23	*
24	* Authors: Dave Airlie	24	* Authors: Dave Airlie
25	* Alex Deucher	25	* Alex Deucher
26	* Jerome Glisse	26	* Jerome Glisse
27	*/	27	*/
28	#include	28	#include
29	#include "drmP.h"	29	#include "drmP.h"
30	#include "drm.h"	30	#include "drm.h"
31	#include "radeon_reg.h"	31	#include "radeon_reg.h"
32	#include "radeon.h"	32	#include "radeon.h"
33	#include "radeon_drm.h"	33	#include "radeon_drm.h"
34		34
35	#include "r300d.h"	35	#include "r300d.h"
36	#include "rv350d.h"	36	#include "rv350d.h"
37	#include "r300_reg_safe.h"	37	#include "r300_reg_safe.h"
38		38
39	/* This files gather functions specifics to: r300,r350,rv350,rv370,rv380	39	/* This files gather functions specifics to: r300,r350,rv350,rv370,rv380
40	*	40	*
41	* GPU Errata:	41	* GPU Errata:
42	* - HOST_PATH_CNTL: r300 family seems to dislike write to HOST_PATH_CNTL	42	* - HOST_PATH_CNTL: r300 family seems to dislike write to HOST_PATH_CNTL
43	* using MMIO to flush host path read cache, this lead to HARDLOCKUP.	43	* using MMIO to flush host path read cache, this lead to HARDLOCKUP.
44	* However, scheduling such write to the ring seems harmless, i suspect	44	* However, scheduling such write to the ring seems harmless, i suspect
45	* the CP read collide with the flush somehow, or maybe the MC, hard to	45	* the CP read collide with the flush somehow, or maybe the MC, hard to
46	* tell. (Jerome Glisse)	46	* tell. (Jerome Glisse)
47	*/	47	*/
48		48
49	/*	49	/*
50	* rv370,rv380 PCIE GART	50	* rv370,rv380 PCIE GART
51	*/	51	*/
52	static int rv370_debugfs_pcie_gart_info_init(struct radeon_device *rdev);	52	static int rv370_debugfs_pcie_gart_info_init(struct radeon_device *rdev);
53		53
54	void rv370_pcie_gart_tlb_flush(struct radeon_device *rdev)	54	void rv370_pcie_gart_tlb_flush(struct radeon_device *rdev)
55	{	55	{
56	uint32_t tmp;	56	uint32_t tmp;
57	int i;	57	int i;
58		58
59	/* Workaround HW bug do flush 2 times */	59	/* Workaround HW bug do flush 2 times */
60	for (i = 0; i < 2; i++) {	60	for (i = 0; i < 2; i++) {
61	tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_CNTL);	61	tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_CNTL);
62	WREG32_PCIE(RADEON_PCIE_TX_GART_CNTL, tmp \| RADEON_PCIE_TX_GART_INVALIDATE_TLB);	62	WREG32_PCIE(RADEON_PCIE_TX_GART_CNTL, tmp \| RADEON_PCIE_TX_GART_INVALIDATE_TLB);
63	(void)RREG32_PCIE(RADEON_PCIE_TX_GART_CNTL);	63	(void)RREG32_PCIE(RADEON_PCIE_TX_GART_CNTL);
64	WREG32_PCIE(RADEON_PCIE_TX_GART_CNTL, tmp);	64	WREG32_PCIE(RADEON_PCIE_TX_GART_CNTL, tmp);
65	}	65	}
66	mb();	66	mb();
67	}	67	}
68		68
69	int rv370_pcie_gart_set_page(struct radeon_device *rdev, int i, uint64_t addr)	69	int rv370_pcie_gart_set_page(struct radeon_device *rdev, int i, uint64_t addr)
70	{	70	{
71	void __iomem ptr = (void )rdev->gart.table.vram.ptr;	71	void __iomem ptr = (void )rdev->gart.table.vram.ptr;
72		72
73	if (i < 0 \|\| i > rdev->gart.num_gpu_pages) {	73	if (i < 0 \|\| i > rdev->gart.num_gpu_pages) {
74	return -EINVAL;	74	return -EINVAL;
75	}	75	}
76	addr = (lower_32_bits(addr) >> 8) \|	76	addr = (lower_32_bits(addr) >> 8) \|
77	((upper_32_bits(addr) & 0xff) << 24) \|	77	((upper_32_bits(addr) & 0xff) << 24) \|
78	0xc;	78	0xc;
79	/* on x86 we want this to be CPU endian, on powerpc	79	/* on x86 we want this to be CPU endian, on powerpc
80	* on powerpc without HW swappers, it'll get swapped on way	80	* on powerpc without HW swappers, it'll get swapped on way
81	* into VRAM - so no need for cpu_to_le32 on VRAM tables */	81	* into VRAM - so no need for cpu_to_le32 on VRAM tables */
82	writel(addr, ((void __iomem )ptr) + (i 4));	82	writel(addr, ((void __iomem )ptr) + (i 4));
83	return 0;	83	return 0;
84	}	84	}
85		85
86	int rv370_pcie_gart_init(struct radeon_device *rdev)	86	int rv370_pcie_gart_init(struct radeon_device *rdev)
87	{	87	{
88	int r;	88	int r;
89		89
90	if (rdev->gart.table.vram.robj) {	90	if (rdev->gart.table.vram.robj) {
91	WARN(1, "RV370 PCIE GART already initialized.\n");	91	WARN(1, "RV370 PCIE GART already initialized.\n");
92	return 0;	92	return 0;
93	}	93	}
94	/* Initialize common gart structure */	94	/* Initialize common gart structure */
95	r = radeon_gart_init(rdev);	95	r = radeon_gart_init(rdev);
96	if (r)	96	if (r)
97	return r;	97	return r;
98	r = rv370_debugfs_pcie_gart_info_init(rdev);	98	r = rv370_debugfs_pcie_gart_info_init(rdev);
99	if (r)	99	if (r)
100	DRM_ERROR("Failed to register debugfs file for PCIE gart !\n");	100	DRM_ERROR("Failed to register debugfs file for PCIE gart !\n");
101	rdev->gart.table_size = rdev->gart.num_gpu_pages * 4;	101	rdev->gart.table_size = rdev->gart.num_gpu_pages * 4;
102	rdev->asic->gart_tlb_flush = &rv370_pcie_gart_tlb_flush;	102	rdev->asic->gart_tlb_flush = &rv370_pcie_gart_tlb_flush;
103	rdev->asic->gart_set_page = &rv370_pcie_gart_set_page;	103	rdev->asic->gart_set_page = &rv370_pcie_gart_set_page;
104	return radeon_gart_table_vram_alloc(rdev);	104	return radeon_gart_table_vram_alloc(rdev);
105	}	105	}
106		106
107	int rv370_pcie_gart_enable(struct radeon_device *rdev)	107	int rv370_pcie_gart_enable(struct radeon_device *rdev)
108	{	108	{
109	uint32_t table_addr;	109	uint32_t table_addr;
110	uint32_t tmp;	110	uint32_t tmp;
111	int r;	111	int r;
112		112
113	if (rdev->gart.table.vram.robj == NULL) {	113	if (rdev->gart.table.vram.robj == NULL) {
114	dev_err(rdev->dev, "No VRAM object for PCIE GART.\n");	114	dev_err(rdev->dev, "No VRAM object for PCIE GART.\n");
115	return -EINVAL;	115	return -EINVAL;
116	}	116	}
117	r = radeon_gart_table_vram_pin(rdev);	117	r = radeon_gart_table_vram_pin(rdev);
118	if (r)	118	if (r)
119	return r;	119	return r;
120	/* discard memory request outside of configured range */	120	/* discard memory request outside of configured range */
121	tmp = RADEON_PCIE_TX_GART_UNMAPPED_ACCESS_DISCARD;	121	tmp = RADEON_PCIE_TX_GART_UNMAPPED_ACCESS_DISCARD;
122	WREG32_PCIE(RADEON_PCIE_TX_GART_CNTL, tmp);	122	WREG32_PCIE(RADEON_PCIE_TX_GART_CNTL, tmp);
123	WREG32_PCIE(RADEON_PCIE_TX_GART_START_LO, rdev->mc.gtt_location);	123	WREG32_PCIE(RADEON_PCIE_TX_GART_START_LO, rdev->mc.gtt_location);
124	tmp = rdev->mc.gtt_location + rdev->mc.gtt_size - RADEON_GPU_PAGE_SIZE;	124	tmp = rdev->mc.gtt_location + rdev->mc.gtt_size - RADEON_GPU_PAGE_SIZE;
125	WREG32_PCIE(RADEON_PCIE_TX_GART_END_LO, tmp);	125	WREG32_PCIE(RADEON_PCIE_TX_GART_END_LO, tmp);
126	WREG32_PCIE(RADEON_PCIE_TX_GART_START_HI, 0);	126	WREG32_PCIE(RADEON_PCIE_TX_GART_START_HI, 0);
127	WREG32_PCIE(RADEON_PCIE_TX_GART_END_HI, 0);	127	WREG32_PCIE(RADEON_PCIE_TX_GART_END_HI, 0);
128	table_addr = rdev->gart.table_addr;	128	table_addr = rdev->gart.table_addr;
129	WREG32_PCIE(RADEON_PCIE_TX_GART_BASE, table_addr);	129	WREG32_PCIE(RADEON_PCIE_TX_GART_BASE, table_addr);
130	/* FIXME: setup default page */	130	/* FIXME: setup default page */
131	WREG32_PCIE(RADEON_PCIE_TX_DISCARD_RD_ADDR_LO, rdev->mc.vram_location);	131	WREG32_PCIE(RADEON_PCIE_TX_DISCARD_RD_ADDR_LO, rdev->mc.vram_location);
132	WREG32_PCIE(RADEON_PCIE_TX_DISCARD_RD_ADDR_HI, 0);	132	WREG32_PCIE(RADEON_PCIE_TX_DISCARD_RD_ADDR_HI, 0);
133	/* Clear error */	133	/* Clear error */
134	WREG32_PCIE(0x18, 0);	134	WREG32_PCIE(0x18, 0);
135	tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_CNTL);	135	tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_CNTL);
136	tmp \|= RADEON_PCIE_TX_GART_EN;	136	tmp \|= RADEON_PCIE_TX_GART_EN;
137	tmp \|= RADEON_PCIE_TX_GART_UNMAPPED_ACCESS_DISCARD;	137	tmp \|= RADEON_PCIE_TX_GART_UNMAPPED_ACCESS_DISCARD;
138	WREG32_PCIE(RADEON_PCIE_TX_GART_CNTL, tmp);	138	WREG32_PCIE(RADEON_PCIE_TX_GART_CNTL, tmp);
139	rv370_pcie_gart_tlb_flush(rdev);	139	rv370_pcie_gart_tlb_flush(rdev);
140	DRM_INFO("PCIE GART of %uM enabled (table at 0x%08X).\n",	140	DRM_INFO("PCIE GART of %uM enabled (table at 0x%08X).\n",
141	(unsigned)(rdev->mc.gtt_size >> 20), table_addr);	141	(unsigned)(rdev->mc.gtt_size >> 20), table_addr);
142	rdev->gart.ready = true;	142	rdev->gart.ready = true;
143	return 0;	143	return 0;
144	}	144	}
145		145
146	void rv370_pcie_gart_disable(struct radeon_device *rdev)	146	void rv370_pcie_gart_disable(struct radeon_device *rdev)
147	{	147	{
148	u32 tmp;	148	u32 tmp;
149	int r;	149	int r;
150		150
151	tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_CNTL);	151	tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_CNTL);
152	tmp \|= RADEON_PCIE_TX_GART_UNMAPPED_ACCESS_DISCARD;	152	tmp \|= RADEON_PCIE_TX_GART_UNMAPPED_ACCESS_DISCARD;
153	WREG32_PCIE(RADEON_PCIE_TX_GART_CNTL, tmp & ~RADEON_PCIE_TX_GART_EN);	153	WREG32_PCIE(RADEON_PCIE_TX_GART_CNTL, tmp & ~RADEON_PCIE_TX_GART_EN);
154	if (rdev->gart.table.vram.robj) {	154	if (rdev->gart.table.vram.robj) {
155	r = radeon_bo_reserve(rdev->gart.table.vram.robj, false);	155	r = radeon_bo_reserve(rdev->gart.table.vram.robj, false);
156	if (likely(r == 0)) {	156	if (likely(r == 0)) {
157	radeon_bo_kunmap(rdev->gart.table.vram.robj);	157	radeon_bo_kunmap(rdev->gart.table.vram.robj);
158	radeon_bo_unpin(rdev->gart.table.vram.robj);	158	radeon_bo_unpin(rdev->gart.table.vram.robj);
159	radeon_bo_unreserve(rdev->gart.table.vram.robj);	159	radeon_bo_unreserve(rdev->gart.table.vram.robj);
160	}	160	}
161	}	161	}
162	}	162	}
163		163
164	void rv370_pcie_gart_fini(struct radeon_device *rdev)	164	void rv370_pcie_gart_fini(struct radeon_device *rdev)
165	{	165	{
166	rv370_pcie_gart_disable(rdev);	166	rv370_pcie_gart_disable(rdev);
167	radeon_gart_table_vram_free(rdev);	167	radeon_gart_table_vram_free(rdev);
168	radeon_gart_fini(rdev);	168	radeon_gart_fini(rdev);
169	}	169	}
170		170
171	void r300_fence_ring_emit(struct radeon_device *rdev,	171	void r300_fence_ring_emit(struct radeon_device *rdev,
172	struct radeon_fence *fence)	172	struct radeon_fence *fence)
173	{	173	{
174	/* Who ever call radeon_fence_emit should call ring_lock and ask	174	/* Who ever call radeon_fence_emit should call ring_lock and ask
175	* for enough space (today caller are ib schedule and buffer move) */	175	* for enough space (today caller are ib schedule and buffer move) */
176	/* Write SC register so SC & US assert idle */	176	/* Write SC register so SC & US assert idle */
177	radeon_ring_write(rdev, PACKET0(0x43E0, 0));	177	radeon_ring_write(rdev, PACKET0(0x43E0, 0));
178	radeon_ring_write(rdev, 0);	178	radeon_ring_write(rdev, 0);
179	radeon_ring_write(rdev, PACKET0(0x43E4, 0));	179	radeon_ring_write(rdev, PACKET0(0x43E4, 0));
180	radeon_ring_write(rdev, 0);	180	radeon_ring_write(rdev, 0);
181	/* Flush 3D cache */	181	/* Flush 3D cache */
182	radeon_ring_write(rdev, PACKET0(0x4E4C, 0));	182	radeon_ring_write(rdev, PACKET0(0x4E4C, 0));
183	radeon_ring_write(rdev, (2 << 0));	183	radeon_ring_write(rdev, (2 << 0));
184	radeon_ring_write(rdev, PACKET0(0x4F18, 0));	184	radeon_ring_write(rdev, PACKET0(0x4F18, 0));
185	radeon_ring_write(rdev, (1 << 0));	185	radeon_ring_write(rdev, (1 << 0));
186	/* Wait until IDLE & CLEAN */	186	/* Wait until IDLE & CLEAN */
187	radeon_ring_write(rdev, PACKET0(0x1720, 0));	187	radeon_ring_write(rdev, PACKET0(0x1720, 0));
188	radeon_ring_write(rdev, (1 << 17) \| (1 << 16) \| (1 << 9));	188	radeon_ring_write(rdev, (1 << 17) \| (1 << 16) \| (1 << 9));
189	radeon_ring_write(rdev, PACKET0(RADEON_HOST_PATH_CNTL, 0));	189	radeon_ring_write(rdev, PACKET0(RADEON_HOST_PATH_CNTL, 0));
190	radeon_ring_write(rdev, rdev->config.r300.hdp_cntl \|	190	radeon_ring_write(rdev, rdev->config.r300.hdp_cntl \|
191	RADEON_HDP_READ_BUFFER_INVALIDATE);	191	RADEON_HDP_READ_BUFFER_INVALIDATE);
192	radeon_ring_write(rdev, PACKET0(RADEON_HOST_PATH_CNTL, 0));	192	radeon_ring_write(rdev, PACKET0(RADEON_HOST_PATH_CNTL, 0));
193	radeon_ring_write(rdev, rdev->config.r300.hdp_cntl);	193	radeon_ring_write(rdev, rdev->config.r300.hdp_cntl);
194	/* Emit fence sequence & fire IRQ */	194	/* Emit fence sequence & fire IRQ */
195	radeon_ring_write(rdev, PACKET0(rdev->fence_drv.scratch_reg, 0));	195	radeon_ring_write(rdev, PACKET0(rdev->fence_drv.scratch_reg, 0));
196	radeon_ring_write(rdev, fence->seq);	196	radeon_ring_write(rdev, fence->seq);
197	radeon_ring_write(rdev, PACKET0(RADEON_GEN_INT_STATUS, 0));	197	radeon_ring_write(rdev, PACKET0(RADEON_GEN_INT_STATUS, 0));
198	radeon_ring_write(rdev, RADEON_SW_INT_FIRE);	198	radeon_ring_write(rdev, RADEON_SW_INT_FIRE);
199	}	199	}
200		200
201		201
202	#if 0	202	#if 0
203		203
204		204
205	int r300_copy_dma(struct radeon_device *rdev,	205	int r300_copy_dma(struct radeon_device *rdev,
206	uint64_t src_offset,	206	uint64_t src_offset,
207	uint64_t dst_offset,	207	uint64_t dst_offset,
208	unsigned num_pages,	208	unsigned num_pages,
209	struct radeon_fence *fence)	209	struct radeon_fence *fence)
210	{	210	{
211	uint32_t size;	211	uint32_t size;
212	uint32_t cur_size;	212	uint32_t cur_size;
213	int i, num_loops;	213	int i, num_loops;
214	int r = 0;	214	int r = 0;
215		215
216	/* radeon pitch is /64 */	216	/* radeon pitch is /64 */
217	size = num_pages << PAGE_SHIFT;	217	size = num_pages << PAGE_SHIFT;
218	num_loops = DIV_ROUND_UP(size, 0x1FFFFF);	218	num_loops = DIV_ROUND_UP(size, 0x1FFFFF);
219	r = radeon_ring_lock(rdev, num_loops * 4 + 64);	219	r = radeon_ring_lock(rdev, num_loops * 4 + 64);
220	if (r) {	220	if (r) {
221	DRM_ERROR("radeon: moving bo (%d).\n", r);	221	DRM_ERROR("radeon: moving bo (%d).\n", r);
222	return r;	222	return r;
223	}	223	}
224	/* Must wait for 2D idle & clean before DMA or hangs might happen */	224	/* Must wait for 2D idle & clean before DMA or hangs might happen */
225	radeon_ring_write(rdev, PACKET0(RADEON_WAIT_UNTIL, 0 ));	225	radeon_ring_write(rdev, PACKET0(RADEON_WAIT_UNTIL, 0 ));
226	radeon_ring_write(rdev, (1 << 16));	226	radeon_ring_write(rdev, (1 << 16));
227	for (i = 0; i < num_loops; i++) {	227	for (i = 0; i < num_loops; i++) {
228	cur_size = size;	228	cur_size = size;
229	if (cur_size > 0x1FFFFF) {	229	if (cur_size > 0x1FFFFF) {
230	cur_size = 0x1FFFFF;	230	cur_size = 0x1FFFFF;
231	}	231	}
232	size -= cur_size;	232	size -= cur_size;
233	radeon_ring_write(rdev, PACKET0(0x720, 2));	233	radeon_ring_write(rdev, PACKET0(0x720, 2));
234	radeon_ring_write(rdev, src_offset);	234	radeon_ring_write(rdev, src_offset);
235	radeon_ring_write(rdev, dst_offset);	235	radeon_ring_write(rdev, dst_offset);
236	radeon_ring_write(rdev, cur_size \| (1 << 31) \| (1 << 30));	236	radeon_ring_write(rdev, cur_size \| (1 << 31) \| (1 << 30));
237	src_offset += cur_size;	237	src_offset += cur_size;
238	dst_offset += cur_size;	238	dst_offset += cur_size;
239	}	239	}
240	radeon_ring_write(rdev, PACKET0(RADEON_WAIT_UNTIL, 0));	240	radeon_ring_write(rdev, PACKET0(RADEON_WAIT_UNTIL, 0));
241	radeon_ring_write(rdev, RADEON_WAIT_DMA_GUI_IDLE);	241	radeon_ring_write(rdev, RADEON_WAIT_DMA_GUI_IDLE);
242	if (fence) {	242	if (fence) {
243	r = radeon_fence_emit(rdev, fence);	243	r = radeon_fence_emit(rdev, fence);
244	}	244	}
245	radeon_ring_unlock_commit(rdev);	245	radeon_ring_unlock_commit(rdev);
246	return r;	246	return r;
247	}	247	}
248		248
249	#endif	249	#endif
250		250
251	void r300_ring_start(struct radeon_device *rdev)	251	void r300_ring_start(struct radeon_device *rdev)
252	{	252	{
253	unsigned gb_tile_config;	253	unsigned gb_tile_config;
254	int r;	254	int r;
255		255
256	/* Sub pixel 1/12 so we can have 4K rendering according to doc */	256	/* Sub pixel 1/12 so we can have 4K rendering according to doc */
257	gb_tile_config = (R300_ENABLE_TILING \| R300_TILE_SIZE_16);	257	gb_tile_config = (R300_ENABLE_TILING \| R300_TILE_SIZE_16);
258	switch(rdev->num_gb_pipes) {	258	switch(rdev->num_gb_pipes) {
259	case 2:	259	case 2:
260	gb_tile_config \|= R300_PIPE_COUNT_R300;	260	gb_tile_config \|= R300_PIPE_COUNT_R300;
261	break;	261	break;
262	case 3:	262	case 3:
263	gb_tile_config \|= R300_PIPE_COUNT_R420_3P;	263	gb_tile_config \|= R300_PIPE_COUNT_R420_3P;
264	break;	264	break;
265	case 4:	265	case 4:
266	gb_tile_config \|= R300_PIPE_COUNT_R420;	266	gb_tile_config \|= R300_PIPE_COUNT_R420;
267	break;	267	break;
268	case 1:	268	case 1:
269	default:	269	default:
270	gb_tile_config \|= R300_PIPE_COUNT_RV350;	270	gb_tile_config \|= R300_PIPE_COUNT_RV350;
271	break;	271	break;
272	}	272	}
273		273
274	r = radeon_ring_lock(rdev, 64);	274	r = radeon_ring_lock(rdev, 64);
275	if (r) {	275	if (r) {
276	return;	276	return;
277	}	277	}
278	radeon_ring_write(rdev, PACKET0(RADEON_ISYNC_CNTL, 0));	278	radeon_ring_write(rdev, PACKET0(RADEON_ISYNC_CNTL, 0));
279	radeon_ring_write(rdev,	279	radeon_ring_write(rdev,
280	RADEON_ISYNC_ANY2D_IDLE3D \|	280	RADEON_ISYNC_ANY2D_IDLE3D \|
281	RADEON_ISYNC_ANY3D_IDLE2D \|	281	RADEON_ISYNC_ANY3D_IDLE2D \|
282	RADEON_ISYNC_WAIT_IDLEGUI \|	282	RADEON_ISYNC_WAIT_IDLEGUI \|
283	RADEON_ISYNC_CPSCRATCH_IDLEGUI);	283	RADEON_ISYNC_CPSCRATCH_IDLEGUI);
284	radeon_ring_write(rdev, PACKET0(R300_GB_TILE_CONFIG, 0));	284	radeon_ring_write(rdev, PACKET0(R300_GB_TILE_CONFIG, 0));
285	radeon_ring_write(rdev, gb_tile_config);	285	radeon_ring_write(rdev, gb_tile_config);
286	radeon_ring_write(rdev, PACKET0(RADEON_WAIT_UNTIL, 0));	286	radeon_ring_write(rdev, PACKET0(RADEON_WAIT_UNTIL, 0));
287	radeon_ring_write(rdev,	287	radeon_ring_write(rdev,
288	RADEON_WAIT_2D_IDLECLEAN \|	288	RADEON_WAIT_2D_IDLECLEAN \|
289	RADEON_WAIT_3D_IDLECLEAN);	289	RADEON_WAIT_3D_IDLECLEAN);
290	radeon_ring_write(rdev, PACKET0(0x170C, 0));	290	radeon_ring_write(rdev, PACKET0(0x170C, 0));
291	radeon_ring_write(rdev, 1 << 31);	291	radeon_ring_write(rdev, 1 << 31);
292	radeon_ring_write(rdev, PACKET0(R300_GB_SELECT, 0));	292	radeon_ring_write(rdev, PACKET0(R300_GB_SELECT, 0));
293	radeon_ring_write(rdev, 0);	293	radeon_ring_write(rdev, 0);
294	radeon_ring_write(rdev, PACKET0(R300_GB_ENABLE, 0));	294	radeon_ring_write(rdev, PACKET0(R300_GB_ENABLE, 0));
295	radeon_ring_write(rdev, 0);	295	radeon_ring_write(rdev, 0);
296	radeon_ring_write(rdev, PACKET0(R300_RB3D_DSTCACHE_CTLSTAT, 0));	296	radeon_ring_write(rdev, PACKET0(R300_RB3D_DSTCACHE_CTLSTAT, 0));
297	radeon_ring_write(rdev, R300_RB3D_DC_FLUSH \| R300_RB3D_DC_FREE);	297	radeon_ring_write(rdev, R300_RB3D_DC_FLUSH \| R300_RB3D_DC_FREE);
298	radeon_ring_write(rdev, PACKET0(R300_RB3D_ZCACHE_CTLSTAT, 0));	298	radeon_ring_write(rdev, PACKET0(R300_RB3D_ZCACHE_CTLSTAT, 0));
299	radeon_ring_write(rdev, R300_ZC_FLUSH \| R300_ZC_FREE);	299	radeon_ring_write(rdev, R300_ZC_FLUSH \| R300_ZC_FREE);
300	radeon_ring_write(rdev, PACKET0(RADEON_WAIT_UNTIL, 0));	300	radeon_ring_write(rdev, PACKET0(RADEON_WAIT_UNTIL, 0));
301	radeon_ring_write(rdev,	301	radeon_ring_write(rdev,
302	RADEON_WAIT_2D_IDLECLEAN \|	302	RADEON_WAIT_2D_IDLECLEAN \|
303	RADEON_WAIT_3D_IDLECLEAN);	303	RADEON_WAIT_3D_IDLECLEAN);
304	radeon_ring_write(rdev, PACKET0(R300_GB_AA_CONFIG, 0));	304	radeon_ring_write(rdev, PACKET0(R300_GB_AA_CONFIG, 0));
305	radeon_ring_write(rdev, 0);	305	radeon_ring_write(rdev, 0);
306	radeon_ring_write(rdev, PACKET0(R300_RB3D_DSTCACHE_CTLSTAT, 0));	306	radeon_ring_write(rdev, PACKET0(R300_RB3D_DSTCACHE_CTLSTAT, 0));
307	radeon_ring_write(rdev, R300_RB3D_DC_FLUSH \| R300_RB3D_DC_FREE);	307	radeon_ring_write(rdev, R300_RB3D_DC_FLUSH \| R300_RB3D_DC_FREE);
308	radeon_ring_write(rdev, PACKET0(R300_RB3D_ZCACHE_CTLSTAT, 0));	308	radeon_ring_write(rdev, PACKET0(R300_RB3D_ZCACHE_CTLSTAT, 0));
309	radeon_ring_write(rdev, R300_ZC_FLUSH \| R300_ZC_FREE);	309	radeon_ring_write(rdev, R300_ZC_FLUSH \| R300_ZC_FREE);
310	radeon_ring_write(rdev, PACKET0(R300_GB_MSPOS0, 0));	310	radeon_ring_write(rdev, PACKET0(R300_GB_MSPOS0, 0));
311	radeon_ring_write(rdev,	311	radeon_ring_write(rdev,
312	((6 << R300_MS_X0_SHIFT) \|	312	((6 << R300_MS_X0_SHIFT) \|
313	(6 << R300_MS_Y0_SHIFT) \|	313	(6 << R300_MS_Y0_SHIFT) \|
314	(6 << R300_MS_X1_SHIFT) \|	314	(6 << R300_MS_X1_SHIFT) \|
315	(6 << R300_MS_Y1_SHIFT) \|	315	(6 << R300_MS_Y1_SHIFT) \|
316	(6 << R300_MS_X2_SHIFT) \|	316	(6 << R300_MS_X2_SHIFT) \|
317	(6 << R300_MS_Y2_SHIFT) \|	317	(6 << R300_MS_Y2_SHIFT) \|
318	(6 << R300_MSBD0_Y_SHIFT) \|	318	(6 << R300_MSBD0_Y_SHIFT) \|
319	(6 << R300_MSBD0_X_SHIFT)));	319	(6 << R300_MSBD0_X_SHIFT)));
320	radeon_ring_write(rdev, PACKET0(R300_GB_MSPOS1, 0));	320	radeon_ring_write(rdev, PACKET0(R300_GB_MSPOS1, 0));
321	radeon_ring_write(rdev,	321	radeon_ring_write(rdev,
322	((6 << R300_MS_X3_SHIFT) \|	322	((6 << R300_MS_X3_SHIFT) \|
323	(6 << R300_MS_Y3_SHIFT) \|	323	(6 << R300_MS_Y3_SHIFT) \|
324	(6 << R300_MS_X4_SHIFT) \|	324	(6 << R300_MS_X4_SHIFT) \|
325	(6 << R300_MS_Y4_SHIFT) \|	325	(6 << R300_MS_Y4_SHIFT) \|
326	(6 << R300_MS_X5_SHIFT) \|	326	(6 << R300_MS_X5_SHIFT) \|
327	(6 << R300_MS_Y5_SHIFT) \|	327	(6 << R300_MS_Y5_SHIFT) \|
328	(6 << R300_MSBD1_SHIFT)));	328	(6 << R300_MSBD1_SHIFT)));
329	radeon_ring_write(rdev, PACKET0(R300_GA_ENHANCE, 0));	329	radeon_ring_write(rdev, PACKET0(R300_GA_ENHANCE, 0));
330	radeon_ring_write(rdev, R300_GA_DEADLOCK_CNTL \| R300_GA_FASTSYNC_CNTL);	330	radeon_ring_write(rdev, R300_GA_DEADLOCK_CNTL \| R300_GA_FASTSYNC_CNTL);
331	radeon_ring_write(rdev, PACKET0(R300_GA_POLY_MODE, 0));	331	radeon_ring_write(rdev, PACKET0(R300_GA_POLY_MODE, 0));
332	radeon_ring_write(rdev,	332	radeon_ring_write(rdev,
333	R300_FRONT_PTYPE_TRIANGE \| R300_BACK_PTYPE_TRIANGE);	333	R300_FRONT_PTYPE_TRIANGE \| R300_BACK_PTYPE_TRIANGE);
334	radeon_ring_write(rdev, PACKET0(R300_GA_ROUND_MODE, 0));	334	radeon_ring_write(rdev, PACKET0(R300_GA_ROUND_MODE, 0));
335	radeon_ring_write(rdev,	335	radeon_ring_write(rdev,
336	R300_GEOMETRY_ROUND_NEAREST \|	336	R300_GEOMETRY_ROUND_NEAREST \|
337	R300_COLOR_ROUND_NEAREST);	337	R300_COLOR_ROUND_NEAREST);
338	radeon_ring_unlock_commit(rdev);	338	radeon_ring_unlock_commit(rdev);
339	}	339	}
340		340
341	void r300_errata(struct radeon_device *rdev)	341	void r300_errata(struct radeon_device *rdev)
342	{	342	{
343	rdev->pll_errata = 0;	343	rdev->pll_errata = 0;
344		344
345	if (rdev->family == CHIP_R300 &&	345	if (rdev->family == CHIP_R300 &&
346	(RREG32(RADEON_CONFIG_CNTL) & RADEON_CFG_ATI_REV_ID_MASK) == RADEON_CFG_ATI_REV_A11) {	346	(RREG32(RADEON_CONFIG_CNTL) & RADEON_CFG_ATI_REV_ID_MASK) == RADEON_CFG_ATI_REV_A11) {
347	rdev->pll_errata \|= CHIP_ERRATA_R300_CG;	347	rdev->pll_errata \|= CHIP_ERRATA_R300_CG;
348	}	348	}
349	}	349	}
350		350
351	int r300_mc_wait_for_idle(struct radeon_device *rdev)	351	int r300_mc_wait_for_idle(struct radeon_device *rdev)
352	{	352	{
353	unsigned i;	353	unsigned i;
354	uint32_t tmp;	354	uint32_t tmp;
355		355
356	for (i = 0; i < rdev->usec_timeout; i++) {	356	for (i = 0; i < rdev->usec_timeout; i++) {
357	/* read MC_STATUS */	357	/* read MC_STATUS */
358	tmp = RREG32(0x0150);	358	tmp = RREG32(0x0150);
359	if (tmp & (1 << 4)) {	359	if (tmp & (1 << 4)) {
360	return 0;	360	return 0;
361	}	361	}
362	DRM_UDELAY(1);	362	DRM_UDELAY(1);
363	}	363	}
364	return -1;	364	return -1;
365	}	365	}
366		366
367	void r300_gpu_init(struct radeon_device *rdev)	367	void r300_gpu_init(struct radeon_device *rdev)
368	{	368	{
369	uint32_t gb_tile_config, tmp;	369	uint32_t gb_tile_config, tmp;
370		370
371	r100_hdp_reset(rdev);	371	r100_hdp_reset(rdev);
372	/* FIXME: rv380 one pipes ? */	372	/* FIXME: rv380 one pipes ? */
373	if ((rdev->family == CHIP_R300) \|\| (rdev->family == CHIP_R350)) {	373	if ((rdev->family == CHIP_R300) \|\| (rdev->family == CHIP_R350)) {
374	/* r300,r350 */	374	/* r300,r350 */
375	rdev->num_gb_pipes = 2;	375	rdev->num_gb_pipes = 2;
376	} else {	376	} else {
377	/* rv350,rv370,rv380 */	377	/* rv350,rv370,rv380 */
378	rdev->num_gb_pipes = 1;	378	rdev->num_gb_pipes = 1;
379	}	379	}
380	rdev->num_z_pipes = 1;	380	rdev->num_z_pipes = 1;
381	gb_tile_config = (R300_ENABLE_TILING \| R300_TILE_SIZE_16);	381	gb_tile_config = (R300_ENABLE_TILING \| R300_TILE_SIZE_16);
382	switch (rdev->num_gb_pipes) {	382	switch (rdev->num_gb_pipes) {
383	case 2:	383	case 2:
384	gb_tile_config \|= R300_PIPE_COUNT_R300;	384	gb_tile_config \|= R300_PIPE_COUNT_R300;
385	break;	385	break;
386	case 3:	386	case 3:
387	gb_tile_config \|= R300_PIPE_COUNT_R420_3P;	387	gb_tile_config \|= R300_PIPE_COUNT_R420_3P;
388	break;	388	break;
389	case 4:	389	case 4:
390	gb_tile_config \|= R300_PIPE_COUNT_R420;	390	gb_tile_config \|= R300_PIPE_COUNT_R420;
391	break;	391	break;
392	default:	392	default:
393	case 1:	393	case 1:
394	gb_tile_config \|= R300_PIPE_COUNT_RV350;	394	gb_tile_config \|= R300_PIPE_COUNT_RV350;
395	break;	395	break;
396	}	396	}
397	WREG32(R300_GB_TILE_CONFIG, gb_tile_config);	397	WREG32(R300_GB_TILE_CONFIG, gb_tile_config);
398		398
399	if (r100_gui_wait_for_idle(rdev)) {	399	if (r100_gui_wait_for_idle(rdev)) {
400	printk(KERN_WARNING "Failed to wait GUI idle while "	400	printk(KERN_WARNING "Failed to wait GUI idle while "
401	"programming pipes. Bad things might happen.\n");	401	"programming pipes. Bad things might happen.\n");
402	}	402	}
403		403
404	tmp = RREG32(0x170C);	404	tmp = RREG32(0x170C);
405	WREG32(0x170C, tmp \| (1 << 31));	405	WREG32(0x170C, tmp \| (1 << 31));
406		406
407	WREG32(R300_RB2D_DSTCACHE_MODE,	407	WREG32(R300_RB2D_DSTCACHE_MODE,
408	R300_DC_AUTOFLUSH_ENABLE \|	408	R300_DC_AUTOFLUSH_ENABLE \|
409	R300_DC_DC_DISABLE_IGNORE_PE);	409	R300_DC_DC_DISABLE_IGNORE_PE);
410		410
411	if (r100_gui_wait_for_idle(rdev)) {	411	if (r100_gui_wait_for_idle(rdev)) {
412	printk(KERN_WARNING "Failed to wait GUI idle while "	412	printk(KERN_WARNING "Failed to wait GUI idle while "
413	"programming pipes. Bad things might happen.\n");	413	"programming pipes. Bad things might happen.\n");
414	}	414	}
415	if (r300_mc_wait_for_idle(rdev)) {	415	if (r300_mc_wait_for_idle(rdev)) {
416	printk(KERN_WARNING "Failed to wait MC idle while "	416	printk(KERN_WARNING "Failed to wait MC idle while "
417	"programming pipes. Bad things might happen.\n");	417	"programming pipes. Bad things might happen.\n");
418	}	418	}
419	DRM_INFO("radeon: %d quad pipes, %d Z pipes initialized.\n",	419	DRM_INFO("radeon: %d quad pipes, %d Z pipes initialized.\n",
420	rdev->num_gb_pipes, rdev->num_z_pipes);	420	rdev->num_gb_pipes, rdev->num_z_pipes);
421	}	421	}
422		422
423	int r300_ga_reset(struct radeon_device *rdev)	423	int r300_ga_reset(struct radeon_device *rdev)
424	{	424	{
425	uint32_t tmp;	425	uint32_t tmp;
426	bool reinit_cp;	426	bool reinit_cp;
427	int i;	427	int i;
428		428
429	reinit_cp = rdev->cp.ready;	429	reinit_cp = rdev->cp.ready;
430	rdev->cp.ready = false;	430	rdev->cp.ready = false;
431	for (i = 0; i < rdev->usec_timeout; i++) {	431	for (i = 0; i < rdev->usec_timeout; i++) {
432	WREG32(RADEON_CP_CSQ_MODE, 0);	432	WREG32(RADEON_CP_CSQ_MODE, 0);
433	WREG32(RADEON_CP_CSQ_CNTL, 0);	433	WREG32(RADEON_CP_CSQ_CNTL, 0);
434	WREG32(RADEON_RBBM_SOFT_RESET, 0x32005);	434	WREG32(RADEON_RBBM_SOFT_RESET, 0x32005);
435	(void)RREG32(RADEON_RBBM_SOFT_RESET);	435	(void)RREG32(RADEON_RBBM_SOFT_RESET);
436	udelay(200);	436	udelay(200);
437	WREG32(RADEON_RBBM_SOFT_RESET, 0);	437	WREG32(RADEON_RBBM_SOFT_RESET, 0);
438	/* Wait to prevent race in RBBM_STATUS */	438	/* Wait to prevent race in RBBM_STATUS */
439	mdelay(1);	439	mdelay(1);
440	tmp = RREG32(RADEON_RBBM_STATUS);	440	tmp = RREG32(RADEON_RBBM_STATUS);
441	if (tmp & ((1 << 20) \| (1 << 26))) {	441	if (tmp & ((1 << 20) \| (1 << 26))) {
442	DRM_ERROR("VAP & CP still busy (RBBM_STATUS=0x%08X)", tmp);	442	DRM_ERROR("VAP & CP still busy (RBBM_STATUS=0x%08X)", tmp);
443	/* GA still busy soft reset it */	443	/* GA still busy soft reset it */
444	WREG32(0x429C, 0x200);	444	WREG32(0x429C, 0x200);
445	WREG32(R300_VAP_PVS_STATE_FLUSH_REG, 0);	445	WREG32(R300_VAP_PVS_STATE_FLUSH_REG, 0);
446	WREG32(0x43E0, 0);	446	WREG32(0x43E0, 0);
447	WREG32(0x43E4, 0);	447	WREG32(0x43E4, 0);
448	WREG32(0x24AC, 0);	448	WREG32(0x24AC, 0);
449	}	449	}
450	/* Wait to prevent race in RBBM_STATUS */	450	/* Wait to prevent race in RBBM_STATUS */
451	mdelay(1);	451	mdelay(1);
452	tmp = RREG32(RADEON_RBBM_STATUS);	452	tmp = RREG32(RADEON_RBBM_STATUS);
453	if (!(tmp & ((1 << 20) \| (1 << 26)))) {	453	if (!(tmp & ((1 << 20) \| (1 << 26)))) {
454	break;	454	break;
455	}	455	}
456	}	456	}
457	for (i = 0; i < rdev->usec_timeout; i++) {	457	for (i = 0; i < rdev->usec_timeout; i++) {
458	tmp = RREG32(RADEON_RBBM_STATUS);	458	tmp = RREG32(RADEON_RBBM_STATUS);
459	if (!(tmp & ((1 << 20) \| (1 << 26)))) {	459	if (!(tmp & ((1 << 20) \| (1 << 26)))) {
460	DRM_INFO("GA reset succeed (RBBM_STATUS=0x%08X)\n",	460	DRM_INFO("GA reset succeed (RBBM_STATUS=0x%08X)\n",
461	tmp);	461	tmp);
462	if (reinit_cp) {	462	if (reinit_cp) {
463	return r100_cp_init(rdev, rdev->cp.ring_size);	463	return r100_cp_init(rdev, rdev->cp.ring_size);
464	}	464	}
465	return 0;	465	return 0;
466	}	466	}
467	DRM_UDELAY(1);	467	DRM_UDELAY(1);
468	}	468	}
469	tmp = RREG32(RADEON_RBBM_STATUS);	469	tmp = RREG32(RADEON_RBBM_STATUS);
470	DRM_ERROR("Failed to reset GA ! (RBBM_STATUS=0x%08X)\n", tmp);	470	DRM_ERROR("Failed to reset GA ! (RBBM_STATUS=0x%08X)\n", tmp);
471	return -1;	471	return -1;
472	}	472	}
473		473
474	int r300_gpu_reset(struct radeon_device *rdev)	474	int r300_gpu_reset(struct radeon_device *rdev)
475	{	475	{
476	uint32_t status;	476	uint32_t status;
477		477
478	/* reset order likely matter */	478	/* reset order likely matter */
479	status = RREG32(RADEON_RBBM_STATUS);	479	status = RREG32(RADEON_RBBM_STATUS);
480	/* reset HDP */	480	/* reset HDP */
481	r100_hdp_reset(rdev);	481	r100_hdp_reset(rdev);
482	/* reset rb2d */	482	/* reset rb2d */
483	if (status & ((1 << 17) \| (1 << 18) \| (1 << 27))) {	483	if (status & ((1 << 17) \| (1 << 18) \| (1 << 27))) {
484	r100_rb2d_reset(rdev);	484	r100_rb2d_reset(rdev);
485	}	485	}
486	/* reset GA */	486	/* reset GA */
487	if (status & ((1 << 20) \| (1 << 26))) {	487	if (status & ((1 << 20) \| (1 << 26))) {
488	r300_ga_reset(rdev);	488	r300_ga_reset(rdev);
489	}	489	}
490	/* reset CP */	490	/* reset CP */
491	status = RREG32(RADEON_RBBM_STATUS);	491	status = RREG32(RADEON_RBBM_STATUS);
492	if (status & (1 << 16)) {	492	if (status & (1 << 16)) {
493	r100_cp_reset(rdev);	493	r100_cp_reset(rdev);
494	}	494	}
495	/* Check if GPU is idle */	495	/* Check if GPU is idle */
496	status = RREG32(RADEON_RBBM_STATUS);	496	status = RREG32(RADEON_RBBM_STATUS);
497	if (status & (1 << 31)) {	497	if (status & (1 << 31)) {
498	DRM_ERROR("Failed to reset GPU (RBBM_STATUS=0x%08X)\n", status);	498	DRM_ERROR("Failed to reset GPU (RBBM_STATUS=0x%08X)\n", status);
499	return -1;	499	return -1;
500	}	500	}

Subversion Repositories Kolibri OS

(root)/drivers/video/drm/radeon/r300.c – Rev 1412 → 1414