xref: /freebsd-11-stable/sys/dev/drm/mach64_dma.c

/* mach64_dma.c -- DMA support for mach64 (Rage Pro) driver -*- linux-c -*- */
/**
 * \file mach64_dma.c
 * DMA support for mach64 (Rage Pro) driver
 *
 * \author Gareth Hughes <gareth@valinux.com>
 * \author Frank C. Earl <fearl@airmail.net>
 * \author Leif Delgass <ldelgass@retinalburn.net>
 * \author Jos�� Fonseca <j_r_fonseca@yahoo.co.uk>
 */

/*-
 * Copyright 2000 Gareth Hughes
 * Copyright 2002 Frank C. Earl
 * Copyright 2002-2003 Leif Delgass
 * All Rights Reserved.
 *
 * 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 (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 NONINFRINGEMENT.  IN NO EVENT SHALL
 * THE COPYRIGHT OWNER(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.
 */

#include <sys/cdefs.h>
__FBSDID("$FreeBSD: head/sys/dev/drm/mach64_dma.c 182080 2008-08-23 20:59:12Z rnoland $");

#include "dev/drm/drmP.h"
#include "dev/drm/drm.h"
#include "dev/drm/mach64_drm.h"
#include "dev/drm/mach64_drv.h"

/*******************************************************************/
/** \name Engine, FIFO control */
/*@{*/

/**
 * Waits for free entries in the FIFO.
 *
 * \note Most writes to Mach64 registers are automatically routed through
 * command FIFO which is 16 entry deep. Prior to writing to any draw engine
 * register one has to ensure that enough FIFO entries are available by calling
 * this function.  Failure to do so may cause the engine to lock.
 *
 * \param dev_priv pointer to device private data structure.
 * \param entries number of free entries in the FIFO to wait for.
 *
 * \returns zero on success, or -EBUSY if the timeout (specificed by
 * drm_mach64_private::usec_timeout) occurs.
 */
int mach64_do_wait_for_fifo(drm_mach64_private_t *dev_priv, int entries)
{
	int slots = 0, i;

	for (i = 0; i < dev_priv->usec_timeout; i++) {
		slots = (MACH64_READ(MACH64_FIFO_STAT) & MACH64_FIFO_SLOT_MASK);
		if (slots <= (0x8000 >> entries))
			return 0;
		DRM_UDELAY(1);
	}

	DRM_INFO("failed! slots=%d entries=%d\n", slots, entries);
	return -EBUSY;
}

/**
 * Wait for the draw engine to be idle.
 */
int mach64_do_wait_for_idle(drm_mach64_private_t *dev_priv)
{
	int i, ret;

	ret = mach64_do_wait_for_fifo(dev_priv, 16);
	if (ret < 0)
		return ret;

	for (i = 0; i < dev_priv->usec_timeout; i++) {
		if (!(MACH64_READ(MACH64_GUI_STAT) & MACH64_GUI_ACTIVE))
			return 0;
		DRM_UDELAY(1);
	}

	DRM_INFO("failed! GUI_STAT=0x%08x\n", MACH64_READ(MACH64_GUI_STAT));
	mach64_dump_ring_info(dev_priv);
	return -EBUSY;
}

/**
 * Wait for free entries in the ring buffer.
 *
 * The Mach64 bus master can be configured to act as a virtual FIFO, using a
 * circular buffer (commonly referred as "ring buffer" in other drivers) with
 * pointers to engine commands. This allows the CPU to do other things while
 * the graphics engine is busy, i.e., DMA mode.
 *
 * This function should be called before writing new entries to the ring
 * buffer.
 *
 * \param dev_priv pointer to device private data structure.
 * \param n number of free entries in the ring buffer to wait for.
 *
 * \returns zero on success, or -EBUSY if the timeout (specificed by
 * drm_mach64_private_t::usec_timeout) occurs.
 *
 * \sa mach64_dump_ring_info()
 */
int mach64_wait_ring(drm_mach64_private_t *dev_priv, int n)
{
	drm_mach64_descriptor_ring_t *ring = &dev_priv->ring;
	int i;

	for (i = 0; i < dev_priv->usec_timeout; i++) {
		mach64_update_ring_snapshot(dev_priv);
		if (ring->space >= n) {
			if (i > 0)
				DRM_DEBUG("%d usecs\n", i);
			return 0;
		}
		DRM_UDELAY(1);
	}

	/* FIXME: This is being ignored... */
	DRM_ERROR("failed!\n");
	mach64_dump_ring_info(dev_priv);
	return -EBUSY;
}

/**
 * Wait until all DMA requests have been processed...
 *
 * \sa mach64_wait_ring()
 */
static int mach64_ring_idle(drm_mach64_private_t *dev_priv)
{
	drm_mach64_descriptor_ring_t *ring = &dev_priv->ring;
	u32 head;
	int i;

	head = ring->head;
	i = 0;
	while (i < dev_priv->usec_timeout) {
		mach64_update_ring_snapshot(dev_priv);
		if (ring->head == ring->tail &&
		    !(MACH64_READ(MACH64_GUI_STAT) & MACH64_GUI_ACTIVE)) {
			if (i > 0)
				DRM_DEBUG("%d usecs\n", i);
			return 0;
		}
		if (ring->head == head) {
			++i;
		} else {
			head = ring->head;
			i = 0;
		}
		DRM_UDELAY(1);
	}

	DRM_INFO("failed! GUI_STAT=0x%08x\n", MACH64_READ(MACH64_GUI_STAT));
	mach64_dump_ring_info(dev_priv);
	return -EBUSY;
}

/**
 * Reset the the ring buffer descriptors.
 *
 * \sa mach64_do_engine_reset()
 */
static void mach64_ring_reset(drm_mach64_private_t *dev_priv)
{
	drm_mach64_descriptor_ring_t *ring = &dev_priv->ring;

	mach64_do_release_used_buffers(dev_priv);
	ring->head_addr = ring->start_addr;
	ring->head = ring->tail = 0;
	ring->space = ring->size;

	MACH64_WRITE(MACH64_BM_GUI_TABLE_CMD,
		     ring->head_addr | MACH64_CIRCULAR_BUF_SIZE_16KB);

	dev_priv->ring_running = 0;
}

/**
 * Ensure the all the queued commands will be processed.
 */
int mach64_do_dma_flush(drm_mach64_private_t *dev_priv)
{
	/* FIXME: It's not necessary to wait for idle when flushing
	 * we just need to ensure the ring will be completely processed
	 * in finite time without another ioctl
	 */
	return mach64_ring_idle(dev_priv);
}

/**
 * Stop all DMA activity.
 */
int mach64_do_dma_idle(drm_mach64_private_t *dev_priv)
{
	int ret;

	/* wait for completion */
	if ((ret = mach64_ring_idle(dev_priv)) < 0) {
		DRM_ERROR("failed BM_GUI_TABLE=0x%08x tail: %u\n",
			  MACH64_READ(MACH64_BM_GUI_TABLE),
			  dev_priv->ring.tail);
		return ret;
	}

	mach64_ring_stop(dev_priv);

	/* clean up after pass */
	mach64_do_release_used_buffers(dev_priv);
	return 0;
}

/**
 * Reset the engine.  This will stop the DMA if it is running.
 */
int mach64_do_engine_reset(drm_mach64_private_t *dev_priv)
{
	u32 tmp;

	DRM_DEBUG("\n");

	/* Kill off any outstanding DMA transfers.
	 */
	tmp = MACH64_READ(MACH64_BUS_CNTL);
	MACH64_WRITE(MACH64_BUS_CNTL, tmp | MACH64_BUS_MASTER_DIS);

	/* Reset the GUI engine (high to low transition).
	 */
	tmp = MACH64_READ(MACH64_GEN_TEST_CNTL);
	MACH64_WRITE(MACH64_GEN_TEST_CNTL, tmp & ~MACH64_GUI_ENGINE_ENABLE);
	/* Enable the GUI engine
	 */
	tmp = MACH64_READ(MACH64_GEN_TEST_CNTL);
	MACH64_WRITE(MACH64_GEN_TEST_CNTL, tmp | MACH64_GUI_ENGINE_ENABLE);

	/* ensure engine is not locked up by clearing any FIFO or HOST errors
	 */
	tmp = MACH64_READ(MACH64_BUS_CNTL);
	MACH64_WRITE(MACH64_BUS_CNTL, tmp | 0x00a00000);

	/* Once GUI engine is restored, disable bus mastering */
	MACH64_WRITE(MACH64_SRC_CNTL, 0);

	/* Reset descriptor ring */
	mach64_ring_reset(dev_priv);

	return 0;
}

/*@}*/


/*******************************************************************/
/** \name Debugging output */
/*@{*/

/**
 * Dump engine registers values.
 */
void mach64_dump_engine_info(drm_mach64_private_t *dev_priv)
{
	DRM_INFO("\n");
	if (!dev_priv->is_pci) {
		DRM_INFO("           AGP_BASE = 0x%08x\n",
			 MACH64_READ(MACH64_AGP_BASE));
		DRM_INFO("           AGP_CNTL = 0x%08x\n",
			 MACH64_READ(MACH64_AGP_CNTL));
	}
	DRM_INFO("     ALPHA_TST_CNTL = 0x%08x\n",
		 MACH64_READ(MACH64_ALPHA_TST_CNTL));
	DRM_INFO("\n");
	DRM_INFO("         BM_COMMAND = 0x%08x\n",
		 MACH64_READ(MACH64_BM_COMMAND));
	DRM_INFO("BM_FRAME_BUF_OFFSET = 0x%08x\n",
		 MACH64_READ(MACH64_BM_FRAME_BUF_OFFSET));
	DRM_INFO("       BM_GUI_TABLE = 0x%08x\n",
		 MACH64_READ(MACH64_BM_GUI_TABLE));
	DRM_INFO("          BM_STATUS = 0x%08x\n",
		 MACH64_READ(MACH64_BM_STATUS));
	DRM_INFO(" BM_SYSTEM_MEM_ADDR = 0x%08x\n",
		 MACH64_READ(MACH64_BM_SYSTEM_MEM_ADDR));
	DRM_INFO("    BM_SYSTEM_TABLE = 0x%08x\n",
		 MACH64_READ(MACH64_BM_SYSTEM_TABLE));
	DRM_INFO("           BUS_CNTL = 0x%08x\n",
		 MACH64_READ(MACH64_BUS_CNTL));
	DRM_INFO("\n");
	/* DRM_INFO( "         CLOCK_CNTL = 0x%08x\n", MACH64_READ( MACH64_CLOCK_CNTL ) ); */
	DRM_INFO("        CLR_CMP_CLR = 0x%08x\n",
		 MACH64_READ(MACH64_CLR_CMP_CLR));
	DRM_INFO("       CLR_CMP_CNTL = 0x%08x\n",
		 MACH64_READ(MACH64_CLR_CMP_CNTL));
	/* DRM_INFO( "        CLR_CMP_MSK = 0x%08x\n", MACH64_READ( MACH64_CLR_CMP_MSK ) ); */
	DRM_INFO("     CONFIG_CHIP_ID = 0x%08x\n",
		 MACH64_READ(MACH64_CONFIG_CHIP_ID));
	DRM_INFO("        CONFIG_CNTL = 0x%08x\n",
		 MACH64_READ(MACH64_CONFIG_CNTL));
	DRM_INFO("       CONFIG_STAT0 = 0x%08x\n",
		 MACH64_READ(MACH64_CONFIG_STAT0));
	DRM_INFO("       CONFIG_STAT1 = 0x%08x\n",
		 MACH64_READ(MACH64_CONFIG_STAT1));
	DRM_INFO("       CONFIG_STAT2 = 0x%08x\n",
		 MACH64_READ(MACH64_CONFIG_STAT2));
	DRM_INFO("            CRC_SIG = 0x%08x\n", MACH64_READ(MACH64_CRC_SIG));
	DRM_INFO("  CUSTOM_MACRO_CNTL = 0x%08x\n",
		 MACH64_READ(MACH64_CUSTOM_MACRO_CNTL));
	DRM_INFO("\n");
	/* DRM_INFO( "           DAC_CNTL = 0x%08x\n", MACH64_READ( MACH64_DAC_CNTL ) ); */
	/* DRM_INFO( "           DAC_REGS = 0x%08x\n", MACH64_READ( MACH64_DAC_REGS ) ); */
	DRM_INFO("        DP_BKGD_CLR = 0x%08x\n",
		 MACH64_READ(MACH64_DP_BKGD_CLR));
	DRM_INFO("        DP_FRGD_CLR = 0x%08x\n",
		 MACH64_READ(MACH64_DP_FRGD_CLR));
	DRM_INFO("             DP_MIX = 0x%08x\n", MACH64_READ(MACH64_DP_MIX));
	DRM_INFO("       DP_PIX_WIDTH = 0x%08x\n",
		 MACH64_READ(MACH64_DP_PIX_WIDTH));
	DRM_INFO("             DP_SRC = 0x%08x\n", MACH64_READ(MACH64_DP_SRC));
	DRM_INFO("      DP_WRITE_MASK = 0x%08x\n",
		 MACH64_READ(MACH64_DP_WRITE_MASK));
	DRM_INFO("         DSP_CONFIG = 0x%08x\n",
		 MACH64_READ(MACH64_DSP_CONFIG));
	DRM_INFO("         DSP_ON_OFF = 0x%08x\n",
		 MACH64_READ(MACH64_DSP_ON_OFF));
	DRM_INFO("           DST_CNTL = 0x%08x\n",
		 MACH64_READ(MACH64_DST_CNTL));
	DRM_INFO("      DST_OFF_PITCH = 0x%08x\n",
		 MACH64_READ(MACH64_DST_OFF_PITCH));
	DRM_INFO("\n");
	/* DRM_INFO( "       EXT_DAC_REGS = 0x%08x\n", MACH64_READ( MACH64_EXT_DAC_REGS ) ); */
	DRM_INFO("       EXT_MEM_CNTL = 0x%08x\n",
		 MACH64_READ(MACH64_EXT_MEM_CNTL));
	DRM_INFO("\n");
	DRM_INFO("          FIFO_STAT = 0x%08x\n",
		 MACH64_READ(MACH64_FIFO_STAT));
	DRM_INFO("\n");
	DRM_INFO("      GEN_TEST_CNTL = 0x%08x\n",
		 MACH64_READ(MACH64_GEN_TEST_CNTL));
	/* DRM_INFO( "              GP_IO = 0x%08x\n", MACH64_READ( MACH64_GP_IO ) ); */
	DRM_INFO("   GUI_CMDFIFO_DATA = 0x%08x\n",
		 MACH64_READ(MACH64_GUI_CMDFIFO_DATA));
	DRM_INFO("  GUI_CMDFIFO_DEBUG = 0x%08x\n",
		 MACH64_READ(MACH64_GUI_CMDFIFO_DEBUG));
	DRM_INFO("           GUI_CNTL = 0x%08x\n",
		 MACH64_READ(MACH64_GUI_CNTL));
	DRM_INFO("           GUI_STAT = 0x%08x\n",
		 MACH64_READ(MACH64_GUI_STAT));
	DRM_INFO("      GUI_TRAJ_CNTL = 0x%08x\n",
		 MACH64_READ(MACH64_GUI_TRAJ_CNTL));
	DRM_INFO("\n");
	DRM_INFO("          HOST_CNTL = 0x%08x\n",
		 MACH64_READ(MACH64_HOST_CNTL));
	DRM_INFO("           HW_DEBUG = 0x%08x\n",
		 MACH64_READ(MACH64_HW_DEBUG));
	DRM_INFO("\n");
	DRM_INFO("    MEM_ADDR_CONFIG = 0x%08x\n",
		 MACH64_READ(MACH64_MEM_ADDR_CONFIG));
	DRM_INFO("       MEM_BUF_CNTL = 0x%08x\n",
		 MACH64_READ(MACH64_MEM_BUF_CNTL));
	DRM_INFO("\n");
	DRM_INFO("           PAT_REG0 = 0x%08x\n",
		 MACH64_READ(MACH64_PAT_REG0));
	DRM_INFO("           PAT_REG1 = 0x%08x\n",
		 MACH64_READ(MACH64_PAT_REG1));
	DRM_INFO("\n");
	DRM_INFO("            SC_LEFT = 0x%08x\n", MACH64_READ(MACH64_SC_LEFT));
	DRM_INFO("           SC_RIGHT = 0x%08x\n",
		 MACH64_READ(MACH64_SC_RIGHT));
	DRM_INFO("             SC_TOP = 0x%08x\n", MACH64_READ(MACH64_SC_TOP));
	DRM_INFO("          SC_BOTTOM = 0x%08x\n",
		 MACH64_READ(MACH64_SC_BOTTOM));
	DRM_INFO("\n");
	DRM_INFO("      SCALE_3D_CNTL = 0x%08x\n",
		 MACH64_READ(MACH64_SCALE_3D_CNTL));
	DRM_INFO("       SCRATCH_REG0 = 0x%08x\n",
		 MACH64_READ(MACH64_SCRATCH_REG0));
	DRM_INFO("       SCRATCH_REG1 = 0x%08x\n",
		 MACH64_READ(MACH64_SCRATCH_REG1));
	DRM_INFO("         SETUP_CNTL = 0x%08x\n",
		 MACH64_READ(MACH64_SETUP_CNTL));
	DRM_INFO("           SRC_CNTL = 0x%08x\n",
		 MACH64_READ(MACH64_SRC_CNTL));
	DRM_INFO("\n");
	DRM_INFO("           TEX_CNTL = 0x%08x\n",
		 MACH64_READ(MACH64_TEX_CNTL));
	DRM_INFO("     TEX_SIZE_PITCH = 0x%08x\n",
		 MACH64_READ(MACH64_TEX_SIZE_PITCH));
	DRM_INFO("       TIMER_CONFIG = 0x%08x\n",
		 MACH64_READ(MACH64_TIMER_CONFIG));
	DRM_INFO("\n");
	DRM_INFO("             Z_CNTL = 0x%08x\n", MACH64_READ(MACH64_Z_CNTL));
	DRM_INFO("        Z_OFF_PITCH = 0x%08x\n",
		 MACH64_READ(MACH64_Z_OFF_PITCH));
	DRM_INFO("\n");
}

#define MACH64_DUMP_CONTEXT	3

/**
 * Used by mach64_dump_ring_info() to dump the contents of the current buffer
 * pointed by the ring head.
 */
static void mach64_dump_buf_info(drm_mach64_private_t *dev_priv,
				 struct drm_buf *buf)
{
	u32 addr = GETBUFADDR(buf);
	u32 used = buf->used >> 2;
	u32 sys_addr = MACH64_READ(MACH64_BM_SYSTEM_MEM_ADDR);
	u32 *p = GETBUFPTR(buf);
	int skipped = 0;

	DRM_INFO("buffer contents:\n");

	while (used) {
		u32 reg, count;

		reg = le32_to_cpu(*p++);
		if (addr <= GETBUFADDR(buf) + MACH64_DUMP_CONTEXT * 4 ||
		    (addr >= sys_addr - MACH64_DUMP_CONTEXT * 4 &&
		     addr <= sys_addr + MACH64_DUMP_CONTEXT * 4) ||
		    addr >=
		    GETBUFADDR(buf) + buf->used - MACH64_DUMP_CONTEXT * 4) {
			DRM_INFO("%08x:  0x%08x\n", addr, reg);
		}
		addr += 4;
		used--;

		count = (reg >> 16) + 1;
		reg = reg & 0xffff;
		reg = MMSELECT(reg);
		while (count && used) {
			if (addr <= GETBUFADDR(buf) + MACH64_DUMP_CONTEXT * 4 ||
			    (addr >= sys_addr - MACH64_DUMP_CONTEXT * 4 &&
			     addr <= sys_addr + MACH64_DUMP_CONTEXT * 4) ||
			    addr >=
			    GETBUFADDR(buf) + buf->used -
			    MACH64_DUMP_CONTEXT * 4) {
				DRM_INFO("%08x:    0x%04x = 0x%08x\n", addr,
					 reg, le32_to_cpu(*p));
				skipped = 0;
			} else {
				if (!skipped) {
					DRM_INFO("  ...\n");
					skipped = 1;
				}
			}
			p++;
			addr += 4;
			used--;

			reg += 4;
			count--;
		}
	}

	DRM_INFO("\n");
}

/**
 * Dump the ring state and contents, including the contents of the buffer being
 * processed by the graphics engine.
 */
void mach64_dump_ring_info(drm_mach64_private_t *dev_priv)
{
	drm_mach64_descriptor_ring_t *ring = &dev_priv->ring;
	int i, skipped;

	DRM_INFO("\n");

	DRM_INFO("ring contents:\n");
	DRM_INFO("  head_addr: 0x%08x head: %u tail: %u\n\n",
		 ring->head_addr, ring->head, ring->tail);

	skipped = 0;
	for (i = 0; i < ring->size / sizeof(u32); i += 4) {
		if (i <= MACH64_DUMP_CONTEXT * 4 ||
		    i >= ring->size / sizeof(u32) - MACH64_DUMP_CONTEXT * 4 ||
		    (i >= ring->tail - MACH64_DUMP_CONTEXT * 4 &&
		     i <= ring->tail + MACH64_DUMP_CONTEXT * 4) ||
		    (i >= ring->head - MACH64_DUMP_CONTEXT * 4 &&
		     i <= ring->head + MACH64_DUMP_CONTEXT * 4)) {
			DRM_INFO("  0x%08x:  0x%08x 0x%08x 0x%08x 0x%08x%s%s\n",
				 (u32)(ring->start_addr + i * sizeof(u32)),
				 le32_to_cpu(((u32 *) ring->start)[i + 0]),
				 le32_to_cpu(((u32 *) ring->start)[i + 1]),
				 le32_to_cpu(((u32 *) ring->start)[i + 2]),
				 le32_to_cpu(((u32 *) ring->start)[i + 3]),
				 i == ring->head ? " (head)" : "",
				 i == ring->tail ? " (tail)" : "");
			skipped = 0;
		} else {
			if (!skipped) {
				DRM_INFO("  ...\n");
				skipped = 1;
			}
		}
	}

	DRM_INFO("\n");

	if (ring->head >= 0 && ring->head < ring->size / sizeof(u32)) {
		struct list_head *ptr;
		u32 addr = le32_to_cpu(((u32 *) ring->start)[ring->head + 1]);

		list_for_each(ptr, &dev_priv->pending) {
			drm_mach64_freelist_t *entry =
			    list_entry(ptr, drm_mach64_freelist_t, list);
			struct drm_buf *buf = entry->buf;

			u32 buf_addr = GETBUFADDR(buf);

			if (buf_addr <= addr && addr < buf_addr + buf->used)
				mach64_dump_buf_info(dev_priv, buf);
		}
	}

	DRM_INFO("\n");
	DRM_INFO("       BM_GUI_TABLE = 0x%08x\n",
		 MACH64_READ(MACH64_BM_GUI_TABLE));
	DRM_INFO("\n");
	DRM_INFO("BM_FRAME_BUF_OFFSET = 0x%08x\n",
		 MACH64_READ(MACH64_BM_FRAME_BUF_OFFSET));
	DRM_INFO(" BM_SYSTEM_MEM_ADDR = 0x%08x\n",
		 MACH64_READ(MACH64_BM_SYSTEM_MEM_ADDR));
	DRM_INFO("         BM_COMMAND = 0x%08x\n",
		 MACH64_READ(MACH64_BM_COMMAND));
	DRM_INFO("\n");
	DRM_INFO("          BM_STATUS = 0x%08x\n",
		 MACH64_READ(MACH64_BM_STATUS));
	DRM_INFO("           BUS_CNTL = 0x%08x\n",
		 MACH64_READ(MACH64_BUS_CNTL));
	DRM_INFO("          FIFO_STAT = 0x%08x\n",
		 MACH64_READ(MACH64_FIFO_STAT));
	DRM_INFO("           GUI_STAT = 0x%08x\n",
		 MACH64_READ(MACH64_GUI_STAT));
	DRM_INFO("           SRC_CNTL = 0x%08x\n",
		 MACH64_READ(MACH64_SRC_CNTL));
}

/*@}*/


/*******************************************************************/
/** \name DMA descriptor ring macros */
/*@{*/

/**
 * Add the end mark to the ring's new tail position.
 *
 * The bus master engine will keep processing the DMA buffers listed in the ring
 * until it finds this mark, making it stop.
 *
 * \sa mach64_clear_dma_eol
 */
static __inline__ void mach64_set_dma_eol(volatile u32 *addr)
{
#if defined(__i386__)
	int nr = 31;

	/* Taken from include/asm-i386/bitops.h linux header */
	__asm__ __volatile__("lock;" "btsl %1,%0":"=m"(*addr)
			     :"Ir"(nr));
#elif defined(__powerpc__)
	u32 old;
	u32 mask = cpu_to_le32(MACH64_DMA_EOL);

	/* Taken from the include/asm-ppc/bitops.h linux header */
	__asm__ __volatile__("\n\
1:	lwarx	%0,0,%3 \n\
	or	%0,%0,%2 \n\
	stwcx.	%0,0,%3 \n\
	bne-	1b":"=&r"(old), "=m"(*addr)
			     :"r"(mask), "r"(addr), "m"(*addr)
			     :"cc");
#elif defined(__alpha__)
	u32 temp;
	u32 mask = MACH64_DMA_EOL;

	/* Taken from the include/asm-alpha/bitops.h linux header */
	__asm__ __volatile__("1:	ldl_l %0,%3\n"
			     "	bis %0,%2,%0\n"
			     "	stl_c %0,%1\n"
			     "	beq %0,2f\n"
			     ".subsection 2\n"
			     "2:	br 1b\n"
			     ".previous":"=&r"(temp), "=m"(*addr)
			     :"Ir"(mask), "m"(*addr));
#else
	u32 mask = cpu_to_le32(MACH64_DMA_EOL);

	*addr |= mask;
#endif
}

/**
 * Remove the end mark from the ring's old tail position.
 *
 * It should be called after calling mach64_set_dma_eol to mark the ring's new
 * tail position.
 *
 * We update the end marks while the bus master engine is in operation. Since
 * the bus master engine may potentially be reading from the same position
 * that we write, we must change atomically to avoid having intermediary bad
 * data.
 */
static __inline__ void mach64_clear_dma_eol(volatile u32 *addr)
{
#if defined(__i386__)
	int nr = 31;

	/* Taken from include/asm-i386/bitops.h linux header */
	__asm__ __volatile__("lock;" "btrl %1,%0":"=m"(*addr)
			     :"Ir"(nr));
#elif defined(__powerpc__)
	u32 old;
	u32 mask = cpu_to_le32(MACH64_DMA_EOL);

	/* Taken from the include/asm-ppc/bitops.h linux header */
	__asm__ __volatile__("\n\
1:	lwarx	%0,0,%3 \n\
	andc	%0,%0,%2 \n\
	stwcx.	%0,0,%3 \n\
	bne-	1b":"=&r"(old), "=m"(*addr)
			     :"r"(mask), "r"(addr), "m"(*addr)
			     :"cc");
#elif defined(__alpha__)
	u32 temp;
	u32 mask = ~MACH64_DMA_EOL;

	/* Taken from the include/asm-alpha/bitops.h linux header */
	__asm__ __volatile__("1:	ldl_l %0,%3\n"
			     "	and %0,%2,%0\n"
			     "	stl_c %0,%1\n"
			     "	beq %0,2f\n"
			     ".subsection 2\n"
			     "2:	br 1b\n"
			     ".previous":"=&r"(temp), "=m"(*addr)
			     :"Ir"(mask), "m"(*addr));
#else
	u32 mask = cpu_to_le32(~MACH64_DMA_EOL);

	*addr &= mask;
#endif
}

#define RING_LOCALS							\
	int _ring_tail, _ring_write; unsigned int _ring_mask; volatile u32 *_ring

#define RING_WRITE_OFS  _ring_write

#define BEGIN_RING(n)							\
	do {								\
		if (MACH64_VERBOSE) {					\
			DRM_INFO( "BEGIN_RING( %d ) \n",		\
				  (n) );				\
		}							\
		if (dev_priv->ring.space <= (n) * sizeof(u32)) {	\
			int ret;					\
			if ((ret = mach64_wait_ring( dev_priv, (n) * sizeof(u32))) < 0 ) { \
				DRM_ERROR( "wait_ring failed, resetting engine\n"); \
				mach64_dump_engine_info( dev_priv );	\
				mach64_do_engine_reset( dev_priv );	\
				return ret;				\
			}						\
		}							\
		dev_priv->ring.space -= (n) * sizeof(u32);		\
		_ring = (u32 *) dev_priv->ring.start;			\
		_ring_tail = _ring_write = dev_priv->ring.tail;		\
		_ring_mask = dev_priv->ring.tail_mask;			\
	} while (0)

#define OUT_RING( x )						\
do {								\
	if (MACH64_VERBOSE) {					\
		DRM_INFO( "   OUT_RING( 0x%08x ) at 0x%x\n",	\
			   (unsigned int)(x), _ring_write );	\
	}							\
	_ring[_ring_write++] = cpu_to_le32( x );		\
	_ring_write &= _ring_mask;				\
} while (0)

#define ADVANCE_RING()							\
do {									\
	if (MACH64_VERBOSE) {						\
		DRM_INFO( "ADVANCE_RING() wr=0x%06x tail=0x%06x\n",	\
			  _ring_write, _ring_tail );			\
	}								\
	DRM_MEMORYBARRIER();						\
	mach64_clear_dma_eol( &_ring[(_ring_tail - 2) & _ring_mask] );	\
	DRM_MEMORYBARRIER();						\
	dev_priv->ring.tail = _ring_write;				\
	mach64_ring_tick( dev_priv, &(dev_priv)->ring );		\
} while (0)

/**
 * Queue a DMA buffer of registers writes into the ring buffer.
 */
int mach64_add_buf_to_ring(drm_mach64_private_t *dev_priv,
                           drm_mach64_freelist_t *entry)
{
	int bytes, pages, remainder;
	u32 address, page;
	int i;
	struct drm_buf *buf = entry->buf;
	RING_LOCALS;

	bytes = buf->used;
	address = GETBUFADDR( buf );
	pages = (bytes + MACH64_DMA_CHUNKSIZE - 1) / MACH64_DMA_CHUNKSIZE;

	BEGIN_RING( pages * 4 );

	for ( i = 0 ; i < pages-1 ; i++ ) {
		page = address + i * MACH64_DMA_CHUNKSIZE;
		OUT_RING( MACH64_APERTURE_OFFSET + MACH64_BM_ADDR );
		OUT_RING( page );
		OUT_RING( MACH64_DMA_CHUNKSIZE | MACH64_DMA_HOLD_OFFSET );
		OUT_RING( 0 );
	}

	/* generate the final descriptor for any remaining commands in this buffer */
	page = address + i * MACH64_DMA_CHUNKSIZE;
	remainder = bytes - i * MACH64_DMA_CHUNKSIZE;

	/* Save dword offset of last descriptor for this buffer.
	 * This is needed to check for completion of the buffer in freelist_get
	 */
	entry->ring_ofs = RING_WRITE_OFS;

	OUT_RING( MACH64_APERTURE_OFFSET + MACH64_BM_ADDR );
	OUT_RING( page );
	OUT_RING( remainder | MACH64_DMA_HOLD_OFFSET | MACH64_DMA_EOL );
	OUT_RING( 0 );

	ADVANCE_RING();

	return 0;
}

/**
 * Queue DMA buffer controlling host data tranfers (e.g., blit).
 *
 * Almost identical to mach64_add_buf_to_ring.
 */
int mach64_add_hostdata_buf_to_ring(drm_mach64_private_t *dev_priv,
                                    drm_mach64_freelist_t *entry)
{
	int bytes, pages, remainder;
	u32 address, page;
	int i;
	struct drm_buf *buf = entry->buf;
	RING_LOCALS;

	bytes = buf->used - MACH64_HOSTDATA_BLIT_OFFSET;
	pages = (bytes + MACH64_DMA_CHUNKSIZE - 1) / MACH64_DMA_CHUNKSIZE;
	address = GETBUFADDR( buf );

	BEGIN_RING( 4 + pages * 4 );

	OUT_RING( MACH64_APERTURE_OFFSET + MACH64_BM_ADDR );
	OUT_RING( address );
	OUT_RING( MACH64_HOSTDATA_BLIT_OFFSET | MACH64_DMA_HOLD_OFFSET );
	OUT_RING( 0 );
	address += MACH64_HOSTDATA_BLIT_OFFSET;

	for ( i = 0 ; i < pages-1 ; i++ ) {
		page = address + i * MACH64_DMA_CHUNKSIZE;
		OUT_RING( MACH64_APERTURE_OFFSET + MACH64_BM_HOSTDATA );
		OUT_RING( page );
		OUT_RING( MACH64_DMA_CHUNKSIZE | MACH64_DMA_HOLD_OFFSET );
		OUT_RING( 0 );
	}

	/* generate the final descriptor for any remaining commands in this buffer */
	page = address + i * MACH64_DMA_CHUNKSIZE;
	remainder = bytes - i * MACH64_DMA_CHUNKSIZE;

	/* Save dword offset of last descriptor for this buffer.
	 * This is needed to check for completion of the buffer in freelist_get
	 */
	entry->ring_ofs = RING_WRITE_OFS;

	OUT_RING( MACH64_APERTURE_OFFSET + MACH64_BM_HOSTDATA );
	OUT_RING( page );
	OUT_RING( remainder | MACH64_DMA_HOLD_OFFSET | MACH64_DMA_EOL );
	OUT_RING( 0 );

	ADVANCE_RING();

	return 0;
}

/*@}*/


/*******************************************************************/
/** \name DMA test and initialization */
/*@{*/

/**
 * Perform a simple DMA operation using the pattern registers to test whether
 * DMA works.
 *
 * \return zero if successful.
 *
 * \note This function was the testbed for many experiences regarding Mach64
 * DMA operation. It is left here since it so tricky to get DMA operating
 * properly in some architectures and hardware.
 */
static int mach64_bm_dma_test(struct drm_device * dev)
{
	drm_mach64_private_t *dev_priv = dev->dev_private;
	drm_dma_handle_t *cpu_addr_dmah;
	u32 data_addr;
	u32 *table, *data;
	u32 expected[2];
	u32 src_cntl, pat_reg0, pat_reg1;
	int i, count, failed;

	DRM_DEBUG("\n");

	table = (u32 *) dev_priv->ring.start;

	/* FIXME: get a dma buffer from the freelist here */
	DRM_DEBUG("Allocating data memory ...\n");
	cpu_addr_dmah =
	    drm_pci_alloc(dev, 0x1000, 0x1000, 0xfffffffful);
	if (!cpu_addr_dmah) {
		DRM_INFO("data-memory allocation failed!\n");
		return -ENOMEM;
	} else {
		data = (u32 *) cpu_addr_dmah->vaddr;
		data_addr = (u32) cpu_addr_dmah->busaddr;
	}

	/* Save the X server's value for SRC_CNTL and restore it
	 * in case our test fails.  This prevents the X server
	 * from disabling it's cache for this register
	 */
	src_cntl = MACH64_READ(MACH64_SRC_CNTL);
	pat_reg0 = MACH64_READ(MACH64_PAT_REG0);
	pat_reg1 = MACH64_READ(MACH64_PAT_REG1);

	mach64_do_wait_for_fifo(dev_priv, 3);

	MACH64_WRITE(MACH64_SRC_CNTL, 0);
	MACH64_WRITE(MACH64_PAT_REG0, 0x11111111);
	MACH64_WRITE(MACH64_PAT_REG1, 0x11111111);

	mach64_do_wait_for_idle(dev_priv);

	for (i = 0; i < 2; i++) {
		u32 reg;
		reg = MACH64_READ((MACH64_PAT_REG0 + i * 4));
		DRM_DEBUG("(Before DMA Transfer) reg %d = 0x%08x\n", i, reg);
		if (reg != 0x11111111) {
			DRM_INFO("Error initializing test registers\n");
			DRM_INFO("resetting engine ...\n");
			mach64_do_engine_reset(dev_priv);
			DRM_INFO("freeing data buffer memory.\n");
			drm_pci_free(dev, cpu_addr_dmah);
			return -EIO;
		}
	}

	/* fill up a buffer with sets of 2 consecutive writes starting with PAT_REG0 */
	count = 0;

	data[count++] = cpu_to_le32(DMAREG(MACH64_PAT_REG0) | (1 << 16));
	data[count++] = expected[0] = 0x22222222;
	data[count++] = expected[1] = 0xaaaaaaaa;

	while (count < 1020) {
		data[count++] =
		    cpu_to_le32(DMAREG(MACH64_PAT_REG0) | (1 << 16));
		data[count++] = 0x22222222;
		data[count++] = 0xaaaaaaaa;
	}
	data[count++] = cpu_to_le32(DMAREG(MACH64_SRC_CNTL) | (0 << 16));
	data[count++] = 0;

	DRM_DEBUG("Preparing table ...\n");
	table[MACH64_DMA_FRAME_BUF_OFFSET] = cpu_to_le32(MACH64_BM_ADDR +
							 MACH64_APERTURE_OFFSET);
	table[MACH64_DMA_SYS_MEM_ADDR] = cpu_to_le32(data_addr);
	table[MACH64_DMA_COMMAND] = cpu_to_le32(count * sizeof(u32)
						| MACH64_DMA_HOLD_OFFSET
						| MACH64_DMA_EOL);
	table[MACH64_DMA_RESERVED] = 0;

	DRM_DEBUG("table[0] = 0x%08x\n", table[0]);
	DRM_DEBUG("table[1] = 0x%08x\n", table[1]);
	DRM_DEBUG("table[2] = 0x%08x\n", table[2]);
	DRM_DEBUG("table[3] = 0x%08x\n", table[3]);

	for (i = 0; i < 6; i++) {
		DRM_DEBUG(" data[%d] = 0x%08x\n", i, data[i]);
	}
	DRM_DEBUG(" ...\n");
	for (i = count - 5; i < count; i++) {
		DRM_DEBUG(" data[%d] = 0x%08x\n", i, data[i]);
	}

	DRM_MEMORYBARRIER();

	DRM_DEBUG("waiting for idle...\n");
	if ((i = mach64_do_wait_for_idle(dev_priv))) {
		DRM_INFO("mach64_do_wait_for_idle failed (result=%d)\n", i);
		DRM_INFO("resetting engine ...\n");
		mach64_do_engine_reset(dev_priv);
		mach64_do_wait_for_fifo(dev_priv, 3);
		MACH64_WRITE(MACH64_SRC_CNTL, src_cntl);
		MACH64_WRITE(MACH64_PAT_REG0, pat_reg0);
		MACH64_WRITE(MACH64_PAT_REG1, pat_reg1);
		DRM_INFO("freeing data buffer memory.\n");
		drm_pci_free(dev, cpu_addr_dmah);
		return i;
	}
	DRM_DEBUG("waiting for idle...done\n");

	DRM_DEBUG("BUS_CNTL = 0x%08x\n", MACH64_READ(MACH64_BUS_CNTL));
	DRM_DEBUG("SRC_CNTL = 0x%08x\n", MACH64_READ(MACH64_SRC_CNTL));
	DRM_DEBUG("\n");
	DRM_DEBUG("data bus addr = 0x%08x\n", data_addr);
	DRM_DEBUG("table bus addr = 0x%08x\n", dev_priv->ring.start_addr);

	DRM_DEBUG("starting DMA transfer...\n");
	MACH64_WRITE(MACH64_BM_GUI_TABLE_CMD,
		     dev_priv->ring.start_addr | MACH64_CIRCULAR_BUF_SIZE_16KB);

	MACH64_WRITE(MACH64_SRC_CNTL,
		     MACH64_SRC_BM_ENABLE | MACH64_SRC_BM_SYNC |
		     MACH64_SRC_BM_OP_SYSTEM_TO_REG);

	/* Kick off the transfer */
	DRM_DEBUG("starting DMA transfer... done.\n");
	MACH64_WRITE(MACH64_DST_HEIGHT_WIDTH, 0);

	DRM_DEBUG("waiting for idle...\n");

	if ((i = mach64_do_wait_for_idle(dev_priv))) {
		/* engine locked up, dump register state and reset */
		DRM_INFO("mach64_do_wait_for_idle failed (result=%d)\n", i);
		mach64_dump_engine_info(dev_priv);
		DRM_INFO("resetting engine ...\n");
		mach64_do_engine_reset(dev_priv);
		mach64_do_wait_for_fifo(dev_priv, 3);
		MACH64_WRITE(MACH64_SRC_CNTL, src_cntl);
		MACH64_WRITE(MACH64_PAT_REG0, pat_reg0);
		MACH64_WRITE(MACH64_PAT_REG1, pat_reg1);
		DRM_INFO("freeing data buffer memory.\n");
		drm_pci_free(dev, cpu_addr_dmah);
		return i;
	}

	DRM_DEBUG("waiting for idle...done\n");

	/* restore SRC_CNTL */
	mach64_do_wait_for_fifo(dev_priv, 1);
	MACH64_WRITE(MACH64_SRC_CNTL, src_cntl);

	failed = 0;

	/* Check register values to see if the GUI master operation succeeded */
	for (i = 0; i < 2; i++) {
		u32 reg;
		reg = MACH64_READ((MACH64_PAT_REG0 + i * 4));
		DRM_DEBUG("(After DMA Transfer) reg %d = 0x%08x\n", i, reg);
		if (reg != expected[i]) {
			failed = -1;
		}
	}

	/* restore pattern registers */
	mach64_do_wait_for_fifo(dev_priv, 2);
	MACH64_WRITE(MACH64_PAT_REG0, pat_reg0);
	MACH64_WRITE(MACH64_PAT_REG1, pat_reg1);

	DRM_DEBUG("freeing data buffer memory.\n");
	drm_pci_free(dev, cpu_addr_dmah);
	DRM_DEBUG("returning ...\n");

	return failed;
}

/**
 * Called during the DMA initialization ioctl to initialize all the necessary
 * software and hardware state for DMA operation.
 */
static int mach64_do_dma_init(struct drm_device * dev, drm_mach64_init_t * init)
{
	drm_mach64_private_t *dev_priv;
	u32 tmp;
	int i, ret;

	DRM_DEBUG("\n");

	dev_priv = drm_alloc(sizeof(drm_mach64_private_t), DRM_MEM_DRIVER);
	if (dev_priv == NULL)
		return -ENOMEM;

	memset(dev_priv, 0, sizeof(drm_mach64_private_t));

	dev_priv->is_pci = init->is_pci;

	dev_priv->fb_bpp = init->fb_bpp;
	dev_priv->front_offset = init->front_offset;
	dev_priv->front_pitch = init->front_pitch;
	dev_priv->back_offset = init->back_offset;
	dev_priv->back_pitch = init->back_pitch;

	dev_priv->depth_bpp = init->depth_bpp;
	dev_priv->depth_offset = init->depth_offset;
	dev_priv->depth_pitch = init->depth_pitch;

	dev_priv->front_offset_pitch = (((dev_priv->front_pitch / 8) << 22) |
					(dev_priv->front_offset >> 3));
	dev_priv->back_offset_pitch = (((dev_priv->back_pitch / 8) << 22) |
				       (dev_priv->back_offset >> 3));
	dev_priv->depth_offset_pitch = (((dev_priv->depth_pitch / 8) << 22) |
					(dev_priv->depth_offset >> 3));

	dev_priv->usec_timeout = 1000000;

	/* Set up the freelist, placeholder list and pending list */
	INIT_LIST_HEAD(&dev_priv->free_list);
	INIT_LIST_HEAD(&dev_priv->placeholders);
	INIT_LIST_HEAD(&dev_priv->pending);

	dev_priv->sarea = drm_getsarea(dev);
	if (!dev_priv->sarea) {
		DRM_ERROR("can not find sarea!\n");
		dev->dev_private = (void *)dev_priv;
		mach64_do_cleanup_dma(dev);
		return -EINVAL;
	}
	dev_priv->fb = drm_core_findmap(dev, init->fb_offset);
	if (!dev_priv->fb) {
		DRM_ERROR("can not find frame buffer map!\n");
		dev->dev_private = (void *)dev_priv;
		mach64_do_cleanup_dma(dev);
		return -EINVAL;
	}
	dev_priv->mmio = drm_core_findmap(dev, init->mmio_offset);
	if (!dev_priv->mmio) {
		DRM_ERROR("can not find mmio map!\n");
		dev->dev_private = (void *)dev_priv;
		mach64_do_cleanup_dma(dev);
		return -EINVAL;
	}

	dev_priv->ring_map = drm_core_findmap(dev, init->ring_offset);
	if (!dev_priv->ring_map) {
		DRM_ERROR("can not find ring map!\n");
		dev->dev_private = (void *)dev_priv;
		mach64_do_cleanup_dma(dev);
		return -EINVAL;
	}

	dev_priv->sarea_priv = (drm_mach64_sarea_t *)
	    ((u8 *) dev_priv->sarea->handle + init->sarea_priv_offset);

	if (!dev_priv->is_pci) {
		drm_core_ioremap(dev_priv->ring_map, dev);
		if (!dev_priv->ring_map->handle) {
			DRM_ERROR("can not ioremap virtual address for"
				  " descriptor ring\n");
			dev->dev_private = (void *)dev_priv;
			mach64_do_cleanup_dma(dev);
			return -ENOMEM;
		}
		dev->agp_buffer_token = init->buffers_offset;
		dev->agp_buffer_map =
		    drm_core_findmap(dev, init->buffers_offset);
		if (!dev->agp_buffer_map) {
			DRM_ERROR("can not find dma buffer map!\n");
			dev->dev_private = (void *)dev_priv;
			mach64_do_cleanup_dma(dev);
			return -EINVAL;
		}
		/* there might be a nicer way to do this -
		   dev isn't passed all the way though the mach64 - DA */
		dev_priv->dev_buffers = dev->agp_buffer_map;

		drm_core_ioremap(dev->agp_buffer_map, dev);
		if (!dev->agp_buffer_map->handle) {
			DRM_ERROR("can not ioremap virtual address for"
				  " dma buffer\n");
			dev->dev_private = (void *)dev_priv;
			mach64_do_cleanup_dma(dev);
			return -ENOMEM;
		}
		dev_priv->agp_textures =
		    drm_core_findmap(dev, init->agp_textures_offset);
		if (!dev_priv->agp_textures) {
			DRM_ERROR("can not find agp texture region!\n");
			dev->dev_private = (void *)dev_priv;
			mach64_do_cleanup_dma(dev);
			return -EINVAL;
		}
	}

	dev->dev_private = (void *)dev_priv;

	dev_priv->driver_mode = init->dma_mode;

	/* changing the FIFO size from the default causes problems with DMA */
	tmp = MACH64_READ(MACH64_GUI_CNTL);
	if ((tmp & MACH64_CMDFIFO_SIZE_MASK) != MACH64_CMDFIFO_SIZE_128) {
		DRM_INFO("Setting FIFO size to 128 entries\n");
		/* FIFO must be empty to change the FIFO depth */
		if ((ret = mach64_do_wait_for_idle(dev_priv))) {
			DRM_ERROR
			    ("wait for idle failed before changing FIFO depth!\n");
			mach64_do_cleanup_dma(dev);
			return ret;
		}
		MACH64_WRITE(MACH64_GUI_CNTL, ((tmp & ~MACH64_CMDFIFO_SIZE_MASK)
					       | MACH64_CMDFIFO_SIZE_128));
		/* need to read GUI_STAT for proper sync according to docs */
		if ((ret = mach64_do_wait_for_idle(dev_priv))) {
			DRM_ERROR
			    ("wait for idle failed when changing FIFO depth!\n");
			mach64_do_cleanup_dma(dev);
			return ret;
		}
	}

	dev_priv->ring.size = 0x4000;	/* 16KB */
	dev_priv->ring.start = dev_priv->ring_map->handle;
	dev_priv->ring.start_addr = (u32) dev_priv->ring_map->offset;

	memset(dev_priv->ring.start, 0, dev_priv->ring.size);
	DRM_INFO("descriptor ring: cpu addr %p, bus addr: 0x%08x\n",
		 dev_priv->ring.start, dev_priv->ring.start_addr);

	ret = 0;
	if (dev_priv->driver_mode != MACH64_MODE_MMIO) {

		/* enable block 1 registers and bus mastering */
		MACH64_WRITE(MACH64_BUS_CNTL, ((MACH64_READ(MACH64_BUS_CNTL)
						| MACH64_BUS_EXT_REG_EN)
					       & ~MACH64_BUS_MASTER_DIS));

		/* try a DMA GUI-mastering pass and fall back to MMIO if it fails */
		DRM_DEBUG("Starting DMA test...\n");
		if ((ret = mach64_bm_dma_test(dev))) {
			dev_priv->driver_mode = MACH64_MODE_MMIO;
		}
	}

	switch (dev_priv->driver_mode) {
	case MACH64_MODE_MMIO:
		MACH64_WRITE(MACH64_BUS_CNTL, (MACH64_READ(MACH64_BUS_CNTL)
					       | MACH64_BUS_EXT_REG_EN
					       | MACH64_BUS_MASTER_DIS));
		if (init->dma_mode == MACH64_MODE_MMIO)
			DRM_INFO("Forcing pseudo-DMA mode\n");
		else
			DRM_INFO
			    ("DMA test failed (ret=%d), using pseudo-DMA mode\n",
			     ret);
		break;
	case MACH64_MODE_DMA_SYNC:
		DRM_INFO("DMA test succeeded, using synchronous DMA mode\n");
		break;
	case MACH64_MODE_DMA_ASYNC:
	default:
		DRM_INFO("DMA test succeeded, using asynchronous DMA mode\n");
	}

	dev_priv->ring_running = 0;

	/* setup offsets for physical address of table start and end */
	dev_priv->ring.head_addr = dev_priv->ring.start_addr;
	dev_priv->ring.head = dev_priv->ring.tail = 0;
	dev_priv->ring.tail_mask = (dev_priv->ring.size / sizeof(u32)) - 1;
	dev_priv->ring.space = dev_priv->ring.size;

	/* setup physical address and size of descriptor table */
	mach64_do_wait_for_fifo(dev_priv, 1);
	MACH64_WRITE(MACH64_BM_GUI_TABLE_CMD,
		     (dev_priv->ring.
		      head_addr | MACH64_CIRCULAR_BUF_SIZE_16KB));

	/* init frame counter */
	dev_priv->sarea_priv->frames_queued = 0;
	for (i = 0; i < MACH64_MAX_QUEUED_FRAMES; i++) {
		dev_priv->frame_ofs[i] = ~0;	/* All ones indicates placeholder */
	}

	/* Allocate the DMA buffer freelist */
	if ((ret = mach64_init_freelist(dev))) {
		DRM_ERROR("Freelist allocation failed\n");
		mach64_do_cleanup_dma(dev);
		return ret;
	}

	return 0;
}

/*******************************************************************/
/** MMIO Pseudo-DMA (intended primarily for debugging, not performance)
 */

int mach64_do_dispatch_pseudo_dma(drm_mach64_private_t *dev_priv)
{
	drm_mach64_descriptor_ring_t *ring = &dev_priv->ring;
	volatile u32 *ring_read;
	struct list_head *ptr;
	drm_mach64_freelist_t *entry;
	struct drm_buf *buf = NULL;
	u32 *buf_ptr;
	u32 used, reg, target;
	int fifo, count, found, ret, no_idle_wait;

	fifo = count = reg = no_idle_wait = 0;
	target = MACH64_BM_ADDR;

	if ((ret = mach64_do_wait_for_idle(dev_priv)) < 0) {
		DRM_INFO("idle failed before pseudo-dma dispatch, resetting engine\n");
		mach64_dump_engine_info(dev_priv);
		mach64_do_engine_reset(dev_priv);
		return ret;
	}

	ring_read = (u32 *) ring->start;

	while (ring->tail != ring->head) {
		u32 buf_addr, new_target, offset;
		u32 bytes, remaining, head, eol;

		head = ring->head;

		new_target =
		    le32_to_cpu(ring_read[head++]) - MACH64_APERTURE_OFFSET;
		buf_addr = le32_to_cpu(ring_read[head++]);
		eol = le32_to_cpu(ring_read[head]) & MACH64_DMA_EOL;
		bytes = le32_to_cpu(ring_read[head++])
		    & ~(MACH64_DMA_HOLD_OFFSET | MACH64_DMA_EOL);
		head++;
		head &= ring->tail_mask;

		/* can't wait for idle between a blit setup descriptor
		 * and a HOSTDATA descriptor or the engine will lock
		 */
		if (new_target == MACH64_BM_HOSTDATA
		    && target == MACH64_BM_ADDR)
			no_idle_wait = 1;

		target = new_target;

		found = 0;
		offset = 0;
		list_for_each(ptr, &dev_priv->pending) {
			entry = list_entry(ptr, drm_mach64_freelist_t, list);
			buf = entry->buf;
			offset = buf_addr - GETBUFADDR(buf);
			if (offset >= 0 && offset < MACH64_BUFFER_SIZE) {
				found = 1;
				break;
			}
		}

		if (!found || buf == NULL) {
			DRM_ERROR
			    ("Couldn't find pending buffer: head: %u tail: %u buf_addr: 0x%08x %s\n",
			     head, ring->tail, buf_addr, (eol ? "eol" : ""));
			mach64_dump_ring_info(dev_priv);
			mach64_do_engine_reset(dev_priv);
			return -EINVAL;
		}

		/* Hand feed the buffer to the card via MMIO, waiting for the fifo
		 * every 16 writes
		 */
		DRM_DEBUG("target: (0x%08x) %s\n", target,
			  (target ==
			   MACH64_BM_HOSTDATA ? "BM_HOSTDATA" : "BM_ADDR"));
		DRM_DEBUG("offset: %u bytes: %u used: %u\n", offset, bytes,
			  buf->used);

		remaining = (buf->used - offset) >> 2;	/* dwords remaining in buffer */
		used = bytes >> 2;	/* dwords in buffer for this descriptor */
		buf_ptr = (u32 *) ((char *)GETBUFPTR(buf) + offset);

		while (used) {

			if (count == 0) {
				if (target == MACH64_BM_HOSTDATA) {
					reg = DMAREG(MACH64_HOST_DATA0);
					count =
					    (remaining > 16) ? 16 : remaining;
					fifo = 0;
				} else {
					reg = le32_to_cpu(*buf_ptr++);
					used--;
					count = (reg >> 16) + 1;
				}

				reg = reg & 0xffff;
				reg = MMSELECT(reg);
			}
			while (count && used) {
				if (!fifo) {
					if (no_idle_wait) {
						if ((ret =
						     mach64_do_wait_for_fifo
						     (dev_priv, 16)) < 0) {
							no_idle_wait = 0;
							return ret;
						}
					} else {
						if ((ret =
						     mach64_do_wait_for_idle
						     (dev_priv)) < 0) {
							return ret;
						}
					}
					fifo = 16;
				}
				--fifo;
				MACH64_WRITE(reg, le32_to_cpu(*buf_ptr++));
				used--;
				remaining--;

				reg += 4;
				count--;
			}
		}
		ring->head = head;
		ring->head_addr = ring->start_addr + (ring->head * sizeof(u32));
		ring->space += (4 * sizeof(u32));
	}

	if ((ret = mach64_do_wait_for_idle(dev_priv)) < 0) {
		return ret;
	}
	MACH64_WRITE(MACH64_BM_GUI_TABLE_CMD,
		     ring->head_addr | MACH64_CIRCULAR_BUF_SIZE_16KB);

	DRM_DEBUG("completed\n");
	return 0;
}

/*@}*/


/*******************************************************************/
/** \name DMA cleanup */
/*@{*/

int mach64_do_cleanup_dma(struct drm_device * dev)
{
	DRM_DEBUG("\n");

	/* Make sure interrupts are disabled here because the uninstall ioctl
	 * may not have been called from userspace and after dev_private
	 * is freed, it's too late.
	 */
	if (dev->irq)
		drm_irq_uninstall(dev);

	if (dev->dev_private) {
		drm_mach64_private_t *dev_priv = dev->dev_private;

		if (!dev_priv->is_pci) {
			if (dev_priv->ring_map)
				drm_core_ioremapfree(dev_priv->ring_map, dev);

			if (dev->agp_buffer_map) {
				drm_core_ioremapfree(dev->agp_buffer_map, dev);
				dev->agp_buffer_map = NULL;
			}
		}

		mach64_destroy_freelist(dev);

		drm_free(dev_priv, sizeof(drm_mach64_private_t),
			 DRM_MEM_DRIVER);
		dev->dev_private = NULL;
	}

	return 0;
}

/*@}*/


/*******************************************************************/
/** \name IOCTL handlers */
/*@{*/

int mach64_dma_init(struct drm_device *dev, void *data,
		    struct drm_file *file_priv)
{
	drm_mach64_init_t *init = data;

	DRM_DEBUG("\n");

	LOCK_TEST_WITH_RETURN(dev, file_priv);

	switch (init->func) {
	case DRM_MACH64_INIT_DMA:
		return mach64_do_dma_init(dev, init);
	case DRM_MACH64_CLEANUP_DMA:
		return mach64_do_cleanup_dma(dev);
	}

	return -EINVAL;
}

int mach64_dma_idle(struct drm_device *dev, void *data,
		    struct drm_file *file_priv)
{
	drm_mach64_private_t *dev_priv = dev->dev_private;

	DRM_DEBUG("\n");

	LOCK_TEST_WITH_RETURN(dev, file_priv);

	return mach64_do_dma_idle(dev_priv);
}

int mach64_dma_flush(struct drm_device *dev, void *data,
		     struct drm_file *file_priv)
{
	drm_mach64_private_t *dev_priv = dev->dev_private;

	DRM_DEBUG("\n");

	LOCK_TEST_WITH_RETURN(dev, file_priv);

	return mach64_do_dma_flush(dev_priv);
}

int mach64_engine_reset(struct drm_device *dev, void *data,
			struct drm_file *file_priv)
{
	drm_mach64_private_t *dev_priv = dev->dev_private;

	DRM_DEBUG("\n");

	LOCK_TEST_WITH_RETURN(dev, file_priv);

	return mach64_do_engine_reset(dev_priv);
}

/*@}*/


/*******************************************************************/
/** \name Freelist management */
/*@{*/

int mach64_init_freelist(struct drm_device * dev)
{
	struct drm_device_dma *dma = dev->dma;
	drm_mach64_private_t *dev_priv = dev->dev_private;
	drm_mach64_freelist_t *entry;
	struct list_head *ptr;
	int i;

	DRM_DEBUG("adding %d buffers to freelist\n", dma->buf_count);

	for (i = 0; i < dma->buf_count; i++) {
		if ((entry =
		     (drm_mach64_freelist_t *)
		     drm_alloc(sizeof(drm_mach64_freelist_t),
			       DRM_MEM_BUFLISTS)) == NULL)
			return -ENOMEM;
		memset(entry, 0, sizeof(drm_mach64_freelist_t));
		entry->buf = dma->buflist[i];
		ptr = &entry->list;
		list_add_tail(ptr, &dev_priv->free_list);
	}

	return 0;
}

void mach64_destroy_freelist(struct drm_device * dev)
{
	drm_mach64_private_t *dev_priv = dev->dev_private;
	drm_mach64_freelist_t *entry;
	struct list_head *ptr;
	struct list_head *tmp;

	DRM_DEBUG("\n");

	list_for_each_safe(ptr, tmp, &dev_priv->pending) {
		list_del(ptr);
		entry = list_entry(ptr, drm_mach64_freelist_t, list);
		drm_free(entry, sizeof(*entry), DRM_MEM_BUFLISTS);
	}
	list_for_each_safe(ptr, tmp, &dev_priv->placeholders) {
		list_del(ptr);
		entry = list_entry(ptr, drm_mach64_freelist_t, list);
		drm_free(entry, sizeof(*entry), DRM_MEM_BUFLISTS);
	}

	list_for_each_safe(ptr, tmp, &dev_priv->free_list) {
		list_del(ptr);
		entry = list_entry(ptr, drm_mach64_freelist_t, list);
		drm_free(entry, sizeof(*entry), DRM_MEM_BUFLISTS);
	}
}

/* IMPORTANT: This function should only be called when the engine is idle or locked up,
 * as it assumes all buffers in the pending list have been completed by the hardware.
 */
int mach64_do_release_used_buffers(drm_mach64_private_t *dev_priv)
{
	struct list_head *ptr;
	struct list_head *tmp;
	drm_mach64_freelist_t *entry;
	int i;

	if (list_empty(&dev_priv->pending))
		return 0;

	/* Iterate the pending list and move all buffers into the freelist... */
	i = 0;
	list_for_each_safe(ptr, tmp, &dev_priv->pending) {
		entry = list_entry(ptr, drm_mach64_freelist_t, list);
		if (entry->discard) {
			entry->buf->pending = 0;
			list_del(ptr);
			list_add_tail(ptr, &dev_priv->free_list);
			i++;
		}
	}

	DRM_DEBUG("released %d buffers from pending list\n", i);

	return 0;
}

static int mach64_do_reclaim_completed(drm_mach64_private_t *dev_priv)
{
	drm_mach64_descriptor_ring_t *ring = &dev_priv->ring;
	struct list_head *ptr;
	struct list_head *tmp;
	drm_mach64_freelist_t *entry;
	u32 head, tail, ofs;

	mach64_ring_tick(dev_priv, ring);
	head = ring->head;
	tail = ring->tail;

	if (head == tail) {
#if MACH64_EXTRA_CHECKING
		if (MACH64_READ(MACH64_GUI_STAT) & MACH64_GUI_ACTIVE) {
			DRM_ERROR("Empty ring with non-idle engine!\n");
			mach64_dump_ring_info(dev_priv);
			return -1;
		}
#endif
		/* last pass is complete, so release everything */
		mach64_do_release_used_buffers(dev_priv);
		DRM_DEBUG("idle engine, freed all buffers.\n");
		if (list_empty(&dev_priv->free_list)) {
			DRM_ERROR("Freelist empty with idle engine\n");
			return -1;
		}
		return 0;
	}
	/* Look for a completed buffer and bail out of the loop
	 * as soon as we find one -- don't waste time trying
	 * to free extra bufs here, leave that to do_release_used_buffers
	 */
	list_for_each_safe(ptr, tmp, &dev_priv->pending) {
		entry = list_entry(ptr, drm_mach64_freelist_t, list);
		ofs = entry->ring_ofs;
		if (entry->discard &&
		    ((head < tail && (ofs < head || ofs >= tail)) ||
		     (head > tail && (ofs < head && ofs >= tail)))) {
#if MACH64_EXTRA_CHECKING
			int i;

			for (i = head; i != tail; i = (i + 4) & ring->tail_mask)
			{
				u32 o1 = le32_to_cpu(((u32 *) ring->
						 start)[i + 1]);
				u32 o2 = GETBUFADDR(entry->buf);

				if (o1 == o2) {
					DRM_ERROR
					    ("Attempting to free used buffer: "
					     "i=%d  buf=0x%08x\n",
					     i, o1);
					mach64_dump_ring_info(dev_priv);
					return -1;
				}
			}
#endif
			/* found a processed buffer */
			entry->buf->pending = 0;
			list_del(ptr);
			list_add_tail(ptr, &dev_priv->free_list);
			DRM_DEBUG
			    ("freed processed buffer (head=%d tail=%d "
			     "buf ring ofs=%d).\n",
			     head, tail, ofs);
			return 0;
		}
	}

	return 1;
}

struct drm_buf *mach64_freelist_get(drm_mach64_private_t *dev_priv)
{
	drm_mach64_descriptor_ring_t *ring = &dev_priv->ring;
	drm_mach64_freelist_t *entry;
	struct list_head *ptr;
	int t;

	if (list_empty(&dev_priv->free_list)) {
		if (list_empty(&dev_priv->pending)) {
			DRM_ERROR
			    ("Couldn't get buffer - pending and free lists empty\n");
			t = 0;
			list_for_each(ptr, &dev_priv->placeholders) {
				t++;
			}
			DRM_INFO("Placeholders: %d\n", t);
			return NULL;
		}

		for (t = 0; t < dev_priv->usec_timeout; t++) {
			int ret;

			ret = mach64_do_reclaim_completed(dev_priv);
			if (ret == 0)
				goto _freelist_entry_found;
			if (ret < 0)
				return NULL;

			DRM_UDELAY(1);
		}
		mach64_dump_ring_info(dev_priv);
		DRM_ERROR
		    ("timeout waiting for buffers: ring head_addr: 0x%08x head: %d tail: %d\n",
		     ring->head_addr, ring->head, ring->tail);
		return NULL;
	}

      _freelist_entry_found:
	ptr = dev_priv->free_list.next;
	list_del(ptr);
	entry = list_entry(ptr, drm_mach64_freelist_t, list);
	entry->buf->used = 0;
	list_add_tail(ptr, &dev_priv->placeholders);
	return entry->buf;
}

int mach64_freelist_put(drm_mach64_private_t *dev_priv, struct drm_buf *copy_buf)
{
	struct list_head *ptr;
	drm_mach64_freelist_t *entry;

#if MACH64_EXTRA_CHECKING
	list_for_each(ptr, &dev_priv->pending) {
		entry = list_entry(ptr, drm_mach64_freelist_t, list);
		if (copy_buf == entry->buf) {
			DRM_ERROR("Trying to release a pending buf\n");
			return -EFAULT;
		}
	}
#endif
	ptr = dev_priv->placeholders.next;
	entry = list_entry(ptr, drm_mach64_freelist_t, list);
	copy_buf->pending = 0;
	copy_buf->used = 0;
	entry->buf = copy_buf;
	entry->discard = 1;
	list_del(ptr);
	list_add_tail(ptr, &dev_priv->free_list);

	return 0;
}

/*@}*/


/*******************************************************************/
/** \name DMA buffer request and submission IOCTL handler */
/*@{*/

static int mach64_dma_get_buffers(struct drm_device *dev,
				  struct drm_file *file_priv,
				  struct drm_dma * d)
{
	int i;
	struct drm_buf *buf;
	drm_mach64_private_t *dev_priv = dev->dev_private;

	for (i = d->granted_count; i < d->request_count; i++) {
		buf = mach64_freelist_get(dev_priv);
#if MACH64_EXTRA_CHECKING
		if (!buf)
			return -EFAULT;
#else
		if (!buf)
			return -EAGAIN;
#endif

		buf->file_priv = file_priv;

		if (DRM_COPY_TO_USER(&d->request_indices[i], &buf->idx,
				     sizeof(buf->idx)))
			return -EFAULT;
		if (DRM_COPY_TO_USER(&d->request_sizes[i], &buf->total,
				     sizeof(buf->total)))
			return -EFAULT;

		d->granted_count++;
	}
	return 0;
}

int mach64_dma_buffers(struct drm_device *dev, void *data,
		       struct drm_file *file_priv)
{
	struct drm_device_dma *dma = dev->dma;
	struct drm_dma *d = data;
	int ret = 0;

	LOCK_TEST_WITH_RETURN(dev, file_priv);

	/* Please don't send us buffers.
	 */
	if (d->send_count != 0) {
		DRM_ERROR("Process %d trying to send %d buffers via drmDMA\n",
			  DRM_CURRENTPID, d->send_count);
		return -EINVAL;
	}

	/* We'll send you buffers.
	 */
	if (d->request_count < 0 || d->request_count > dma->buf_count) {
		DRM_ERROR("Process %d trying to get %d buffers (of %d max)\n",
			  DRM_CURRENTPID, d->request_count, dma->buf_count);
		ret = -EINVAL;
	}

	d->granted_count = 0;

	if (d->request_count) {
		ret = mach64_dma_get_buffers(dev, file_priv, d);
	}

	return ret;
}

void mach64_driver_lastclose(struct drm_device * dev)
{
	mach64_do_cleanup_dma(dev);
}

/*@}*/