gma500: introduce the framebuffer support code

We support 2D acceleration on some devices but we try and do tricks with
the GTT as a starting point as this is far faster. The GTT logic could be
improved further but for most display sizes it already makes a pretty good
decision.

Signed-off-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

drivers/gpu/drm/gma500/accel_2d.c

@@ -0,0 +1,362 @@
+/**************************************************************************
+ * Copyright (c) 2007-2011, Intel Corporation.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License along with
+ * this program; if not, write to the Free Software Foundation, Inc.,
+ * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+ *
+ * Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
+ * develop this driver.
+ *
+ **************************************************************************/
+
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/errno.h>
+#include <linux/string.h>
+#include <linux/mm.h>
+#include <linux/tty.h>
+#include <linux/slab.h>
+#include <linux/delay.h>
+#include <linux/fb.h>
+#include <linux/init.h>
+#include <linux/console.h>
+
+#include <drm/drmP.h>
+#include <drm/drm.h>
+#include <drm/drm_crtc.h>
+
+#include "psb_drv.h"
+#include "psb_reg.h"
+#include "framebuffer.h"
+
+/**
+ *	psb_spank		-	reset the 2D engine
+ *	@dev_priv: our PSB DRM device
+ *
+ *	Soft reset the graphics engine and then reload the necessary registers.
+ *	We use this at initialisation time but it will become relevant for
+ *	accelerated X later
+ */
+void psb_spank(struct drm_psb_private *dev_priv)
+{
+	PSB_WSGX32(_PSB_CS_RESET_BIF_RESET | _PSB_CS_RESET_DPM_RESET |
+		_PSB_CS_RESET_TA_RESET | _PSB_CS_RESET_USE_RESET |
+		_PSB_CS_RESET_ISP_RESET | _PSB_CS_RESET_TSP_RESET |
+		_PSB_CS_RESET_TWOD_RESET, PSB_CR_SOFT_RESET);
+	PSB_RSGX32(PSB_CR_SOFT_RESET);
+
+	msleep(1);
+
+	PSB_WSGX32(0, PSB_CR_SOFT_RESET);
+	wmb();
+	PSB_WSGX32(PSB_RSGX32(PSB_CR_BIF_CTRL) | _PSB_CB_CTRL_CLEAR_FAULT,
+		   PSB_CR_BIF_CTRL);
+	wmb();
+	(void) PSB_RSGX32(PSB_CR_BIF_CTRL);
+
+	msleep(1);
+	PSB_WSGX32(PSB_RSGX32(PSB_CR_BIF_CTRL) & ~_PSB_CB_CTRL_CLEAR_FAULT,
+		   PSB_CR_BIF_CTRL);
+	(void) PSB_RSGX32(PSB_CR_BIF_CTRL);
+	PSB_WSGX32(dev_priv->gtt.gatt_start, PSB_CR_BIF_TWOD_REQ_BASE);
+}
+
+/**
+ *	psb2_2d_wait_available	-	wait for FIFO room
+ *	@dev_priv: our DRM device
+ *	@size: size (in dwords) of the command we want to issue
+ *
+ *	Wait until there is room to load the FIFO with our data. If the
+ *	device is not responding then reset it
+ */
+static int psb_2d_wait_available(struct drm_psb_private *dev_priv,
+			  unsigned size)
+{
+	uint32_t avail = PSB_RSGX32(PSB_CR_2D_SOCIF);
+	unsigned long t = jiffies + HZ;
+
+	while (avail < size) {
+		avail = PSB_RSGX32(PSB_CR_2D_SOCIF);
+		if (time_after(jiffies, t)) {
+			psb_spank(dev_priv);
+			return -EIO;
+		}
+	}
+	return 0;
+}
+
+/**
+ *	psb_2d_submit		-	submit a 2D command
+ *	@dev_priv: our DRM device
+ *	@cmdbuf: command to issue
+ *	@size: length (in dwords)
+ *
+ *	Issue one or more 2D commands to the accelerator. This needs to be
+ *	serialized later when we add the GEM interfaces for acceleration
+ */
+static int psbfb_2d_submit(struct drm_psb_private *dev_priv, uint32_t *cmdbuf,
+								unsigned size)
+{
+	int ret = 0;
+	int i;
+	unsigned submit_size;
+
+	mutex_lock(&dev_priv->mutex_2d);
+	while (size > 0) {
+		submit_size = (size < 0x60) ? size : 0x60;
+		size -= submit_size;
+		ret = psb_2d_wait_available(dev_priv, submit_size);
+		if (ret)
+		        break;
+
+		submit_size <<= 2;
+
+		for (i = 0; i < submit_size; i += 4)
+			PSB_WSGX32(*cmdbuf++, PSB_SGX_2D_SLAVE_PORT + i);
+
+		(void)PSB_RSGX32(PSB_SGX_2D_SLAVE_PORT + i - 4);
+	}
+	mutex_unlock(&dev_priv->mutex_2d);
+	return ret;
+}
+
+
+/**
+ *	psb_accel_2d_copy_direction	-	compute blit order
+ *	@xdir: X direction of move
+ *	@ydir: Y direction of move
+ *
+ *	Compute the correct order setings to ensure that an overlapping blit
+ *	correctly copies all the pixels.
+ */
+static u32 psb_accel_2d_copy_direction(int xdir, int ydir)
+{
+	if (xdir < 0)
+		return (ydir < 0) ? PSB_2D_COPYORDER_BR2TL :
+						PSB_2D_COPYORDER_TR2BL;
+	else
+		return (ydir < 0) ? PSB_2D_COPYORDER_BL2TR :
+						PSB_2D_COPYORDER_TL2BR;
+}
+
+/**
+ *	psb_accel_2d_copy		-	accelerated 2D copy
+ *	@dev_priv: our DRM device
+ *	@src_offset in bytes
+ *	@src_stride in bytes
+ *	@src_format psb 2D format defines
+ *	@dst_offset in bytes
+ *	@dst_stride in bytes
+ *	@dst_format psb 2D format defines
+ *	@src_x offset in pixels
+ *	@src_y offset in pixels
+ *	@dst_x offset in pixels
+ *	@dst_y offset in pixels
+ *	@size_x of the copied area
+ *	@size_y of the copied area
+ *
+ *	Format and issue a 2D accelerated copy command.
+ */
+static int psb_accel_2d_copy(struct drm_psb_private *dev_priv,
+			     uint32_t src_offset, uint32_t src_stride,
+			     uint32_t src_format, uint32_t dst_offset,
+			     uint32_t dst_stride, uint32_t dst_format,
+			     uint16_t src_x, uint16_t src_y,
+			     uint16_t dst_x, uint16_t dst_y,
+			     uint16_t size_x, uint16_t size_y)
+{
+	uint32_t blit_cmd;
+	uint32_t buffer[10];
+	uint32_t *buf;
+	uint32_t direction;
+
+	buf = buffer;
+
+	direction =
+	    psb_accel_2d_copy_direction(src_x - dst_x, src_y - dst_y);
+
+	if (direction == PSB_2D_COPYORDER_BR2TL ||
+	    direction == PSB_2D_COPYORDER_TR2BL) {
+		src_x += size_x - 1;
+		dst_x += size_x - 1;
+	}
+	if (direction == PSB_2D_COPYORDER_BR2TL ||
+	    direction == PSB_2D_COPYORDER_BL2TR) {
+		src_y += size_y - 1;
+		dst_y += size_y - 1;
+	}
+
+	blit_cmd =
+	    PSB_2D_BLIT_BH |
+	    PSB_2D_ROT_NONE |
+	    PSB_2D_DSTCK_DISABLE |
+	    PSB_2D_SRCCK_DISABLE |
+	    PSB_2D_USE_PAT | PSB_2D_ROP3_SRCCOPY | direction;
+
+	*buf++ = PSB_2D_FENCE_BH;
+	*buf++ =
+	    PSB_2D_DST_SURF_BH | dst_format | (dst_stride <<
+					       PSB_2D_DST_STRIDE_SHIFT);
+	*buf++ = dst_offset;
+	*buf++ =
+	    PSB_2D_SRC_SURF_BH | src_format | (src_stride <<
+					       PSB_2D_SRC_STRIDE_SHIFT);
+	*buf++ = src_offset;
+	*buf++ =
+	    PSB_2D_SRC_OFF_BH | (src_x << PSB_2D_SRCOFF_XSTART_SHIFT) |
+	    (src_y << PSB_2D_SRCOFF_YSTART_SHIFT);
+	*buf++ = blit_cmd;
+	*buf++ =
+	    (dst_x << PSB_2D_DST_XSTART_SHIFT) | (dst_y <<
+						  PSB_2D_DST_YSTART_SHIFT);
+	*buf++ =
+	    (size_x << PSB_2D_DST_XSIZE_SHIFT) | (size_y <<
+						  PSB_2D_DST_YSIZE_SHIFT);
+	*buf++ = PSB_2D_FLUSH_BH;
+
+	return psbfb_2d_submit(dev_priv, buffer, buf - buffer);
+}
+
+/**
+ *	psbfb_copyarea_accel	-	copyarea acceleration for /dev/fb
+ *	@info: our framebuffer
+ *	@a: copyarea parameters from the framebuffer core
+ *
+ *	Perform a 2D copy via the accelerator
+ */
+static void psbfb_copyarea_accel(struct fb_info *info,
+				 const struct fb_copyarea *a)
+{
+	struct psb_fbdev *fbdev = info->par;
+	struct psb_framebuffer *psbfb = &fbdev->pfb;
+	struct drm_device *dev = psbfb->base.dev;
+	struct drm_framebuffer *fb = fbdev->psb_fb_helper.fb;
+	struct drm_psb_private *dev_priv = dev->dev_private;
+	uint32_t offset;
+	uint32_t stride;
+	uint32_t src_format;
+	uint32_t dst_format;
+
+	if (!fb)
+		return;
+
+	offset = psbfb->gtt->offset;
+	stride = fb->pitch;
+
+	switch (fb->depth) {
+	case 8:
+		src_format = PSB_2D_SRC_332RGB;
+		dst_format = PSB_2D_DST_332RGB;
+		break;
+	case 15:
+		src_format = PSB_2D_SRC_555RGB;
+		dst_format = PSB_2D_DST_555RGB;
+		break;
+	case 16:
+		src_format = PSB_2D_SRC_565RGB;
+		dst_format = PSB_2D_DST_565RGB;
+		break;
+	case 24:
+	case 32:
+		/* this is wrong but since we don't do blending its okay */
+		src_format = PSB_2D_SRC_8888ARGB;
+		dst_format = PSB_2D_DST_8888ARGB;
+		break;
+	default:
+		/* software fallback */
+		cfb_copyarea(info, a);
+		return;
+	}
+
+	if (!gma_power_begin(dev, false)) {
+		cfb_copyarea(info, a);
+		return;
+	}
+	psb_accel_2d_copy(dev_priv,
+			  offset, stride, src_format,
+			  offset, stride, dst_format,
+			  a->sx, a->sy, a->dx, a->dy, a->width, a->height);
+	gma_power_end(dev);
+}
+
+/**
+ *	psbfb_copyarea	-	2D copy interface
+ *	@info: our framebuffer
+ *	@region: region to copy
+ *
+ *	Copy an area of the framebuffer console either by the accelerator
+ *	or directly using the cfb helpers according to the request
+ */
+void psbfb_copyarea(struct fb_info *info,
+			   const struct fb_copyarea *region)
+{
+	if (unlikely(info->state != FBINFO_STATE_RUNNING))
+		return;
+
+	/* Avoid the 8 pixel erratum */
+	if (region->width == 8 || region->height == 8 ||
+		(info->flags & FBINFO_HWACCEL_DISABLED))
+		return cfb_copyarea(info, region);
+
+	psbfb_copyarea_accel(info, region);
+}
+
+/**
+ *	psbfb_sync	-	synchronize 2D
+ *	@info: our framebuffer
+ *
+ *	Wait for the 2D engine to quiesce so that we can do CPU
+ *	access to the framebuffer again
+ */
+int psbfb_sync(struct fb_info *info)
+{
+	struct psb_fbdev *fbdev = info->par;
+	struct psb_framebuffer *psbfb = &fbdev->pfb;
+	struct drm_device *dev = psbfb->base.dev;
+	struct drm_psb_private *dev_priv = dev->dev_private;
+	unsigned long _end = jiffies + DRM_HZ;
+	int busy = 0;
+
+	mutex_lock(&dev_priv->mutex_2d);
+	/*
+	 * First idle the 2D engine.
+	 */
+
+	if ((PSB_RSGX32(PSB_CR_2D_SOCIF) == _PSB_C2_SOCIF_EMPTY) &&
+	    ((PSB_RSGX32(PSB_CR_2D_BLIT_STATUS) & _PSB_C2B_STATUS_BUSY) == 0))
+		goto out;
+
+	do {
+		busy = (PSB_RSGX32(PSB_CR_2D_SOCIF) != _PSB_C2_SOCIF_EMPTY);
+		cpu_relax();
+	} while (busy && !time_after_eq(jiffies, _end));
+
+	if (busy)
+		busy = (PSB_RSGX32(PSB_CR_2D_SOCIF) != _PSB_C2_SOCIF_EMPTY);
+	if (busy)
+		goto out;
+
+	do {
+		busy = ((PSB_RSGX32(PSB_CR_2D_BLIT_STATUS) &
+						_PSB_C2B_STATUS_BUSY) != 0);
+		cpu_relax();
+	} while (busy && !time_after_eq(jiffies, _end));
+	if (busy)
+		busy = ((PSB_RSGX32(PSB_CR_2D_BLIT_STATUS) &
+					_PSB_C2B_STATUS_BUSY) != 0);
+
+out:
+	mutex_unlock(&dev_priv->mutex_2d);
+	return (busy) ? -EBUSY : 0;
+}