neutrino/src/driver/fbaccel.cpp

/*
	Neutrino-GUI  -   DBoxII-Project

	Copyright (C) 2001 Steffen Hehn 'McClean'
                      2003 thegoodguy
	Copyright (C) 2007-2013 Stefan Seyfried

	Framebuffer acceleration hardware abstraction functions.
	The various hardware dependent framebuffer acceleration functions
	are represented in this class.

	License: GPL

	This program is free software; you can redistribute it and/or modify
	it under the terms of the GNU General Public License as published by
	the Free Software Foundation; either version 2 of the License, or
	(at your option) any later version.

	This program is distributed in the hope that 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, see <http://www.gnu.org/licenses/>.
*/

#ifdef HAVE_CONFIG_H
#include <config.h>
#endif

#include <driver/framebuffer_ng.h>
#include <driver/fbaccel.h>

#include <stdio.h>
#include <fcntl.h>
#include <sys/ioctl.h>
#include <unistd.h>
#include <sys/mman.h>
#include <memory.h>
#include <math.h>
#include <limits.h>

#include <linux/kd.h>

#include <stdlib.h>

#if HAVE_COOL_HARDWARE
#include <cs_api.h>
#include <cnxtfb.h>
#endif
#if HAVE_SPARK_HARDWARE
#include <linux/stmfb.h>
#include <bpamem.h>
#endif
#if HAVE_GENERIC_HARDWARE
#include <glfb.h>
extern GLFramebuffer *glfb;
#endif

#include <driver/abstime.h>
#include <system/set_threadname.h>

#if HAVE_COOL_HARDWARE || HAVE_TRIPLEDRAGON
#define NEED_BLIT_THREAD 0
#else
#define NEED_BLIT_THREAD 1
#endif

/* note that it is *not* enough to just change those values */
#define DEFAULT_XRES 1280
#define DEFAULT_YRES 720
#define DEFAULT_BPP  32

//#undef USE_NEVIS_GXA //FIXME
/*******************************************************************************/
#ifdef USE_NEVIS_GXA
#define GXA_POINT(x, y)         (((y) & 0x0FFF) << 16) | ((x) & 0x0FFF)
#define GXA_SRC_BMP_SEL(x)      (x << 8)
#define GXA_DST_BMP_SEL(x)      (x << 5)
#define GXA_PARAM_COUNT(x)      (x << 2)

#define GXA_CMD_REG		0x001C
#define GXA_FG_COLOR_REG	0x0020
#define GXA_BG_COLOR_REG        0x0024
#define GXA_LINE_CONTROL_REG    0x0038
#define GXA_BMP1_TYPE_REG       0x0048
#define GXA_BMP1_ADDR_REG       0x004C
#define GXA_BMP2_TYPE_REG       0x0050
#define GXA_BMP2_ADDR_REG       0x0054
#define GXA_BMP3_TYPE_REG       0x0058
#define GXA_BMP3_ADDR_REG       0x005C
#define GXA_BMP4_TYPE_REG       0x0060
#define GXA_BMP4_ADDR_REG       0x0064
#define GXA_BMP5_TYPE_REG       0x0068
#define GXA_BMP5_ADDR_REG       0x006C
#define GXA_BMP6_TYPE_REG       0x0070
#define GXA_BMP7_TYPE_REG       0x0078
#define GXA_DEPTH_REG		0x00F4
#define GXA_CONTENT_ID_REG      0x0144

#define GXA_CMD_BLT             0x00010800
#define GXA_CMD_NOT_ALPHA       0x00011000
#define GXA_CMD_NOT_TEXT        0x00018000
#define GXA_CMD_QMARK		0x00001000

#define GXA_BLEND_CFG_REG       0x003C
#define GXA_CFG_REG             0x0030
#define GXA_CFG2_REG            0x00FC
/*
static unsigned int _read_gxa(volatile unsigned char *base_addr, unsigned int offset)
{
	return *(volatile unsigned int *)(base_addr + offset);
}
*/

static unsigned int _mark = 0;

static void _write_gxa(volatile unsigned char *base_addr, unsigned int offset, unsigned int value)
{
	while ((*(volatile unsigned int *)(base_addr + GXA_DEPTH_REG)) & 0x40000000) {};
	*(volatile unsigned int *)(base_addr + offset) = value;
}

/* this adds a tagged marker into the GXA queue. Once this comes out
   of the other end of the queue, all commands before it are finished */
void CFbAccel::add_gxa_sync_marker(void)
{
	unsigned int cmd = GXA_CMD_QMARK | GXA_PARAM_COUNT(1);
	// TODO: locking?
	_mark++;
	_mark &= 0x0000001F; /* bit 0x20 crashes the kernel, if set */
	_write_gxa(gxa_base, cmd, _mark);
	//fprintf(stderr, "%s: wrote %02x\n", __FUNCTION__, _mark);
}

/* wait until the current marker comes out of the GXA command queue */
void CFbAccel::waitForIdle(void)
{
	unsigned int cfg, count = 0;
	do {
		cfg = *(volatile unsigned int *)(gxa_base + GXA_CMD_REG);
		cfg >>= 24;	/* the token is stored in bits 31...24 */
		if (cfg == _mark)
			break;
		/* usleep is too coarse, because of CONFIG_HZ=100 in kernel
		   so use sched_yield to at least give other threads a chance to run */
		sched_yield();
		//fprintf(stderr, "%s: read  %02x, expected %02x\n", __FUNCTION__, cfg, _mark);
	} while(++count < 8192); /* don't deadlock here if there is an error */

	if (count > 2048) /* more than 2000 are unlikely, even for large BMP6 blits */
		fprintf(stderr, "CFbAccel::waitForIdle: count is big (%d)!\n", count);
}
#elif HAVE_TRIPLEDRAGON
#include <directfb.h>
#include <tdgfx/stb04gfx.h>
extern IDirectFB *dfb;
extern IDirectFBSurface *dfbdest;
extern int gfxfd;
void CFbAccel::waitForIdle(void)
{
#if 0
	struct timeval ts, te;
	gettimeofday(&ts, NULL);
#endif
	/* does not work: DFBResult r = dfb->WaitForSync(dfb); */
	ioctl(gfxfd, STB04GFX_ENGINE_SYNC);
#if 0
	gettimeofday(&te, NULL);
	printf("STB04GFX_ENGINE_SYNC took %lld us\n", (te.tv_sec * 1000000LL + te.tv_usec) - (ts.tv_sec * 1000000LL + ts.tv_usec));
#endif
}
#elif HAVE_SPARK_HARDWARE

static int bpafd = -1;
static size_t lbb_sz = 1920 * 1080;	/* offset from fb start in 'pixels' */
static size_t lbb_off = lbb_sz * sizeof(fb_pixel_t);	/* same in bytes */
static int backbuf_sz = 0;

void CFbAccel::waitForIdle(void)
{
#if 0	/* blits too often and does not seem to be necessary */
	blit_mutex.lock();
	if (blit_pending)
	{
		blit_mutex.unlock();
		_blit();
		return;
	}
	blit_mutex.unlock();
#endif
	OpenThreads::ScopedLock<OpenThreads::Mutex> m_lock(mutex);
	ioctl(fb->fd, STMFBIO_SYNC_BLITTER);
}
#else
void CFbAccel::waitForIdle(void)
{
}
#endif

CFbAccel::CFbAccel(CFrameBuffer *_fb)
{
	blit_thread = false;
	fb = _fb;
	init();
	lastcol = 0xffffffff;
	lbb = fb->lfb;	/* the memory area to draw to... */
#ifdef HAVE_SPARK_HARDWARE
	if (fb->available < 12*1024*1024)
	{
		/* for old installations that did not upgrade their module config
		 * it will still work good enough to display the message below */
		fprintf(stderr, "[neutrino] WARNING: not enough framebuffer memory available!\n");
		fprintf(stderr, "[neutrino]          I need at least 12MB.\n");
		FILE *f = fopen("/tmp/infobar.txt", "w");
		if (f) {
			fprintf(f, "NOT ENOUGH FRAMEBUFFER MEMORY!");
			fclose(f);
		}
		lbb_sz = 0;
		lbb_off = 0;
	}
	lbb = fb->lfb + lbb_sz;
	bpafd = open("/dev/bpamem0", O_RDWR | O_CLOEXEC);
	if (bpafd < 0)
	{
		fprintf(stderr, "[neutrino] FB: cannot open /dev/bpamem0: %m\n");
		return;
	}
	backbuf_sz = 1280 * 720 * sizeof(fb_pixel_t);
	BPAMemAllocMemData bpa_data;
	bpa_data.bpa_part = (char *)"LMI_VID";
	bpa_data.mem_size = backbuf_sz;
	int res;
	res = ioctl(bpafd, BPAMEMIO_ALLOCMEM, &bpa_data);
	if (res)
	{
		fprintf(stderr, "[neutrino] FB: cannot allocate from bpamem: %m\n");
		fprintf(stderr, "backbuf_sz: %d\n", backbuf_sz);
		close(bpafd);
		bpafd = -1;
		return;
	}
	close(bpafd);

	char bpa_mem_device[30];
	sprintf(bpa_mem_device, "/dev/bpamem%d", bpa_data.device_num);
	bpafd = open(bpa_mem_device, O_RDWR | O_CLOEXEC);
	if (bpafd < 0)
	{
		fprintf(stderr, "[neutrino] FB: cannot open secondary %s: %m\n", bpa_mem_device);
		return;
	}

	backbuffer = (fb_pixel_t *)mmap(0, bpa_data.mem_size, PROT_WRITE|PROT_READ, MAP_SHARED, bpafd, 0);
	if (backbuffer == MAP_FAILED)
	{
		fprintf(stderr, "[neutrino] FB: cannot map from bpamem: %m\n");
		ioctl(bpafd, BPAMEMIO_FREEMEM);
		close(bpafd);
		bpafd = -1;
		return;
	}
#ifdef PARTIAL_BLIT
	to_blit.xs = to_blit.ys = INT_MAX;
	to_blit.xe = to_blit.ye = 0;
	last_xres = 0;
#endif
#endif

#ifdef USE_NEVIS_GXA
	/* Open /dev/mem for HW-register access */
	devmem_fd = open("/dev/mem", O_RDWR | O_SYNC | O_CLOEXEC);
	if (devmem_fd < 0) {
		perror("CFbAccel open /dev/mem");
		goto error;
	}

	/* mmap the GXA's base address */
	gxa_base = (volatile unsigned char*)mmap(0, 0x00040000, PROT_READ|PROT_WRITE, MAP_SHARED, devmem_fd, 0xE0600000);
	if (gxa_base == (void *)-1) {
		perror("CFbAccel mmap /dev/mem");
		goto error;
	}

	setupGXA();
 error:
	/* TODO: what to do here? does this really happen? */
	;
#endif /* USE_NEVIS_GXA */

#if NEED_BLIT_THREAD
	/* start the autoblit-thread (run() function) */
	OpenThreads::Thread::start();
#endif
};

CFbAccel::~CFbAccel()
{
	if (blit_thread)
	{
		blit_thread = false;
		blit(); /* wakes up the thread */
		OpenThreads::Thread::join();
	}
#if HAVE_SPARK_HARDWARE
	if (backbuffer)
	{
		fprintf(stderr, "CFbAccel: unmap backbuffer\n");
		munmap(backbuffer, backbuf_sz);
	}
	if (bpafd != -1)
	{
		fprintf(stderr, "CFbAccel: BPAMEMIO_FREEMEM\n");
		ioctl(bpafd, BPAMEMIO_FREEMEM);
		close(bpafd);
	}
#endif
#ifdef USE_NEVIS_GXA
	if (gxa_base != MAP_FAILED)
		munmap((void *)gxa_base, 0x40000);
	if (devmem_fd != -1)
		close(devmem_fd);
#endif
#if ! HAVE_GENERIC_HARDWARE
	if (fb->lfb)
		munmap(fb->lfb, fb->available);
	if (fb->fd > -1)
		close(fb->fd);
#endif
}

void CFbAccel::update()
{
#ifndef HAVE_SPARK_HARDWARE
	int needmem = fb->stride * fb->yRes * 2;
	if (fb->available >= needmem)
	{
		backbuffer = fb->lfb + fb->stride / sizeof(fb_pixel_t) * fb->yRes;
		return;
	}
	fprintf(stderr, "CFbAccel: not enough FB memory (have %d, need %d)\n", fb->available, needmem);
	backbuffer = fb->lfb; /* will not work well, but avoid crashes */
#endif
}

void CFbAccel::setColor(fb_pixel_t col)
{
#if HAVE_TRIPLEDRAGON
	if (col == lastcol)
		return;
	char *c = (char *)&col;
	dfbdest->SetColor(dfbdest, c[1], c[2], c[3], c[0]);
	lastcol = col;
#elif defined USE_NEVIS_GXA
	if (col == lastcol)
		return;
	_write_gxa(gxa_base, GXA_FG_COLOR_REG, (unsigned int)col); /* setup the drawing color */
	lastcol = col;
#else
	(void)col; /* avoid "unused parameter" compiler warning */
#endif
}

void CFbAccel::paintRect(const int x, const int y, const int dx, const int dy, const fb_pixel_t col)
{
#if HAVE_TRIPLEDRAGON
	setColor(col);
	dfbdest->FillRectangle(dfbdest, x, y, dx, dy);
#elif defined(USE_NEVIS_GXA)
	OpenThreads::ScopedLock<OpenThreads::Mutex> m_lock(mutex);
	unsigned int cmd = GXA_CMD_BLT | GXA_CMD_NOT_TEXT | GXA_CMD_NOT_ALPHA |
			   GXA_SRC_BMP_SEL(6) | GXA_DST_BMP_SEL(2) | GXA_PARAM_COUNT(2);

	_write_gxa(gxa_base, GXA_BG_COLOR_REG, (unsigned int)col); /* setup the drawing color */
	_write_gxa(gxa_base, GXA_BMP6_TYPE_REG, (3 << 16) | (1 << 27)); /* 3 == 32bpp, 1<<27 == fill */
	_write_gxa(gxa_base, cmd, GXA_POINT(x, y));	/* destination pos */
	_write_gxa(gxa_base, cmd, GXA_POINT(dx, dy));	/* destination size */
	_write_gxa(gxa_base, GXA_BG_COLOR_REG, (unsigned int)fb->backgroundColor);

	/* the GXA seems to do asynchronous rendering, so we add a sync marker
	   to which the fontrenderer code can synchronize */
	add_gxa_sync_marker();
#elif HAVE_SPARK_HARDWARE
	if (dx <= 0 || dy <= 0)
		return;

	// The STM blitter introduces considerable overhead probably not worth for single lines. --martii
	if (dx == 1) {
		waitForIdle();
		fb_pixel_t *fbs = fb->getFrameBufferPointer() + (DEFAULT_XRES * y) + x;
		fb_pixel_t *fbe = fbs + DEFAULT_XRES * dy;
		while (fbs < fbe) {
			*fbs = col;
			fbs += DEFAULT_XRES;
		}
		mark(x , y, x + 1, y + dy);
		return;
	}
	if (dy == 1) {
		waitForIdle();
		fb_pixel_t *fbs = fb->getFrameBufferPointer() + (DEFAULT_XRES * y) + x;
		fb_pixel_t *fbe = fbs + dx;
		while (fbs < fbe)
			*fbs++ = col;
		mark(x , y, x + dx, y + 1);
		return;
	}

	/* function has const parameters, so copy them here... */
	int width = dx;
	int height = dy;
	int xx = x;
	int yy = y;
	/* maybe we should just return instead of fixing this up... */
	if (x < 0) {
		fprintf(stderr, "[neutrino] fb::%s: x < 0 (%d)\n", __func__, x);
		width += x;
		if (width <= 0)
			return;
		xx = 0;
	}

	if (y < 0) {
		fprintf(stderr, "[neutrino] fb::%s: y < 0 (%d)\n", __func__, y);
		height += y;
		if (height <= 0)
			return;
		yy = 0;
	}

	int right = xx + width;
	int bottom = yy + height;

	if (right > (int)fb->xRes) {
		if (xx >= (int)fb->xRes) {
			fprintf(stderr, "[neutrino] fb::%s: x >= xRes (%d > %d)\n", __func__, xx, fb->xRes);
			return;
		}
		fprintf(stderr, "[neutrino] fb::%s: x+w > xRes! (%d+%d > %d)\n", __func__, xx, width, fb->xRes);
		right = fb->xRes;
	}
	if (bottom > (int)fb->yRes) {
		if (yy >= (int)fb->yRes) {
			fprintf(stderr, "[neutrino] fb::%s: y >= yRes (%d > %d)\n", __func__, yy, fb->yRes);
			return;
		}
		fprintf(stderr, "[neutrino] fb::%s: y+h > yRes! (%d+%d > %d)\n", __func__, yy, height, fb->yRes);
		bottom = fb->yRes;
	}

	STMFBIO_BLT_DATA bltData;
	memset(&bltData, 0, sizeof(STMFBIO_BLT_DATA));

	bltData.operation  = BLT_OP_FILL;
	bltData.dstOffset  = lbb_off;
	bltData.dstPitch   = fb->stride;

	bltData.dst_left   = xx;
	bltData.dst_top    = yy;
	bltData.dst_right  = right;
	bltData.dst_bottom = bottom;

	bltData.dstFormat  = SURF_ARGB8888;
	bltData.srcFormat  = SURF_ARGB8888;
	bltData.dstMemBase = STMFBGP_FRAMEBUFFER;
	bltData.srcMemBase = STMFBGP_FRAMEBUFFER;
	bltData.colour     = col;

	mark(xx, yy, bltData.dst_right, bltData.dst_bottom);
	OpenThreads::ScopedLock<OpenThreads::Mutex> m_lock(mutex);
	if (ioctl(fb->fd, STMFBIO_BLT, &bltData ) < 0)
		fprintf(stderr, "blitRect FBIO_BLIT: %m x:%d y:%d w:%d h:%d s:%d\n", xx,yy,width,height,fb->stride);
#else
	int line = 0;
	int swidth = fb->stride / sizeof(fb_pixel_t);
	fb_pixel_t *fbp = fb->getFrameBufferPointer() + (swidth * y);
	int pos;
	while (line < dy)
	{
		for (pos = x; pos < x + dx; pos++)
			*(fbp + pos) = col;
		fbp += swidth;
		line++;
	}
#endif
	blit();
}

void CFbAccel::paintPixel(const int x, const int y, const fb_pixel_t col)
{
#if HAVE_TRIPLEDRAGON
	setColor(col);
	dfbdest->DrawLine(dfbdest, x, y, x, y);
#elif defined (USE_NEVIS_GXA)
	paintLine(x, y, x + 1, y, col);
#else
	fb_pixel_t *pos = fb->getFrameBufferPointer();
	pos += (fb->stride / sizeof(fb_pixel_t)) * y;
	pos += x;
	*pos = col;
#endif
}

void CFbAccel::paintLine(int xa, int ya, int xb, int yb, const fb_pixel_t col)
{
#if HAVE_TRIPLEDRAGON
	setColor(col);
	dfbdest->DrawLine(dfbdest, xa, ya, xb, yb);
#elif defined(USE_NEVIS_GXA)
	OpenThreads::ScopedLock<OpenThreads::Mutex> m_lock(mutex);
	/* draw a single vertical line from point xa/ya to xb/yb */
	unsigned int cmd = GXA_CMD_NOT_TEXT | GXA_SRC_BMP_SEL(2) | GXA_DST_BMP_SEL(2) | GXA_PARAM_COUNT(2) | GXA_CMD_NOT_ALPHA;

	setColor(col);
	_write_gxa(gxa_base, GXA_LINE_CONTROL_REG, 0x00000404);	/* X is major axis, skip last pixel */
	_write_gxa(gxa_base, cmd, GXA_POINT(xb, yb));		/* end point */
	_write_gxa(gxa_base, cmd, GXA_POINT(xa, ya));		/* start point */
#else
	int dx = abs (xa - xb);
	int dy = abs (ya - yb);
	if (dy == 0) /* horizontal line */
	{
		/* paintRect actually is 1 pixel short to the right,
		 * but that's bug-compatibility with the GXA code */
		paintRect(xa, ya, xb - xa, 1, col);
		return;
	}
	if (dx == 0) /* vertical line */
	{
		paintRect(xa, ya, 1, yb - ya, col);
		return;
	}
	int x;
	int y;
	int End;
	int step;

	if (dx > dy)
	{
		int p = 2 * dy - dx;
		int twoDy = 2 * dy;
		int twoDyDx = 2 * (dy-dx);

		if (xa > xb)
		{
			x = xb;
			y = yb;
			End = xa;
			step = ya < yb ? -1 : 1;
		}
		else
		{
			x = xa;
			y = ya;
			End = xb;
			step = yb < ya ? -1 : 1;
		}

		paintPixel(x, y, col);

		while (x < End)
		{
			x++;
			if (p < 0)
				p += twoDy;
			else
			{
				y += step;
				p += twoDyDx;
			}
			paintPixel(x, y, col);
		}
	}
	else
	{
		int p = 2 * dx - dy;
		int twoDx = 2 * dx;
		int twoDxDy = 2 * (dx-dy);

		if (ya > yb)
		{
			x = xb;
			y = yb;
			End = ya;
			step = xa < xb ? -1 : 1;
		}
		else
		{
			x = xa;
			y = ya;
			End = yb;
			step = xb < xa ? -1 : 1;
		}

		paintPixel(x, y, col);

		while (y < End)
		{
			y++;
			if (p < 0)
				p += twoDx;
			else
			{
				x += step;
				p += twoDxDy;
			}
			paintPixel(x, y, col);
		}
	}
	mark(xa, ya, xb, yb);
	blit();
#endif
}

#if !HAVE_TRIPLEDRAGON
void CFbAccel::blit2FB(void *fbbuff, uint32_t width, uint32_t height, uint32_t xoff, uint32_t yoff, uint32_t xp, uint32_t yp, bool transp)
{
#if !HAVE_SPARK_HARDWARE
	int  xc, yc;
	xc = (width > fb->xRes) ? fb->xRes : width;
	yc = (height > fb->yRes) ? fb->yRes : height;
#endif
#ifdef USE_NEVIS_GXA
	(void)transp;
	u32 cmd;
	void *uKva;

	uKva = cs_phys_addr(fbbuff);
	//printf("CFbAccel::blit2FB: data %x Kva %x\n", (int) fbbuff, (int) uKva);

	if (uKva != NULL) {
		OpenThreads::ScopedLock<OpenThreads::Mutex> m_lock(mutex);
		cmd = GXA_CMD_BLT | GXA_CMD_NOT_TEXT | GXA_SRC_BMP_SEL(1) | GXA_DST_BMP_SEL(2) | GXA_PARAM_COUNT(3);

		_write_gxa(gxa_base, GXA_BMP1_TYPE_REG, (3 << 16) | width);
		_write_gxa(gxa_base, GXA_BMP1_ADDR_REG, (unsigned int)uKva);

		_write_gxa(gxa_base, cmd, GXA_POINT(xoff, yoff)); /* destination pos */
		_write_gxa(gxa_base, cmd, GXA_POINT(xc, yc));     /* source width, FIXME real or adjusted xc, yc ? */
		_write_gxa(gxa_base, cmd, GXA_POINT(xp, yp));     /* source pos */

		return;
	}
#elif HAVE_SPARK_HARDWARE
	int x, y, dw, dh;
	x = xoff;
	y = yoff;
	dw = width - xp;
	dh = height - yp;

	size_t mem_sz = width * height * sizeof(fb_pixel_t);
	unsigned long ulFlags = 0;
	if (!transp) /* transp == false (default): use transparency from source alphachannel */
		ulFlags = BLT_OP_FLAGS_BLEND_SRC_ALPHA|BLT_OP_FLAGS_BLEND_DST_MEMORY; // we need alpha blending

	STMFBIO_BLT_EXTERN_DATA blt_data;
	memset(&blt_data, 0, sizeof(STMFBIO_BLT_EXTERN_DATA));
	blt_data.operation  = BLT_OP_COPY;
	blt_data.ulFlags    = ulFlags;
	blt_data.srcOffset  = 0;
	blt_data.srcPitch   = width * 4;
	blt_data.dstOffset  = lbb_off;
	blt_data.dstPitch   = fb->stride;
	blt_data.src_left   = xp;
	blt_data.src_top    = yp;
	blt_data.src_right  = width;
	blt_data.src_bottom = height;
	blt_data.dst_left   = x;
	blt_data.dst_top    = y;
	blt_data.dst_right  = x + dw;
	blt_data.dst_bottom = y + dh;
	blt_data.srcFormat  = SURF_ARGB8888;
	blt_data.dstFormat  = SURF_ARGB8888;
	blt_data.srcMemBase = (char *)backbuffer;
	blt_data.dstMemBase = (char *)fb->lfb;
	blt_data.srcMemSize = mem_sz;
	blt_data.dstMemSize = fb->stride * fb->yRes + lbb_off;

	mark(x, y, blt_data.dst_right, blt_data.dst_bottom);
	OpenThreads::ScopedLock<OpenThreads::Mutex> m_lock(mutex);
	ioctl(fb->fd, STMFBIO_SYNC_BLITTER);
	if (fbbuff != backbuffer)
		memmove(backbuffer, fbbuff, mem_sz);
	// icons are so small that they will still be in cache
	msync(backbuffer, backbuf_sz, MS_SYNC);

	if (ioctl(fb->fd, STMFBIO_BLT_EXTERN, &blt_data) < 0)
		perror("CFbAccel blit2FB STMFBIO_BLT_EXTERN");
	return;
#else
	fb_pixel_t *data = (fb_pixel_t *) fbbuff;

	uint8_t *d = (uint8_t *)lbb + xoff * sizeof(fb_pixel_t) + fb->stride * yoff;
	fb_pixel_t * d2;

	for (int count = 0; count < yc; count++ ) {
		fb_pixel_t *pixpos = &data[(count + yp) * width];
		d2 = (fb_pixel_t *) d;
		for (int count2 = 0; count2 < xc; count2++ ) {
			fb_pixel_t pix = *(pixpos + xp);
			if (transp || (pix & 0xff000000) == 0xff000000)
				*d2 = pix;
			else {
				uint8_t *in = (uint8_t *)(pixpos + xp);
				uint8_t *out = (uint8_t *)d2;
				int a = in[3];	/* TODO: big/little endian */
				*out = (*out + ((*in - *out) * a) / 256);
				in++; out++;
				*out = (*out + ((*in - *out) * a) / 256);
				in++; out++;
				*out = (*out + ((*in - *out) * a) / 256);
			}
			d2++;
			pixpos++;
		}
		d += fb->stride;
	}
	blit();
#if 0
	for(int i = 0; i < yc; i++){
		memmove(clfb + (i + yoff)*stride + xoff*4, ip + (i + yp)*width + xp, xc*4);
	}
#endif
#endif
}
#else
void CFbAccel::blit2FB(void *fbbuff, uint32_t width, uint32_t height, uint32_t xoff, uint32_t yoff, uint32_t xp, uint32_t yp, bool transp)
{
	DFBRectangle src;
	DFBResult err;
	IDirectFBSurface *surf;
	DFBSurfaceDescription dsc;

	src.x = xp;
	src.y = yp;
	src.w = width - xp;
	src.h = height - yp;

	dsc.flags  = (DFBSurfaceDescriptionFlags)(DSDESC_CAPS | DSDESC_WIDTH | DSDESC_HEIGHT | DSDESC_PREALLOCATED);
	dsc.caps   = DSCAPS_NONE;
	dsc.width  = width;
	dsc.height = height;
	dsc.preallocated[0].data  = fbbuff;
	dsc.preallocated[0].pitch = width * sizeof(fb_pixel_t);
	err = dfb->CreateSurface(dfb, &dsc, &surf);
	/* TODO: maybe we should not die if this fails? */
	if (err != DFB_OK) {
		fprintf(stderr, "CFbAccel::blit2FB: ");
		DirectFBErrorFatal("dfb->CreateSurface(dfb, &dsc, &surf)", err);
	}

	if (transp)
	{
		surf->SetSrcColorKey(surf, 0, 0, 0);
		dfbdest->SetBlittingFlags(dfbdest, DSBLIT_SRC_COLORKEY);
	}
	else
		dfbdest->SetBlittingFlags(dfbdest, DSBLIT_BLEND_ALPHACHANNEL);

	dfbdest->Blit(dfbdest, surf, &src, xoff, yoff);
	surf->Release(surf);
	return;
}
#endif

#ifdef USE_NEVIS_GXA
void CFbAccel::setupGXA()
{
	// We (re)store the GXA regs here in case DFB override them and was not
	// able to restore them.
	_write_gxa(gxa_base, GXA_BMP2_TYPE_REG, (3 << 16) | (unsigned int)fb->screeninfo.xres);
	_write_gxa(gxa_base, GXA_BMP2_ADDR_REG, (unsigned int) fb->fix.smem_start);
	_write_gxa(gxa_base, GXA_BLEND_CFG_REG, 0x00089064);
	//	TODO check mono-flip, bit 8
	_write_gxa(gxa_base, GXA_CFG_REG, 0x100 | (1 << 12) | (1 << 29));
	_write_gxa(gxa_base, GXA_CFG2_REG, 0x1FF);
	_write_gxa(gxa_base, GXA_BG_COLOR_REG, (unsigned int)fb->backgroundColor);
	add_gxa_sync_marker();
}
#endif

#define BLIT_INTERVAL_MIN 40
#define BLIT_INTERVAL_MAX 250
void CFbAccel::run()
{
	printf("CFbAccel::run start\n");
	time_t last_blit = 0;
	blit_pending = false;
	blit_thread = true;
	blit_mutex.lock();
	set_threadname("fb::autoblit");
	while (blit_thread) {
		blit_cond.wait(&blit_mutex, blit_pending ? BLIT_INTERVAL_MIN : BLIT_INTERVAL_MAX);
		time_t now = time_monotonic_ms();
		if (now - last_blit < BLIT_INTERVAL_MIN)
		{
			blit_pending = true;
			//printf("CFbAccel::run: skipped, time %ld\n", now - last_blit);
		}
		else
		{
			blit_pending = false;
			blit_mutex.unlock();
			_blit();
			blit_mutex.lock();
			last_blit = now;
		}
	}
	blit_mutex.unlock();
	printf("CFbAccel::run end\n");
}

void CFbAccel::blit()
{
	//printf("CFbAccel::blit\n");
	blit_mutex.lock();
	blit_cond.signal();
	blit_mutex.unlock();
}

#if HAVE_SPARK_HARDWARE
void CFbAccel::_blit()
{
#if 0
	static time_t last = 0;
	time_t now = time_monotonic_ms();
	printf("%s %ld\n", __func__, now - last);
	last = now;
#endif
	OpenThreads::ScopedLock<OpenThreads::Mutex> m_lock(mutex);
#ifdef PARTIAL_BLIT
	if (to_blit.xs == INT_MAX)
		return;

	int srcXa = to_blit.xs;
	int srcYa = to_blit.ys;
	int srcXb = to_blit.xe;
	int srcYb = to_blit.ye;
#else
	const int srcXa = 0;
	const int srcYa = 0;
	int srcXb = fb->xRes;
	int srcYb = fb->yRes;
#endif
	STMFBIO_BLT_DATA  bltData;
	memset(&bltData, 0, sizeof(STMFBIO_BLT_DATA));

	bltData.operation  = BLT_OP_COPY;
	//bltData.ulFlags  = BLT_OP_FLAGS_BLEND_SRC_ALPHA | BLT_OP_FLAGS_BLEND_DST_MEMORY; // we need alpha blending
	// src
	bltData.srcOffset  = lbb_off;
	bltData.srcPitch   = fb->stride;

	bltData.src_left   = srcXa;
	bltData.src_top    = srcYa;
	bltData.src_right  = srcXb;
	bltData.src_bottom = srcYb;

	bltData.srcFormat = SURF_BGRA8888;
	bltData.srcMemBase = STMFBGP_FRAMEBUFFER;

	/* calculate dst/blit factor */
	fb_var_screeninfo s;
	if (ioctl(fb->fd, FBIOGET_VSCREENINFO, &s) == -1)
		perror("CFbAccel <FBIOGET_VSCREENINFO>");

#ifdef PARTIAL_BLIT
	if (s.xres != last_xres) /* fb resolution has changed -> clear artifacts */
	{
		last_xres = s.xres;
		bltData.src_left   = 0;
		bltData.src_top    = 0;
		bltData.src_right  = fb->xRes;
		bltData.src_bottom = fb->yRes;
	}

	double xFactor = (double)s.xres/(double)fb->xRes;
	double yFactor = (double)s.yres/(double)fb->yRes;

	int desXa = xFactor * bltData.src_left;
	int desYa = yFactor * bltData.src_top;
	int desXb = xFactor * bltData.src_right;
	int desYb = yFactor * bltData.src_bottom;
#else
	const int desXa = 0;
	const int desYa = 0;
	int desXb = s.xres;
	int desYb = s.yres;
#endif

	/* dst */
	bltData.dstOffset  = 0;
	bltData.dstPitch   = s.xres * 4;

	bltData.dst_left   = desXa;
	bltData.dst_top    = desYa;
	bltData.dst_right  = desXb;
	bltData.dst_bottom = desYb;

	bltData.dstFormat = SURF_BGRA8888;
	bltData.dstMemBase = STMFBGP_FRAMEBUFFER;

	//printf("CFbAccel::blit: sx:%d sy:%d sxe:%d sye: %d dx:%d dy:%d dxe:%d dye:%d\n", srcXa, srcYa, srcXb, srcYb, desXa, desYa, desXb, desYb);
	if ((bltData.dst_right > s.xres) || (bltData.dst_bottom > s.yres))
		printf("CFbAccel::blit: values out of range desXb:%d desYb:%d\n",
			bltData.dst_right, bltData.dst_bottom);

	if(ioctl(fb->fd, STMFBIO_SYNC_BLITTER) < 0)
		perror("CFbAccel::blit ioctl STMFBIO_SYNC_BLITTER 1");
	msync(lbb, fb->xRes * 4 * fb->yRes, MS_SYNC);
	if (ioctl(fb->fd, STMFBIO_BLT, &bltData ) < 0)
		perror("STMFBIO_BLT");
	if(ioctl(fb->fd, STMFBIO_SYNC_BLITTER) < 0)
		perror("CFbAccel::blit ioctl STMFBIO_SYNC_BLITTER 2");

#ifdef PARTIAL_BLIT
	to_blit.xs = to_blit.ys = INT_MAX;
	to_blit.xe = to_blit.ye = 0;
#endif
}

#elif HAVE_AZBOX_HARDWARE

#ifndef FBIO_WAITFORVSYNC
#define FBIO_WAITFORVSYNC _IOW('F', 0x20, __u32)
#endif
#ifndef FBIO_BLIT
#define FBIO_BLIT 0x22
#define FBIO_SET_MANUAL_BLIT _IOW('F', 0x21, __u8)
#endif
static bool autoblit = getenv("AZBOX_KERNEL_BLIT") ? true : false;
void CFbAccel::_blit()
{
	if (autoblit)
		return;
	// blit
	if (ioctl(fb->fd, FBIO_BLIT) < 0)
		perror("CFbAccel FBIO_BLIT");
#if 0
	// sync bliter
	int c = 0;
	if( ioctl(fd, FBIO_WAITFORVSYNC, &c) < 0 )
		perror("FBIO_WAITFORVSYNC");
#endif
}

#else
/* not azbox and not spark -> no blit() needed */
void CFbAccel::_blit()
{
#if HAVE_GENERIC_HARDWARE
	if (glfb)
		glfb->blit();
#endif
}
#endif

/* not really used yet */
#ifdef PARTIAL_BLIT
void CFbAccel::mark(int xs, int ys, int xe, int ye)
{
	OpenThreads::ScopedLock<OpenThreads::Mutex> m_lock(mutex);
	if (xs < to_blit.xs)
		to_blit.xs = xs;
	if (ys < to_blit.ys)
		to_blit.ys = ys;
	if (xe > to_blit.xe) {
		if (xe >= (int)fb->xRes)
			to_blit.xe = fb->xRes - 1;
		else
			to_blit.xe = xe;
	}
	if (ye > to_blit.ye) {
		if (ye >= (int)fb->xRes)
			to_blit.ye = fb->yRes - 1;
		else
			to_blit.ye = ye;
	}
}
#else
void CFbAccel::mark(int, int, int, int)
{
}
#endif

bool CFbAccel::init(void)
{
	fb_pixel_t *lfb;
	fb->lfb = NULL;
	fb->fd = -1;
#if HAVE_GENERIC_HARDWARE
	if (!glfb) {
		fprintf(stderr, "CFbAccel::init: GL Framebuffer is not set up? we are doomed...\n");
		return false;
	}
	fb->screeninfo = glfb->getScreenInfo();
	fb->stride = 4 * fb->screeninfo.xres;
	fb->available = glfb->getOSDBuffer()->size(); /* allocated in glfb constructor */
	lfb = reinterpret_cast<fb_pixel_t*>(glfb->getOSDBuffer()->data());
#else
	int fd;
#if HAVE_TRIPLEDRAGON
	/* kernel is too old for O_CLOEXEC :-( */
	fd = open("/dev/fb0", O_RDWR);
	if (fd != -1)
		fcntl(fd, F_SETFD, FD_CLOEXEC);
#else
	fd = open("/dev/fb0", O_RDWR|O_CLOEXEC);
#endif
	if (fd < 0) {
		perror("open /dev/fb0");
		return false;
	}
	fb->fd = fd;

	if (ioctl(fd, FBIOGET_VSCREENINFO, &fb->screeninfo) < 0) {
		perror("FBIOGET_VSCREENINFO");
		return false;
	}

	if (ioctl(fd, FBIOGET_FSCREENINFO, &fb->fix) < 0) {
		perror("FBIOGET_FSCREENINFO");
		return false;
	}

	fb->available = fb->fix.smem_len;
	printf("%dk video mem\n", fb->available / 1024);
	lfb = (fb_pixel_t *)mmap(0, fb->available, PROT_WRITE|PROT_READ, MAP_SHARED, fd, 0);

	if (lfb == MAP_FAILED) {
		perror("mmap");
		return false;;
	}
#endif /* ! GENERIC_HARDWARE */

	memset(lfb, 0, fb->available);
	fb->lfb = lfb;
	return true;
}

/* wrong name... */
int CFbAccel::setMode(void)
{
	int fd = fb->fd;
	t_fb_var_screeninfo *si = &fb->screeninfo;
#if HAVE_AZBOX_HARDWARE
	// set auto blit if AZBOX_KERNEL_BLIT environment variable is set
	unsigned char tmp = getenv("AZBOX_KERNEL_BLIT") ? 0 : 1;
	if (ioctl(fd, FBIO_SET_MANUAL_BLIT, &tmp) < 0)
		perror("FBIO_SET_MANUAL_BLIT");

	const unsigned int nxRes = DEFAULT_XRES;
	const unsigned int nyRes = DEFAULT_YRES;
	const unsigned int nbpp  = DEFAULT_BPP;
	si->xres_virtual = si->xres = nxRes;
	si->yres_virtual = (si->yres = nyRes) * 2;
	si->height = 0;
	si->width = 0;
	si->xoffset = si->yoffset = 0;
	si->bits_per_pixel = nbpp;

	si->transp.offset = 24;
	si->transp.length = 8;
	si->red.offset = 16;
	si->red.length = 8;
	si->green.offset = 8;
	si->green.length = 8;
	si->blue.offset = 0;
	si->blue.length = 8;

	if (ioctl(fd, FBIOPUT_VSCREENINFO, si) < 0) {
		// try single buffering
		si->yres_virtual = si->yres = nyRes;
		if (ioctl(fd, FBIOPUT_VSCREENINFO, si) < 0)
		perror("FBIOPUT_VSCREENINFO");
		printf("FB: double buffering not available.\n");
	}
	else
		printf("FB: double buffering available!\n");

	ioctl(fd, FBIOGET_VSCREENINFO, si);

	if (si->xres != nxRes || si->yres != nyRes || si->bits_per_pixel != nbpp)
	{
		printf("SetMode failed: wanted: %dx%dx%d, got %dx%dx%d\n",
		       nxRes, nyRes, nbpp,
		       si->xres, si->yres, si->bits_per_pixel);
	}
#endif
#if HAVE_SPARK_HARDWARE
	/* it's all fake... :-) */
	si->xres = si->xres_virtual = DEFAULT_XRES;
	si->yres = si->yres_virtual = DEFAULT_YRES;
	si->bits_per_pixel = DEFAULT_BPP;
	fb->stride = si->xres * si->bits_per_pixel / 8;
#else
#if ! HAVE_GENERIC_HARDWARE
	fb_fix_screeninfo _fix;

	if (ioctl(fd, FBIOGET_FSCREENINFO, &_fix) < 0) {
		perror("FBIOGET_FSCREENINFO");
		return -1;
	}
	fb->stride = _fix.line_length;
#endif
#endif
#if HAVE_COOL_HARDWARE
	if (ioctl(fd, FBIOBLANK, FB_BLANK_UNBLANK) < 0)
		printf("screen unblanking failed\n");
#endif
	/* avoid compiler warnings on various platforms */
	(void) fd;
	(void) si;
	return 0;
}