linux_old1/drivers/video/fbdev/neofb.c

2236 lines
55 KiB
C
Raw Normal View History

/*
* linux/drivers/video/neofb.c -- NeoMagic Framebuffer Driver
*
* Copyright (c) 2001-2002 Denis Oliver Kropp <dok@directfb.org>
*
*
* Card specific code is based on XFree86's neomagic driver.
* Framebuffer framework code is based on code of cyber2000fb.
*
* This file is subject to the terms and conditions of the GNU General
* Public License. See the file COPYING in the main directory of this
* archive for more details.
*
*
* 0.4.1
* - Cosmetic changes (dok)
*
* 0.4
* - Toshiba Libretto support, allow modes larger than LCD size if
* LCD is disabled, keep BIOS settings if internal/external display
* haven't been enabled explicitly
* (Thomas J. Moore <dark@mama.indstate.edu>)
*
* 0.3.3
* - Porting over to new fbdev api. (jsimmons)
*
* 0.3.2
* - got rid of all floating point (dok)
*
* 0.3.1
* - added module license (dok)
*
* 0.3
* - hardware accelerated clear and move for 2200 and above (dok)
* - maximum allowed dotclock is handled now (dok)
*
* 0.2.1
* - correct panning after X usage (dok)
* - added module and kernel parameters (dok)
* - no stretching if external display is enabled (dok)
*
* 0.2
* - initial version (dok)
*
*
* TODO
* - ioctl for internal/external switching
* - blanking
* - 32bit depth support, maybe impossible
* - disable pan-on-sync, need specs
*
* BUGS
* - white margin on bootup like with tdfxfb (colormap problem?)
*
*/
#include <linux/module.h>
#include <linux/kernel.h>
#include <linux/errno.h>
#include <linux/string.h>
#include <linux/mm.h>
#include <linux/slab.h>
#include <linux/delay.h>
#include <linux/fb.h>
#include <linux/pci.h>
#include <linux/init.h>
#ifdef CONFIG_TOSHIBA
#include <linux/toshiba.h>
#endif
#include <asm/io.h>
#include <asm/irq.h>
#include <asm/pgtable.h>
#include <video/vga.h>
#include <video/neomagic.h>
#define NEOFB_VERSION "0.4.2"
/* --------------------------------------------------------------------- */
static bool internal;
static bool external;
static bool libretto;
static bool nostretch;
static bool nopciburst;
static char *mode_option = NULL;
#ifdef MODULE
MODULE_AUTHOR("(c) 2001-2002 Denis Oliver Kropp <dok@convergence.de>");
MODULE_LICENSE("GPL");
MODULE_DESCRIPTION("FBDev driver for NeoMagic PCI Chips");
module_param(internal, bool, 0);
MODULE_PARM_DESC(internal, "Enable output on internal LCD Display.");
module_param(external, bool, 0);
MODULE_PARM_DESC(external, "Enable output on external CRT.");
module_param(libretto, bool, 0);
MODULE_PARM_DESC(libretto, "Force Libretto 100/110 800x480 LCD.");
module_param(nostretch, bool, 0);
MODULE_PARM_DESC(nostretch,
"Disable stretching of modes smaller than LCD.");
module_param(nopciburst, bool, 0);
MODULE_PARM_DESC(nopciburst, "Disable PCI burst mode.");
module_param(mode_option, charp, 0);
MODULE_PARM_DESC(mode_option, "Preferred video mode ('640x480-8@60', etc)");
#endif
/* --------------------------------------------------------------------- */
static biosMode bios8[] = {
{320, 240, 0x40},
{300, 400, 0x42},
{640, 400, 0x20},
{640, 480, 0x21},
{800, 600, 0x23},
{1024, 768, 0x25},
};
static biosMode bios16[] = {
{320, 200, 0x2e},
{320, 240, 0x41},
{300, 400, 0x43},
{640, 480, 0x31},
{800, 600, 0x34},
{1024, 768, 0x37},
};
static biosMode bios24[] = {
{640, 480, 0x32},
{800, 600, 0x35},
{1024, 768, 0x38}
};
#ifdef NO_32BIT_SUPPORT_YET
/* FIXME: guessed values, wrong */
static biosMode bios32[] = {
{640, 480, 0x33},
{800, 600, 0x36},
{1024, 768, 0x39}
};
#endif
static inline void write_le32(int regindex, u32 val, const struct neofb_par *par)
{
writel(val, par->neo2200 + par->cursorOff + regindex);
}
static int neoFindMode(int xres, int yres, int depth)
{
int xres_s;
int i, size;
biosMode *mode;
switch (depth) {
case 8:
size = ARRAY_SIZE(bios8);
mode = bios8;
break;
case 16:
size = ARRAY_SIZE(bios16);
mode = bios16;
break;
case 24:
size = ARRAY_SIZE(bios24);
mode = bios24;
break;
#ifdef NO_32BIT_SUPPORT_YET
case 32:
size = ARRAY_SIZE(bios32);
mode = bios32;
break;
#endif
default:
return 0;
}
for (i = 0; i < size; i++) {
if (xres <= mode[i].x_res) {
xres_s = mode[i].x_res;
for (; i < size; i++) {
if (mode[i].x_res != xres_s)
return mode[i - 1].mode;
if (yres <= mode[i].y_res)
return mode[i].mode;
}
}
}
return mode[size - 1].mode;
}
/*
* neoCalcVCLK --
*
* Determine the closest clock frequency to the one requested.
*/
#define MAX_N 127
#define MAX_D 31
#define MAX_F 1
static void neoCalcVCLK(const struct fb_info *info,
struct neofb_par *par, long freq)
{
int n, d, f;
int n_best = 0, d_best = 0, f_best = 0;
long f_best_diff = 0x7ffff;
for (f = 0; f <= MAX_F; f++)
for (d = 0; d <= MAX_D; d++)
for (n = 0; n <= MAX_N; n++) {
long f_out;
long f_diff;
f_out = ((14318 * (n + 1)) / (d + 1)) >> f;
f_diff = abs(f_out - freq);
if (f_diff <= f_best_diff) {
f_best_diff = f_diff;
n_best = n;
d_best = d;
f_best = f;
}
if (f_out > freq)
break;
}
if (info->fix.accel == FB_ACCEL_NEOMAGIC_NM2200 ||
info->fix.accel == FB_ACCEL_NEOMAGIC_NM2230 ||
info->fix.accel == FB_ACCEL_NEOMAGIC_NM2360 ||
info->fix.accel == FB_ACCEL_NEOMAGIC_NM2380) {
/* NOT_DONE: We are trying the full range of the 2200 clock.
We should be able to try n up to 2047 */
par->VCLK3NumeratorLow = n_best;
par->VCLK3NumeratorHigh = (f_best << 7);
} else
par->VCLK3NumeratorLow = n_best | (f_best << 7);
par->VCLK3Denominator = d_best;
#ifdef NEOFB_DEBUG
printk(KERN_DEBUG "neoVCLK: f:%ld NumLow=%d NumHi=%d Den=%d Df=%ld\n",
freq,
par->VCLK3NumeratorLow,
par->VCLK3NumeratorHigh,
par->VCLK3Denominator, f_best_diff);
#endif
}
/*
* vgaHWInit --
* Handle the initialization, etc. of a screen.
* Return FALSE on failure.
*/
static int vgaHWInit(const struct fb_var_screeninfo *var,
struct neofb_par *par)
{
int hsync_end = var->xres + var->right_margin + var->hsync_len;
int htotal = (hsync_end + var->left_margin) >> 3;
int vsync_start = var->yres + var->lower_margin;
int vsync_end = vsync_start + var->vsync_len;
int vtotal = vsync_end + var->upper_margin;
par->MiscOutReg = 0x23;
if (!(var->sync & FB_SYNC_HOR_HIGH_ACT))
par->MiscOutReg |= 0x40;
if (!(var->sync & FB_SYNC_VERT_HIGH_ACT))
par->MiscOutReg |= 0x80;
/*
* Time Sequencer
*/
par->Sequencer[0] = 0x00;
par->Sequencer[1] = 0x01;
par->Sequencer[2] = 0x0F;
par->Sequencer[3] = 0x00; /* Font select */
par->Sequencer[4] = 0x0E; /* Misc */
/*
* CRTC Controller
*/
par->CRTC[0] = htotal - 5;
par->CRTC[1] = (var->xres >> 3) - 1;
par->CRTC[2] = (var->xres >> 3) - 1;
par->CRTC[3] = ((htotal - 1) & 0x1F) | 0x80;
par->CRTC[4] = ((var->xres + var->right_margin) >> 3);
par->CRTC[5] = (((htotal - 1) & 0x20) << 2)
| (((hsync_end >> 3)) & 0x1F);
par->CRTC[6] = (vtotal - 2) & 0xFF;
par->CRTC[7] = (((vtotal - 2) & 0x100) >> 8)
| (((var->yres - 1) & 0x100) >> 7)
| ((vsync_start & 0x100) >> 6)
| (((var->yres - 1) & 0x100) >> 5)
| 0x10 | (((vtotal - 2) & 0x200) >> 4)
| (((var->yres - 1) & 0x200) >> 3)
| ((vsync_start & 0x200) >> 2);
par->CRTC[8] = 0x00;
par->CRTC[9] = (((var->yres - 1) & 0x200) >> 4) | 0x40;
if (var->vmode & FB_VMODE_DOUBLE)
par->CRTC[9] |= 0x80;
par->CRTC[10] = 0x00;
par->CRTC[11] = 0x00;
par->CRTC[12] = 0x00;
par->CRTC[13] = 0x00;
par->CRTC[14] = 0x00;
par->CRTC[15] = 0x00;
par->CRTC[16] = vsync_start & 0xFF;
par->CRTC[17] = (vsync_end & 0x0F) | 0x20;
par->CRTC[18] = (var->yres - 1) & 0xFF;
par->CRTC[19] = var->xres_virtual >> 4;
par->CRTC[20] = 0x00;
par->CRTC[21] = (var->yres - 1) & 0xFF;
par->CRTC[22] = (vtotal - 1) & 0xFF;
par->CRTC[23] = 0xC3;
par->CRTC[24] = 0xFF;
/*
* are these unnecessary?
* vgaHWHBlankKGA(mode, regp, 0, KGA_FIX_OVERSCAN | KGA_ENABLE_ON_ZERO);
* vgaHWVBlankKGA(mode, regp, 0, KGA_FIX_OVERSCAN | KGA_ENABLE_ON_ZERO);
*/
/*
* Graphics Display Controller
*/
par->Graphics[0] = 0x00;
par->Graphics[1] = 0x00;
par->Graphics[2] = 0x00;
par->Graphics[3] = 0x00;
par->Graphics[4] = 0x00;
par->Graphics[5] = 0x40;
par->Graphics[6] = 0x05; /* only map 64k VGA memory !!!! */
par->Graphics[7] = 0x0F;
par->Graphics[8] = 0xFF;
par->Attribute[0] = 0x00; /* standard colormap translation */
par->Attribute[1] = 0x01;
par->Attribute[2] = 0x02;
par->Attribute[3] = 0x03;
par->Attribute[4] = 0x04;
par->Attribute[5] = 0x05;
par->Attribute[6] = 0x06;
par->Attribute[7] = 0x07;
par->Attribute[8] = 0x08;
par->Attribute[9] = 0x09;
par->Attribute[10] = 0x0A;
par->Attribute[11] = 0x0B;
par->Attribute[12] = 0x0C;
par->Attribute[13] = 0x0D;
par->Attribute[14] = 0x0E;
par->Attribute[15] = 0x0F;
par->Attribute[16] = 0x41;
par->Attribute[17] = 0xFF;
par->Attribute[18] = 0x0F;
par->Attribute[19] = 0x00;
par->Attribute[20] = 0x00;
return 0;
}
static void vgaHWLock(struct vgastate *state)
{
/* Protect CRTC[0-7] */
vga_wcrt(state->vgabase, 0x11, vga_rcrt(state->vgabase, 0x11) | 0x80);
}
static void vgaHWUnlock(void)
{
/* Unprotect CRTC[0-7] */
vga_wcrt(NULL, 0x11, vga_rcrt(NULL, 0x11) & ~0x80);
}
static void neoLock(struct vgastate *state)
{
vga_wgfx(state->vgabase, 0x09, 0x00);
vgaHWLock(state);
}
static void neoUnlock(void)
{
vgaHWUnlock();
vga_wgfx(NULL, 0x09, 0x26);
}
/*
* VGA Palette management
*/
static int paletteEnabled = 0;
static inline void VGAenablePalette(void)
{
vga_r(NULL, VGA_IS1_RC);
vga_w(NULL, VGA_ATT_W, 0x00);
paletteEnabled = 1;
}
static inline void VGAdisablePalette(void)
{
vga_r(NULL, VGA_IS1_RC);
vga_w(NULL, VGA_ATT_W, 0x20);
paletteEnabled = 0;
}
static inline void VGAwATTR(u8 index, u8 value)
{
if (paletteEnabled)
index &= ~0x20;
else
index |= 0x20;
vga_r(NULL, VGA_IS1_RC);
vga_wattr(NULL, index, value);
}
static void vgaHWProtect(int on)
{
unsigned char tmp;
tmp = vga_rseq(NULL, 0x01);
if (on) {
/*
* Turn off screen and disable sequencer.
*/
vga_wseq(NULL, 0x00, 0x01); /* Synchronous Reset */
vga_wseq(NULL, 0x01, tmp | 0x20); /* disable the display */
VGAenablePalette();
} else {
/*
* Reenable sequencer, then turn on screen.
*/
vga_wseq(NULL, 0x01, tmp & ~0x20); /* reenable display */
vga_wseq(NULL, 0x00, 0x03); /* clear synchronousreset */
VGAdisablePalette();
}
}
static void vgaHWRestore(const struct fb_info *info,
const struct neofb_par *par)
{
int i;
vga_w(NULL, VGA_MIS_W, par->MiscOutReg);
for (i = 1; i < 5; i++)
vga_wseq(NULL, i, par->Sequencer[i]);
/* Ensure CRTC registers 0-7 are unlocked by clearing bit 7 or CRTC[17] */
vga_wcrt(NULL, 17, par->CRTC[17] & ~0x80);
for (i = 0; i < 25; i++)
vga_wcrt(NULL, i, par->CRTC[i]);
for (i = 0; i < 9; i++)
vga_wgfx(NULL, i, par->Graphics[i]);
VGAenablePalette();
for (i = 0; i < 21; i++)
VGAwATTR(i, par->Attribute[i]);
VGAdisablePalette();
}
/* -------------------- Hardware specific routines ------------------------- */
/*
* Hardware Acceleration for Neo2200+
*/
static inline int neo2200_sync(struct fb_info *info)
{
struct neofb_par *par = info->par;
while (readl(&par->neo2200->bltStat) & 1)
cpu_relax();
return 0;
}
static inline void neo2200_wait_fifo(struct fb_info *info,
int requested_fifo_space)
{
// ndev->neo.waitfifo_calls++;
// ndev->neo.waitfifo_sum += requested_fifo_space;
/* FIXME: does not work
if (neo_fifo_space < requested_fifo_space)
{
neo_fifo_waitcycles++;
while (1)
{
neo_fifo_space = (neo2200->bltStat >> 8);
if (neo_fifo_space >= requested_fifo_space)
break;
}
}
else
{
neo_fifo_cache_hits++;
}
neo_fifo_space -= requested_fifo_space;
*/
neo2200_sync(info);
}
static inline void neo2200_accel_init(struct fb_info *info,
struct fb_var_screeninfo *var)
{
struct neofb_par *par = info->par;
Neo2200 __iomem *neo2200 = par->neo2200;
u32 bltMod, pitch;
neo2200_sync(info);
switch (var->bits_per_pixel) {
case 8:
bltMod = NEO_MODE1_DEPTH8;
pitch = var->xres_virtual;
break;
case 15:
case 16:
bltMod = NEO_MODE1_DEPTH16;
pitch = var->xres_virtual * 2;
break;
case 24:
bltMod = NEO_MODE1_DEPTH24;
pitch = var->xres_virtual * 3;
break;
default:
printk(KERN_ERR
"neofb: neo2200_accel_init: unexpected bits per pixel!\n");
return;
}
writel(bltMod << 16, &neo2200->bltStat);
writel((pitch << 16) | pitch, &neo2200->pitch);
}
/* --------------------------------------------------------------------- */
static int
neofb_open(struct fb_info *info, int user)
{
struct neofb_par *par = info->par;
if (!par->ref_count) {
memset(&par->state, 0, sizeof(struct vgastate));
par->state.flags = VGA_SAVE_MODE | VGA_SAVE_FONTS;
save_vga(&par->state);
}
par->ref_count++;
return 0;
}
static int
neofb_release(struct fb_info *info, int user)
{
struct neofb_par *par = info->par;
if (!par->ref_count)
return -EINVAL;
if (par->ref_count == 1) {
restore_vga(&par->state);
}
par->ref_count--;
return 0;
}
static int
neofb_check_var(struct fb_var_screeninfo *var, struct fb_info *info)
{
struct neofb_par *par = info->par;
int memlen, vramlen;
int mode_ok = 0;
DBG("neofb_check_var");
if (PICOS2KHZ(var->pixclock) > par->maxClock)
return -EINVAL;
/* Is the mode larger than the LCD panel? */
if (par->internal_display &&
((var->xres > par->NeoPanelWidth) ||
(var->yres > par->NeoPanelHeight))) {
printk(KERN_INFO
"Mode (%dx%d) larger than the LCD panel (%dx%d)\n",
var->xres, var->yres, par->NeoPanelWidth,
par->NeoPanelHeight);
return -EINVAL;
}
/* Is the mode one of the acceptable sizes? */
if (!par->internal_display)
mode_ok = 1;
else {
switch (var->xres) {
case 1280:
if (var->yres == 1024)
mode_ok = 1;
break;
case 1024:
if (var->yres == 768)
mode_ok = 1;
break;
case 800:
if (var->yres == (par->libretto ? 480 : 600))
mode_ok = 1;
break;
case 640:
if (var->yres == 480)
mode_ok = 1;
break;
}
}
if (!mode_ok) {
printk(KERN_INFO
"Mode (%dx%d) won't display properly on LCD\n",
var->xres, var->yres);
return -EINVAL;
}
var->red.msb_right = 0;
var->green.msb_right = 0;
var->blue.msb_right = 0;
var->transp.msb_right = 0;
var->transp.offset = 0;
var->transp.length = 0;
switch (var->bits_per_pixel) {
case 8: /* PSEUDOCOLOUR, 256 */
var->red.offset = 0;
var->red.length = 8;
var->green.offset = 0;
var->green.length = 8;
var->blue.offset = 0;
var->blue.length = 8;
break;
case 16: /* DIRECTCOLOUR, 64k */
var->red.offset = 11;
var->red.length = 5;
var->green.offset = 5;
var->green.length = 6;
var->blue.offset = 0;
var->blue.length = 5;
break;
case 24: /* TRUECOLOUR, 16m */
var->red.offset = 16;
var->red.length = 8;
var->green.offset = 8;
var->green.length = 8;
var->blue.offset = 0;
var->blue.length = 8;
break;
#ifdef NO_32BIT_SUPPORT_YET
case 32: /* TRUECOLOUR, 16m */
var->transp.offset = 24;
var->transp.length = 8;
var->red.offset = 16;
var->red.length = 8;
var->green.offset = 8;
var->green.length = 8;
var->blue.offset = 0;
var->blue.length = 8;
break;
#endif
default:
printk(KERN_WARNING "neofb: no support for %dbpp\n",
var->bits_per_pixel);
return -EINVAL;
}
vramlen = info->fix.smem_len;
if (vramlen > 4 * 1024 * 1024)
vramlen = 4 * 1024 * 1024;
if (var->xres_virtual < var->xres)
var->xres_virtual = var->xres;
memlen = var->xres_virtual * var->bits_per_pixel * var->yres_virtual >> 3;
if (memlen > vramlen) {
var->yres_virtual = vramlen * 8 / (var->xres_virtual *
var->bits_per_pixel);
memlen = var->xres_virtual * var->bits_per_pixel *
var->yres_virtual / 8;
}
/* we must round yres/xres down, we already rounded y/xres_virtual up
if it was possible. We should return -EINVAL, but I disagree */
if (var->yres_virtual < var->yres)
var->yres = var->yres_virtual;
if (var->xoffset + var->xres > var->xres_virtual)
var->xoffset = var->xres_virtual - var->xres;
if (var->yoffset + var->yres > var->yres_virtual)
var->yoffset = var->yres_virtual - var->yres;
var->nonstd = 0;
var->height = -1;
var->width = -1;
if (var->bits_per_pixel >= 24 || !par->neo2200)
var->accel_flags &= ~FB_ACCELF_TEXT;
return 0;
}
static int neofb_set_par(struct fb_info *info)
{
struct neofb_par *par = info->par;
unsigned char temp;
int i, clock_hi = 0;
int lcd_stretch;
int hoffset, voffset;
int vsync_start, vtotal;
DBG("neofb_set_par");
neoUnlock();
vgaHWProtect(1); /* Blank the screen */
vsync_start = info->var.yres + info->var.lower_margin;
vtotal = vsync_start + info->var.vsync_len + info->var.upper_margin;
/*
* This will allocate the datastructure and initialize all of the
* generic VGA registers.
*/
if (vgaHWInit(&info->var, par))
return -EINVAL;
/*
* The default value assigned by vgaHW.c is 0x41, but this does
* not work for NeoMagic.
*/
par->Attribute[16] = 0x01;
switch (info->var.bits_per_pixel) {
case 8:
par->CRTC[0x13] = info->var.xres_virtual >> 3;
par->ExtCRTOffset = info->var.xres_virtual >> 11;
par->ExtColorModeSelect = 0x11;
break;
case 16:
par->CRTC[0x13] = info->var.xres_virtual >> 2;
par->ExtCRTOffset = info->var.xres_virtual >> 10;
par->ExtColorModeSelect = 0x13;
break;
case 24:
par->CRTC[0x13] = (info->var.xres_virtual * 3) >> 3;
par->ExtCRTOffset = (info->var.xres_virtual * 3) >> 11;
par->ExtColorModeSelect = 0x14;
break;
#ifdef NO_32BIT_SUPPORT_YET
case 32: /* FIXME: guessed values */
par->CRTC[0x13] = info->var.xres_virtual >> 1;
par->ExtCRTOffset = info->var.xres_virtual >> 9;
par->ExtColorModeSelect = 0x15;
break;
#endif
default:
break;
}
par->ExtCRTDispAddr = 0x10;
/* Vertical Extension */
par->VerticalExt = (((vtotal - 2) & 0x400) >> 10)
| (((info->var.yres - 1) & 0x400) >> 9)
| (((vsync_start) & 0x400) >> 8)
| (((vsync_start) & 0x400) >> 7);
/* Fast write bursts on unless disabled. */
if (par->pci_burst)
par->SysIfaceCntl1 = 0x30;
else
par->SysIfaceCntl1 = 0x00;
par->SysIfaceCntl2 = 0xc0; /* VESA Bios sets this to 0x80! */
[PATCH] neofb: avoid resetting display config on unblank (v2) There were two mistakes in the register-read-on-(un)blank approach. - First, without proper register (un)locking the value read back will always be zero, and this is what I missed entirely until just now. Due to this, the logic could not be verified at all and I tried some bogus checks which are completely stupid. - Second, the LCD status bit will always be set to zero when the backlight has been turned off. Reading the value back during unblank will disable the LCD unconditionally, regardless of the state it is supposed to be in, since we set it to zero beforehand. So this is what we do now: - create a new variable in struct neofb_par, and use that to determine whether to read back registers (initialized to true) - before actually blanking the screen, read back the register to sense any possible change made through Fn key combo - use proper neoUnlock() / neoLock() to actually read something - every call to neofb_blank() determines if we read back next time: blanking disables readback, unblanking (FB_BLANK_UNBLANK) enables it This should give us a nice and clean state machine. Has been thoroughly tested on a Dell Latitude CPiA / NM220 Chip docked to a C/Dock2 with attached CRT in all possible combinations of LCD/CRT on/off. I changed the config via Fn key, let the console blank, unblanked by keypress - works flawlessly. Signed-off-by: Christian Trefzer <ctrefzer@gmx.de> Cc: "Antonino A. Daplas" <adaplas@pol.net> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-02-16 07:17:34 +08:00
/* Initialize: by default, we want display config register to be read */
par->PanelDispCntlRegRead = 1;
/* Enable any user specified display devices. */
par->PanelDispCntlReg1 = 0x00;
if (par->internal_display)
par->PanelDispCntlReg1 |= 0x02;
if (par->external_display)
par->PanelDispCntlReg1 |= 0x01;
/* If the user did not specify any display devices, then... */
if (par->PanelDispCntlReg1 == 0x00) {
/* Default to internal (i.e., LCD) only. */
par->PanelDispCntlReg1 = vga_rgfx(NULL, 0x20) & 0x03;
}
/* If we are using a fixed mode, then tell the chip we are. */
switch (info->var.xres) {
case 1280:
par->PanelDispCntlReg1 |= 0x60;
break;
case 1024:
par->PanelDispCntlReg1 |= 0x40;
break;
case 800:
par->PanelDispCntlReg1 |= 0x20;
break;
case 640:
default:
break;
}
/* Setup shadow register locking. */
switch (par->PanelDispCntlReg1 & 0x03) {
case 0x01: /* External CRT only mode: */
par->GeneralLockReg = 0x00;
/* We need to program the VCLK for external display only mode. */
par->ProgramVCLK = 1;
break;
case 0x02: /* Internal LCD only mode: */
case 0x03: /* Simultaneous internal/external (LCD/CRT) mode: */
par->GeneralLockReg = 0x01;
/* Don't program the VCLK when using the LCD. */
par->ProgramVCLK = 0;
break;
}
/*
* If the screen is to be stretched, turn on stretching for the
* various modes.
*
* OPTION_LCD_STRETCH means stretching should be turned off!
*/
par->PanelDispCntlReg2 = 0x00;
par->PanelDispCntlReg3 = 0x00;
if (par->lcd_stretch && (par->PanelDispCntlReg1 == 0x02) && /* LCD only */
(info->var.xres != par->NeoPanelWidth)) {
switch (info->var.xres) {
case 320: /* Needs testing. KEM -- 24 May 98 */
case 400: /* Needs testing. KEM -- 24 May 98 */
case 640:
case 800:
case 1024:
lcd_stretch = 1;
par->PanelDispCntlReg2 |= 0xC6;
break;
default:
lcd_stretch = 0;
/* No stretching in these modes. */
}
} else
lcd_stretch = 0;
/*
* If the screen is to be centerd, turn on the centering for the
* various modes.
*/
par->PanelVertCenterReg1 = 0x00;
par->PanelVertCenterReg2 = 0x00;
par->PanelVertCenterReg3 = 0x00;
par->PanelVertCenterReg4 = 0x00;
par->PanelVertCenterReg5 = 0x00;
par->PanelHorizCenterReg1 = 0x00;
par->PanelHorizCenterReg2 = 0x00;
par->PanelHorizCenterReg3 = 0x00;
par->PanelHorizCenterReg4 = 0x00;
par->PanelHorizCenterReg5 = 0x00;
if (par->PanelDispCntlReg1 & 0x02) {
if (info->var.xres == par->NeoPanelWidth) {
/*
* No centering required when the requested display width
* equals the panel width.
*/
} else {
par->PanelDispCntlReg2 |= 0x01;
par->PanelDispCntlReg3 |= 0x10;
/* Calculate the horizontal and vertical offsets. */
if (!lcd_stretch) {
hoffset =
((par->NeoPanelWidth -
info->var.xres) >> 4) - 1;
voffset =
((par->NeoPanelHeight -
info->var.yres) >> 1) - 2;
} else {
/* Stretched modes cannot be centered. */
hoffset = 0;
voffset = 0;
}
switch (info->var.xres) {
case 320: /* Needs testing. KEM -- 24 May 98 */
par->PanelHorizCenterReg3 = hoffset;
par->PanelVertCenterReg2 = voffset;
break;
case 400: /* Needs testing. KEM -- 24 May 98 */
par->PanelHorizCenterReg4 = hoffset;
par->PanelVertCenterReg1 = voffset;
break;
case 640:
par->PanelHorizCenterReg1 = hoffset;
par->PanelVertCenterReg3 = voffset;
break;
case 800:
par->PanelHorizCenterReg2 = hoffset;
par->PanelVertCenterReg4 = voffset;
break;
case 1024:
par->PanelHorizCenterReg5 = hoffset;
par->PanelVertCenterReg5 = voffset;
break;
case 1280:
default:
/* No centering in these modes. */
break;
}
}
}
par->biosMode =
neoFindMode(info->var.xres, info->var.yres,
info->var.bits_per_pixel);
/*
* Calculate the VCLK that most closely matches the requested dot
* clock.
*/
neoCalcVCLK(info, par, PICOS2KHZ(info->var.pixclock));
/* Since we program the clocks ourselves, always use VCLK3. */
par->MiscOutReg |= 0x0C;
/* alread unlocked above */
/* BOGUS vga_wgfx(NULL, 0x09, 0x26); */
/* don't know what this is, but it's 0 from bootup anyway */
vga_wgfx(NULL, 0x15, 0x00);
/* was set to 0x01 by my bios in text and vesa modes */
vga_wgfx(NULL, 0x0A, par->GeneralLockReg);
/*
* The color mode needs to be set before calling vgaHWRestore
* to ensure the DAC is initialized properly.
*
* NOTE: Make sure we don't change bits make sure we don't change
* any reserved bits.
*/
temp = vga_rgfx(NULL, 0x90);
switch (info->fix.accel) {
case FB_ACCEL_NEOMAGIC_NM2070:
temp &= 0xF0; /* Save bits 7:4 */
temp |= (par->ExtColorModeSelect & ~0xF0);
break;
case FB_ACCEL_NEOMAGIC_NM2090:
case FB_ACCEL_NEOMAGIC_NM2093:
case FB_ACCEL_NEOMAGIC_NM2097:
case FB_ACCEL_NEOMAGIC_NM2160:
case FB_ACCEL_NEOMAGIC_NM2200:
case FB_ACCEL_NEOMAGIC_NM2230:
case FB_ACCEL_NEOMAGIC_NM2360:
case FB_ACCEL_NEOMAGIC_NM2380:
temp &= 0x70; /* Save bits 6:4 */
temp |= (par->ExtColorModeSelect & ~0x70);
break;
}
vga_wgfx(NULL, 0x90, temp);
/*
* In some rare cases a lockup might occur if we don't delay
* here. (Reported by Miles Lane)
*/
//mdelay(200);
/*
* Disable horizontal and vertical graphics and text expansions so
* that vgaHWRestore works properly.
*/
temp = vga_rgfx(NULL, 0x25);
temp &= 0x39;
vga_wgfx(NULL, 0x25, temp);
/*
* Sleep for 200ms to make sure that the two operations above have
* had time to take effect.
*/
mdelay(200);
/*
* This function handles restoring the generic VGA registers. */
vgaHWRestore(info, par);
/* linear colormap for non palettized modes */
switch (info->var.bits_per_pixel) {
case 8:
/* PseudoColor, 256 */
info->fix.visual = FB_VISUAL_PSEUDOCOLOR;
break;
case 16:
/* TrueColor, 64k */
info->fix.visual = FB_VISUAL_TRUECOLOR;
for (i = 0; i < 64; i++) {
outb(i, 0x3c8);
outb(i << 1, 0x3c9);
outb(i, 0x3c9);
outb(i << 1, 0x3c9);
}
break;
case 24:
#ifdef NO_32BIT_SUPPORT_YET
case 32:
#endif
/* TrueColor, 16m */
info->fix.visual = FB_VISUAL_TRUECOLOR;
for (i = 0; i < 256; i++) {
outb(i, 0x3c8);
outb(i, 0x3c9);
outb(i, 0x3c9);
outb(i, 0x3c9);
}
break;
}
vga_wgfx(NULL, 0x0E, par->ExtCRTDispAddr);
vga_wgfx(NULL, 0x0F, par->ExtCRTOffset);
temp = vga_rgfx(NULL, 0x10);
temp &= 0x0F; /* Save bits 3:0 */
temp |= (par->SysIfaceCntl1 & ~0x0F); /* VESA Bios sets bit 1! */
vga_wgfx(NULL, 0x10, temp);
vga_wgfx(NULL, 0x11, par->SysIfaceCntl2);
vga_wgfx(NULL, 0x15, 0 /*par->SingleAddrPage */ );
vga_wgfx(NULL, 0x16, 0 /*par->DualAddrPage */ );
temp = vga_rgfx(NULL, 0x20);
switch (info->fix.accel) {
case FB_ACCEL_NEOMAGIC_NM2070:
temp &= 0xFC; /* Save bits 7:2 */
temp |= (par->PanelDispCntlReg1 & ~0xFC);
break;
case FB_ACCEL_NEOMAGIC_NM2090:
case FB_ACCEL_NEOMAGIC_NM2093:
case FB_ACCEL_NEOMAGIC_NM2097:
case FB_ACCEL_NEOMAGIC_NM2160:
temp &= 0xDC; /* Save bits 7:6,4:2 */
temp |= (par->PanelDispCntlReg1 & ~0xDC);
break;
case FB_ACCEL_NEOMAGIC_NM2200:
case FB_ACCEL_NEOMAGIC_NM2230:
case FB_ACCEL_NEOMAGIC_NM2360:
case FB_ACCEL_NEOMAGIC_NM2380:
temp &= 0x98; /* Save bits 7,4:3 */
temp |= (par->PanelDispCntlReg1 & ~0x98);
break;
}
vga_wgfx(NULL, 0x20, temp);
temp = vga_rgfx(NULL, 0x25);
temp &= 0x38; /* Save bits 5:3 */
temp |= (par->PanelDispCntlReg2 & ~0x38);
vga_wgfx(NULL, 0x25, temp);
if (info->fix.accel != FB_ACCEL_NEOMAGIC_NM2070) {
temp = vga_rgfx(NULL, 0x30);
temp &= 0xEF; /* Save bits 7:5 and bits 3:0 */
temp |= (par->PanelDispCntlReg3 & ~0xEF);
vga_wgfx(NULL, 0x30, temp);
}
vga_wgfx(NULL, 0x28, par->PanelVertCenterReg1);
vga_wgfx(NULL, 0x29, par->PanelVertCenterReg2);
vga_wgfx(NULL, 0x2a, par->PanelVertCenterReg3);
if (info->fix.accel != FB_ACCEL_NEOMAGIC_NM2070) {
vga_wgfx(NULL, 0x32, par->PanelVertCenterReg4);
vga_wgfx(NULL, 0x33, par->PanelHorizCenterReg1);
vga_wgfx(NULL, 0x34, par->PanelHorizCenterReg2);
vga_wgfx(NULL, 0x35, par->PanelHorizCenterReg3);
}
if (info->fix.accel == FB_ACCEL_NEOMAGIC_NM2160)
vga_wgfx(NULL, 0x36, par->PanelHorizCenterReg4);
if (info->fix.accel == FB_ACCEL_NEOMAGIC_NM2200 ||
info->fix.accel == FB_ACCEL_NEOMAGIC_NM2230 ||
info->fix.accel == FB_ACCEL_NEOMAGIC_NM2360 ||
info->fix.accel == FB_ACCEL_NEOMAGIC_NM2380) {
vga_wgfx(NULL, 0x36, par->PanelHorizCenterReg4);
vga_wgfx(NULL, 0x37, par->PanelVertCenterReg5);
vga_wgfx(NULL, 0x38, par->PanelHorizCenterReg5);
clock_hi = 1;
}
/* Program VCLK3 if needed. */
if (par->ProgramVCLK && ((vga_rgfx(NULL, 0x9B) != par->VCLK3NumeratorLow)
|| (vga_rgfx(NULL, 0x9F) != par->VCLK3Denominator)
|| (clock_hi && ((vga_rgfx(NULL, 0x8F) & ~0x0f)
!= (par->VCLK3NumeratorHigh &
~0x0F))))) {
vga_wgfx(NULL, 0x9B, par->VCLK3NumeratorLow);
if (clock_hi) {
temp = vga_rgfx(NULL, 0x8F);
temp &= 0x0F; /* Save bits 3:0 */
temp |= (par->VCLK3NumeratorHigh & ~0x0F);
vga_wgfx(NULL, 0x8F, temp);
}
vga_wgfx(NULL, 0x9F, par->VCLK3Denominator);
}
if (par->biosMode)
vga_wcrt(NULL, 0x23, par->biosMode);
vga_wgfx(NULL, 0x93, 0xc0); /* Gives 5x faster framebuffer writes !!! */
/* Program vertical extension register */
if (info->fix.accel == FB_ACCEL_NEOMAGIC_NM2200 ||
info->fix.accel == FB_ACCEL_NEOMAGIC_NM2230 ||
info->fix.accel == FB_ACCEL_NEOMAGIC_NM2360 ||
info->fix.accel == FB_ACCEL_NEOMAGIC_NM2380) {
vga_wcrt(NULL, 0x70, par->VerticalExt);
}
vgaHWProtect(0); /* Turn on screen */
/* Calling this also locks offset registers required in update_start */
neoLock(&par->state);
info->fix.line_length =
info->var.xres_virtual * (info->var.bits_per_pixel >> 3);
switch (info->fix.accel) {
case FB_ACCEL_NEOMAGIC_NM2200:
case FB_ACCEL_NEOMAGIC_NM2230:
case FB_ACCEL_NEOMAGIC_NM2360:
case FB_ACCEL_NEOMAGIC_NM2380:
neo2200_accel_init(info, &info->var);
break;
default:
break;
}
return 0;
}
/*
* Pan or Wrap the Display
*/
static int neofb_pan_display(struct fb_var_screeninfo *var,
struct fb_info *info)
{
struct neofb_par *par = info->par;
struct vgastate *state = &par->state;
int oldExtCRTDispAddr;
int Base;
DBG("neofb_update_start");
Base = (var->yoffset * info->var.xres_virtual + var->xoffset) >> 2;
Base *= (info->var.bits_per_pixel + 7) / 8;
neoUnlock();
/*
* These are the generic starting address registers.
*/
vga_wcrt(state->vgabase, 0x0C, (Base & 0x00FF00) >> 8);
vga_wcrt(state->vgabase, 0x0D, (Base & 0x00FF));
/*
* Make sure we don't clobber some other bits that might already
* have been set. NOTE: NM2200 has a writable bit 3, but it shouldn't
* be needed.
*/
oldExtCRTDispAddr = vga_rgfx(NULL, 0x0E);
vga_wgfx(state->vgabase, 0x0E, (((Base >> 16) & 0x0f) | (oldExtCRTDispAddr & 0xf0)));
neoLock(state);
return 0;
}
static int neofb_setcolreg(u_int regno, u_int red, u_int green, u_int blue,
u_int transp, struct fb_info *fb)
{
if (regno >= fb->cmap.len || regno > 255)
return -EINVAL;
if (fb->var.bits_per_pixel <= 8) {
outb(regno, 0x3c8);
outb(red >> 10, 0x3c9);
outb(green >> 10, 0x3c9);
outb(blue >> 10, 0x3c9);
} else if (regno < 16) {
switch (fb->var.bits_per_pixel) {
case 16:
((u32 *) fb->pseudo_palette)[regno] =
((red & 0xf800)) | ((green & 0xfc00) >> 5) |
((blue & 0xf800) >> 11);
break;
case 24:
((u32 *) fb->pseudo_palette)[regno] =
((red & 0xff00) << 8) | ((green & 0xff00)) |
((blue & 0xff00) >> 8);
break;
#ifdef NO_32BIT_SUPPORT_YET
case 32:
((u32 *) fb->pseudo_palette)[regno] =
((transp & 0xff00) << 16) | ((red & 0xff00) << 8) |
((green & 0xff00)) | ((blue & 0xff00) >> 8);
break;
#endif
default:
return 1;
}
}
return 0;
}
/*
* (Un)Blank the display.
*/
static int neofb_blank(int blank_mode, struct fb_info *info)
{
/*
* Blank the screen if blank_mode != 0, else unblank.
* Return 0 if blanking succeeded, != 0 if un-/blanking failed due to
* e.g. a video mode which doesn't support it. Implements VESA suspend
* and powerdown modes for monitors, and backlight control on LCDs.
* blank_mode == 0: unblanked (backlight on)
* blank_mode == 1: blank (backlight on)
* blank_mode == 2: suspend vsync (backlight off)
* blank_mode == 3: suspend hsync (backlight off)
* blank_mode == 4: powerdown (backlight off)
*
* wms...Enable VESA DPMS compatible powerdown mode
* run "setterm -powersave powerdown" to take advantage
*/
struct neofb_par *par = info->par;
int seqflags, lcdflags, dpmsflags, reg, tmpdisp;
[PATCH] neofb: avoid resetting display config on unblank (v2) There were two mistakes in the register-read-on-(un)blank approach. - First, without proper register (un)locking the value read back will always be zero, and this is what I missed entirely until just now. Due to this, the logic could not be verified at all and I tried some bogus checks which are completely stupid. - Second, the LCD status bit will always be set to zero when the backlight has been turned off. Reading the value back during unblank will disable the LCD unconditionally, regardless of the state it is supposed to be in, since we set it to zero beforehand. So this is what we do now: - create a new variable in struct neofb_par, and use that to determine whether to read back registers (initialized to true) - before actually blanking the screen, read back the register to sense any possible change made through Fn key combo - use proper neoUnlock() / neoLock() to actually read something - every call to neofb_blank() determines if we read back next time: blanking disables readback, unblanking (FB_BLANK_UNBLANK) enables it This should give us a nice and clean state machine. Has been thoroughly tested on a Dell Latitude CPiA / NM220 Chip docked to a C/Dock2 with attached CRT in all possible combinations of LCD/CRT on/off. I changed the config via Fn key, let the console blank, unblanked by keypress - works flawlessly. Signed-off-by: Christian Trefzer <ctrefzer@gmx.de> Cc: "Antonino A. Daplas" <adaplas@pol.net> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-02-16 07:17:34 +08:00
/*
* Read back the register bits related to display configuration. They might
* have been changed underneath the driver via Fn key stroke.
*/
neoUnlock();
tmpdisp = vga_rgfx(NULL, 0x20) & 0x03;
neoLock(&par->state);
/* In case we blank the screen, we want to store the possibly new
* configuration in the driver. During un-blank, we re-apply this setting,
* since the LCD bit will be cleared in order to switch off the backlight.
*/
[PATCH] neofb: avoid resetting display config on unblank (v2) There were two mistakes in the register-read-on-(un)blank approach. - First, without proper register (un)locking the value read back will always be zero, and this is what I missed entirely until just now. Due to this, the logic could not be verified at all and I tried some bogus checks which are completely stupid. - Second, the LCD status bit will always be set to zero when the backlight has been turned off. Reading the value back during unblank will disable the LCD unconditionally, regardless of the state it is supposed to be in, since we set it to zero beforehand. So this is what we do now: - create a new variable in struct neofb_par, and use that to determine whether to read back registers (initialized to true) - before actually blanking the screen, read back the register to sense any possible change made through Fn key combo - use proper neoUnlock() / neoLock() to actually read something - every call to neofb_blank() determines if we read back next time: blanking disables readback, unblanking (FB_BLANK_UNBLANK) enables it This should give us a nice and clean state machine. Has been thoroughly tested on a Dell Latitude CPiA / NM220 Chip docked to a C/Dock2 with attached CRT in all possible combinations of LCD/CRT on/off. I changed the config via Fn key, let the console blank, unblanked by keypress - works flawlessly. Signed-off-by: Christian Trefzer <ctrefzer@gmx.de> Cc: "Antonino A. Daplas" <adaplas@pol.net> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-02-16 07:17:34 +08:00
if (par->PanelDispCntlRegRead) {
par->PanelDispCntlReg1 = tmpdisp;
[PATCH] neofb: avoid resetting display config on unblank (v2) There were two mistakes in the register-read-on-(un)blank approach. - First, without proper register (un)locking the value read back will always be zero, and this is what I missed entirely until just now. Due to this, the logic could not be verified at all and I tried some bogus checks which are completely stupid. - Second, the LCD status bit will always be set to zero when the backlight has been turned off. Reading the value back during unblank will disable the LCD unconditionally, regardless of the state it is supposed to be in, since we set it to zero beforehand. So this is what we do now: - create a new variable in struct neofb_par, and use that to determine whether to read back registers (initialized to true) - before actually blanking the screen, read back the register to sense any possible change made through Fn key combo - use proper neoUnlock() / neoLock() to actually read something - every call to neofb_blank() determines if we read back next time: blanking disables readback, unblanking (FB_BLANK_UNBLANK) enables it This should give us a nice and clean state machine. Has been thoroughly tested on a Dell Latitude CPiA / NM220 Chip docked to a C/Dock2 with attached CRT in all possible combinations of LCD/CRT on/off. I changed the config via Fn key, let the console blank, unblanked by keypress - works flawlessly. Signed-off-by: Christian Trefzer <ctrefzer@gmx.de> Cc: "Antonino A. Daplas" <adaplas@pol.net> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-02-16 07:17:34 +08:00
}
par->PanelDispCntlRegRead = !blank_mode;
switch (blank_mode) {
case FB_BLANK_POWERDOWN: /* powerdown - both sync lines down */
seqflags = VGA_SR01_SCREEN_OFF; /* Disable sequencer */
lcdflags = 0; /* LCD off */
dpmsflags = NEO_GR01_SUPPRESS_HSYNC |
NEO_GR01_SUPPRESS_VSYNC;
#ifdef CONFIG_TOSHIBA
/* Do we still need this ? */
/* attempt to turn off backlight on toshiba; also turns off external */
{
SMMRegisters regs;
regs.eax = 0xff00; /* HCI_SET */
regs.ebx = 0x0002; /* HCI_BACKLIGHT */
regs.ecx = 0x0000; /* HCI_DISABLE */
tosh_smm(&regs);
}
#endif
break;
case FB_BLANK_HSYNC_SUSPEND: /* hsync off */
seqflags = VGA_SR01_SCREEN_OFF; /* Disable sequencer */
lcdflags = 0; /* LCD off */
dpmsflags = NEO_GR01_SUPPRESS_HSYNC;
break;
case FB_BLANK_VSYNC_SUSPEND: /* vsync off */
seqflags = VGA_SR01_SCREEN_OFF; /* Disable sequencer */
lcdflags = 0; /* LCD off */
dpmsflags = NEO_GR01_SUPPRESS_VSYNC;
break;
case FB_BLANK_NORMAL: /* just blank screen (backlight stays on) */
seqflags = VGA_SR01_SCREEN_OFF; /* Disable sequencer */
/*
* During a blank operation with the LID shut, we might store "LCD off"
* by mistake. Due to timing issues, the BIOS may switch the lights
* back on, and we turn it back off once we "unblank".
*
* So here is an attempt to implement ">=" - if we are in the process
* of unblanking, and the LCD bit is unset in the driver but set in the
* register, we must keep it.
*/
lcdflags = ((par->PanelDispCntlReg1 | tmpdisp) & 0x02); /* LCD normal */
dpmsflags = 0x00; /* no hsync/vsync suppression */
break;
case FB_BLANK_UNBLANK: /* unblank */
seqflags = 0; /* Enable sequencer */
lcdflags = ((par->PanelDispCntlReg1 | tmpdisp) & 0x02); /* LCD normal */
dpmsflags = 0x00; /* no hsync/vsync suppression */
#ifdef CONFIG_TOSHIBA
/* Do we still need this ? */
/* attempt to re-enable backlight/external on toshiba */
{
SMMRegisters regs;
regs.eax = 0xff00; /* HCI_SET */
regs.ebx = 0x0002; /* HCI_BACKLIGHT */
regs.ecx = 0x0001; /* HCI_ENABLE */
tosh_smm(&regs);
}
#endif
break;
default: /* Anything else we don't understand; return 1 to tell
* fb_blank we didn't aactually do anything */
return 1;
}
neoUnlock();
reg = (vga_rseq(NULL, 0x01) & ~0x20) | seqflags;
vga_wseq(NULL, 0x01, reg);
reg = (vga_rgfx(NULL, 0x20) & ~0x02) | lcdflags;
vga_wgfx(NULL, 0x20, reg);
reg = (vga_rgfx(NULL, 0x01) & ~0xF0) | 0x80 | dpmsflags;
vga_wgfx(NULL, 0x01, reg);
neoLock(&par->state);
return 0;
}
static void
neo2200_fillrect(struct fb_info *info, const struct fb_fillrect *rect)
{
struct neofb_par *par = info->par;
u_long dst, rop;
dst = rect->dx + rect->dy * info->var.xres_virtual;
rop = rect->rop ? 0x060000 : 0x0c0000;
neo2200_wait_fifo(info, 4);
/* set blt control */
writel(NEO_BC3_FIFO_EN |
NEO_BC0_SRC_IS_FG | NEO_BC3_SKIP_MAPPING |
// NEO_BC3_DST_XY_ADDR |
// NEO_BC3_SRC_XY_ADDR |
rop, &par->neo2200->bltCntl);
switch (info->var.bits_per_pixel) {
case 8:
writel(rect->color, &par->neo2200->fgColor);
break;
case 16:
case 24:
writel(((u32 *) (info->pseudo_palette))[rect->color],
&par->neo2200->fgColor);
break;
}
writel(dst * ((info->var.bits_per_pixel + 7) >> 3),
&par->neo2200->dstStart);
writel((rect->height << 16) | (rect->width & 0xffff),
&par->neo2200->xyExt);
}
static void
neo2200_copyarea(struct fb_info *info, const struct fb_copyarea *area)
{
u32 sx = area->sx, sy = area->sy, dx = area->dx, dy = area->dy;
struct neofb_par *par = info->par;
u_long src, dst, bltCntl;
bltCntl = NEO_BC3_FIFO_EN | NEO_BC3_SKIP_MAPPING | 0x0C0000;
if ((dy > sy) || ((dy == sy) && (dx > sx))) {
/* Start with the lower right corner */
sy += (area->height - 1);
dy += (area->height - 1);
sx += (area->width - 1);
dx += (area->width - 1);
bltCntl |= NEO_BC0_X_DEC | NEO_BC0_DST_Y_DEC | NEO_BC0_SRC_Y_DEC;
}
src = sx * (info->var.bits_per_pixel >> 3) + sy*info->fix.line_length;
dst = dx * (info->var.bits_per_pixel >> 3) + dy*info->fix.line_length;
neo2200_wait_fifo(info, 4);
/* set blt control */
writel(bltCntl, &par->neo2200->bltCntl);
writel(src, &par->neo2200->srcStart);
writel(dst, &par->neo2200->dstStart);
writel((area->height << 16) | (area->width & 0xffff),
&par->neo2200->xyExt);
}
static void
neo2200_imageblit(struct fb_info *info, const struct fb_image *image)
{
struct neofb_par *par = info->par;
int s_pitch = (image->width * image->depth + 7) >> 3;
int scan_align = info->pixmap.scan_align - 1;
int buf_align = info->pixmap.buf_align - 1;
int bltCntl_flags, d_pitch, data_len;
// The data is padded for the hardware
d_pitch = (s_pitch + scan_align) & ~scan_align;
data_len = ((d_pitch * image->height) + buf_align) & ~buf_align;
neo2200_sync(info);
if (image->depth == 1) {
if (info->var.bits_per_pixel == 24 && image->width < 16) {
/* FIXME. There is a bug with accelerated color-expanded
* transfers in 24 bit mode if the image being transferred
* is less than 16 bits wide. This is due to insufficient
* padding when writing the image. We need to adjust
* struct fb_pixmap. Not yet done. */
cfb_imageblit(info, image);
return;
}
bltCntl_flags = NEO_BC0_SRC_MONO;
} else if (image->depth == info->var.bits_per_pixel) {
bltCntl_flags = 0;
} else {
/* We don't currently support hardware acceleration if image
* depth is different from display */
cfb_imageblit(info, image);
return;
}
switch (info->var.bits_per_pixel) {
case 8:
writel(image->fg_color, &par->neo2200->fgColor);
writel(image->bg_color, &par->neo2200->bgColor);
break;
case 16:
case 24:
writel(((u32 *) (info->pseudo_palette))[image->fg_color],
&par->neo2200->fgColor);
writel(((u32 *) (info->pseudo_palette))[image->bg_color],
&par->neo2200->bgColor);
break;
}
writel(NEO_BC0_SYS_TO_VID |
NEO_BC3_SKIP_MAPPING | bltCntl_flags |
// NEO_BC3_DST_XY_ADDR |
0x0c0000, &par->neo2200->bltCntl);
writel(0, &par->neo2200->srcStart);
// par->neo2200->dstStart = (image->dy << 16) | (image->dx & 0xffff);
writel(((image->dx & 0xffff) * (info->var.bits_per_pixel >> 3) +
image->dy * info->fix.line_length), &par->neo2200->dstStart);
writel((image->height << 16) | (image->width & 0xffff),
&par->neo2200->xyExt);
memcpy_toio(par->mmio_vbase + 0x100000, image->data, data_len);
}
static void
neofb_fillrect(struct fb_info *info, const struct fb_fillrect *rect)
{
switch (info->fix.accel) {
case FB_ACCEL_NEOMAGIC_NM2200:
case FB_ACCEL_NEOMAGIC_NM2230:
case FB_ACCEL_NEOMAGIC_NM2360:
case FB_ACCEL_NEOMAGIC_NM2380:
neo2200_fillrect(info, rect);
break;
default:
cfb_fillrect(info, rect);
break;
}
}
static void
neofb_copyarea(struct fb_info *info, const struct fb_copyarea *area)
{
switch (info->fix.accel) {
case FB_ACCEL_NEOMAGIC_NM2200:
case FB_ACCEL_NEOMAGIC_NM2230:
case FB_ACCEL_NEOMAGIC_NM2360:
case FB_ACCEL_NEOMAGIC_NM2380:
neo2200_copyarea(info, area);
break;
default:
cfb_copyarea(info, area);
break;
}
}
static void
neofb_imageblit(struct fb_info *info, const struct fb_image *image)
{
switch (info->fix.accel) {
case FB_ACCEL_NEOMAGIC_NM2200:
case FB_ACCEL_NEOMAGIC_NM2230:
case FB_ACCEL_NEOMAGIC_NM2360:
case FB_ACCEL_NEOMAGIC_NM2380:
neo2200_imageblit(info, image);
break;
default:
cfb_imageblit(info, image);
break;
}
}
static int
neofb_sync(struct fb_info *info)
{
switch (info->fix.accel) {
case FB_ACCEL_NEOMAGIC_NM2200:
case FB_ACCEL_NEOMAGIC_NM2230:
case FB_ACCEL_NEOMAGIC_NM2360:
case FB_ACCEL_NEOMAGIC_NM2380:
neo2200_sync(info);
break;
default:
break;
}
return 0;
}
/*
static void
neofb_draw_cursor(struct fb_info *info, u8 *dst, u8 *src, unsigned int width)
{
//memset_io(info->sprite.addr, 0xff, 1);
}
static int
neofb_cursor(struct fb_info *info, struct fb_cursor *cursor)
{
struct neofb_par *par = (struct neofb_par *) info->par;
* Disable cursor *
write_le32(NEOREG_CURSCNTL, ~NEO_CURS_ENABLE, par);
if (cursor->set & FB_CUR_SETPOS) {
u32 x = cursor->image.dx;
u32 y = cursor->image.dy;
info->cursor.image.dx = x;
info->cursor.image.dy = y;
write_le32(NEOREG_CURSX, x, par);
write_le32(NEOREG_CURSY, y, par);
}
if (cursor->set & FB_CUR_SETSIZE) {
info->cursor.image.height = cursor->image.height;
info->cursor.image.width = cursor->image.width;
}
if (cursor->set & FB_CUR_SETHOT)
info->cursor.hot = cursor->hot;
if (cursor->set & FB_CUR_SETCMAP) {
if (cursor->image.depth == 1) {
u32 fg = cursor->image.fg_color;
u32 bg = cursor->image.bg_color;
info->cursor.image.fg_color = fg;
info->cursor.image.bg_color = bg;
fg = ((fg & 0xff0000) >> 16) | ((fg & 0xff) << 16) | (fg & 0xff00);
bg = ((bg & 0xff0000) >> 16) | ((bg & 0xff) << 16) | (bg & 0xff00);
write_le32(NEOREG_CURSFGCOLOR, fg, par);
write_le32(NEOREG_CURSBGCOLOR, bg, par);
}
}
if (cursor->set & FB_CUR_SETSHAPE)
fb_load_cursor_image(info);
if (info->cursor.enable)
write_le32(NEOREG_CURSCNTL, NEO_CURS_ENABLE, par);
return 0;
}
*/
static struct fb_ops neofb_ops = {
.owner = THIS_MODULE,
.fb_open = neofb_open,
.fb_release = neofb_release,
.fb_check_var = neofb_check_var,
.fb_set_par = neofb_set_par,
.fb_setcolreg = neofb_setcolreg,
.fb_pan_display = neofb_pan_display,
.fb_blank = neofb_blank,
.fb_sync = neofb_sync,
.fb_fillrect = neofb_fillrect,
.fb_copyarea = neofb_copyarea,
.fb_imageblit = neofb_imageblit,
};
/* --------------------------------------------------------------------- */
static struct fb_videomode mode800x480 = {
.xres = 800,
.yres = 480,
.pixclock = 25000,
.left_margin = 88,
.right_margin = 40,
.upper_margin = 23,
.lower_margin = 1,
.hsync_len = 128,
.vsync_len = 4,
.sync = FB_SYNC_HOR_HIGH_ACT | FB_SYNC_VERT_HIGH_ACT,
.vmode = FB_VMODE_NONINTERLACED
};
static int neo_map_mmio(struct fb_info *info, struct pci_dev *dev)
{
struct neofb_par *par = info->par;
DBG("neo_map_mmio");
switch (info->fix.accel) {
case FB_ACCEL_NEOMAGIC_NM2070:
info->fix.mmio_start = pci_resource_start(dev, 0)+
0x100000;
break;
case FB_ACCEL_NEOMAGIC_NM2090:
case FB_ACCEL_NEOMAGIC_NM2093:
info->fix.mmio_start = pci_resource_start(dev, 0)+
0x200000;
break;
case FB_ACCEL_NEOMAGIC_NM2160:
case FB_ACCEL_NEOMAGIC_NM2097:
case FB_ACCEL_NEOMAGIC_NM2200:
case FB_ACCEL_NEOMAGIC_NM2230:
case FB_ACCEL_NEOMAGIC_NM2360:
case FB_ACCEL_NEOMAGIC_NM2380:
info->fix.mmio_start = pci_resource_start(dev, 1);
break;
default:
info->fix.mmio_start = pci_resource_start(dev, 0);
}
info->fix.mmio_len = MMIO_SIZE;
if (!request_mem_region
(info->fix.mmio_start, MMIO_SIZE, "memory mapped I/O")) {
printk("neofb: memory mapped IO in use\n");
return -EBUSY;
}
par->mmio_vbase = ioremap(info->fix.mmio_start, MMIO_SIZE);
if (!par->mmio_vbase) {
printk("neofb: unable to map memory mapped IO\n");
release_mem_region(info->fix.mmio_start,
info->fix.mmio_len);
return -ENOMEM;
} else
printk(KERN_INFO "neofb: mapped io at %p\n",
par->mmio_vbase);
return 0;
}
static void neo_unmap_mmio(struct fb_info *info)
{
struct neofb_par *par = info->par;
DBG("neo_unmap_mmio");
iounmap(par->mmio_vbase);
par->mmio_vbase = NULL;
release_mem_region(info->fix.mmio_start,
info->fix.mmio_len);
}
static int neo_map_video(struct fb_info *info, struct pci_dev *dev,
int video_len)
{
//unsigned long addr;
video: fbdev: neofb: use arch_phys_wc_add() and ioremap_wc() Convert the driver from using the x86 specific MTRR code to the architecture agnostic arch_phys_wc_add(). arch_phys_wc_add() will avoid MTRR if write-combining is available, in order to take advantage of that also ensure the ioremap'd area is requested as write-combining. There are a few motivations for this: a) Take advantage of PAT when available b) Help bury MTRR code away, MTRR is architecture specific and on x86 its replaced by PAT c) Help with the goal of eventually using _PAGE_CACHE_UC over _PAGE_CACHE_UC_MINUS on x86 on ioremap_nocache() (see commit de33c442e titled "x86 PAT: fix performance drop for glx, use UC minus for ioremap(), ioremap_nocache() and pci_mmap_page_range()") The conversion done is expressed by the following Coccinelle SmPL patch, it additionally required manual intervention to address all the #ifdery and removal of redundant things which arch_phys_wc_add() already addresses such as verbose message about when MTRR fails and doing nothing when we didn't get an MTRR. @ mtrr_found @ expression index, base, size; @@ -index = mtrr_add(base, size, MTRR_TYPE_WRCOMB, 1); +index = arch_phys_wc_add(base, size); @ mtrr_rm depends on mtrr_found @ expression mtrr_found.index, mtrr_found.base, mtrr_found.size; @@ -mtrr_del(index, base, size); +arch_phys_wc_del(index); @ mtrr_rm_zero_arg depends on mtrr_found @ expression mtrr_found.index; @@ -mtrr_del(index, 0, 0); +arch_phys_wc_del(index); @ mtrr_rm_fb_info depends on mtrr_found @ struct fb_info *info; expression mtrr_found.index; @@ -mtrr_del(index, info->fix.smem_start, info->fix.smem_len); +arch_phys_wc_del(index); @ ioremap_replace_nocache depends on mtrr_found @ struct fb_info *info; expression base, size; @@ -info->screen_base = ioremap_nocache(base, size); +info->screen_base = ioremap_wc(base, size); @ ioremap_replace_default depends on mtrr_found @ struct fb_info *info; expression base, size; @@ -info->screen_base = ioremap(base, size); +info->screen_base = ioremap_wc(base, size); Generated-by: Coccinelle SmPL Cc: Jean-Christophe Plagniol-Villard <plagnioj@jcrosoft.com> Cc: Tomi Valkeinen <tomi.valkeinen@ti.com> Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Cc: Jingoo Han <jg1.han@samsung.com> Cc: Suresh Siddha <sbsiddha@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Juergen Gross <jgross@suse.com> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Andy Lutomirski <luto@amacapital.net> Cc: Dave Airlie <airlied@redhat.com> Cc: Antonino Daplas <adaplas@gmail.com> Cc: linux-fbdev@vger.kernel.org Cc: linux-kernel@vger.kernel.org Signed-off-by: Luis R. Rodriguez <mcgrof@suse.com> Reviewed-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
2015-04-22 04:16:27 +08:00
struct neofb_par *par = info->par;
DBG("neo_map_video");
info->fix.smem_start = pci_resource_start(dev, 0);
info->fix.smem_len = video_len;
if (!request_mem_region(info->fix.smem_start, info->fix.smem_len,
"frame buffer")) {
printk("neofb: frame buffer in use\n");
return -EBUSY;
}
info->screen_base =
video: fbdev: neofb: use arch_phys_wc_add() and ioremap_wc() Convert the driver from using the x86 specific MTRR code to the architecture agnostic arch_phys_wc_add(). arch_phys_wc_add() will avoid MTRR if write-combining is available, in order to take advantage of that also ensure the ioremap'd area is requested as write-combining. There are a few motivations for this: a) Take advantage of PAT when available b) Help bury MTRR code away, MTRR is architecture specific and on x86 its replaced by PAT c) Help with the goal of eventually using _PAGE_CACHE_UC over _PAGE_CACHE_UC_MINUS on x86 on ioremap_nocache() (see commit de33c442e titled "x86 PAT: fix performance drop for glx, use UC minus for ioremap(), ioremap_nocache() and pci_mmap_page_range()") The conversion done is expressed by the following Coccinelle SmPL patch, it additionally required manual intervention to address all the #ifdery and removal of redundant things which arch_phys_wc_add() already addresses such as verbose message about when MTRR fails and doing nothing when we didn't get an MTRR. @ mtrr_found @ expression index, base, size; @@ -index = mtrr_add(base, size, MTRR_TYPE_WRCOMB, 1); +index = arch_phys_wc_add(base, size); @ mtrr_rm depends on mtrr_found @ expression mtrr_found.index, mtrr_found.base, mtrr_found.size; @@ -mtrr_del(index, base, size); +arch_phys_wc_del(index); @ mtrr_rm_zero_arg depends on mtrr_found @ expression mtrr_found.index; @@ -mtrr_del(index, 0, 0); +arch_phys_wc_del(index); @ mtrr_rm_fb_info depends on mtrr_found @ struct fb_info *info; expression mtrr_found.index; @@ -mtrr_del(index, info->fix.smem_start, info->fix.smem_len); +arch_phys_wc_del(index); @ ioremap_replace_nocache depends on mtrr_found @ struct fb_info *info; expression base, size; @@ -info->screen_base = ioremap_nocache(base, size); +info->screen_base = ioremap_wc(base, size); @ ioremap_replace_default depends on mtrr_found @ struct fb_info *info; expression base, size; @@ -info->screen_base = ioremap(base, size); +info->screen_base = ioremap_wc(base, size); Generated-by: Coccinelle SmPL Cc: Jean-Christophe Plagniol-Villard <plagnioj@jcrosoft.com> Cc: Tomi Valkeinen <tomi.valkeinen@ti.com> Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Cc: Jingoo Han <jg1.han@samsung.com> Cc: Suresh Siddha <sbsiddha@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Juergen Gross <jgross@suse.com> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Andy Lutomirski <luto@amacapital.net> Cc: Dave Airlie <airlied@redhat.com> Cc: Antonino Daplas <adaplas@gmail.com> Cc: linux-fbdev@vger.kernel.org Cc: linux-kernel@vger.kernel.org Signed-off-by: Luis R. Rodriguez <mcgrof@suse.com> Reviewed-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
2015-04-22 04:16:27 +08:00
ioremap_wc(info->fix.smem_start, info->fix.smem_len);
if (!info->screen_base) {
printk("neofb: unable to map screen memory\n");
release_mem_region(info->fix.smem_start,
info->fix.smem_len);
return -ENOMEM;
} else
printk(KERN_INFO "neofb: mapped framebuffer at %p\n",
info->screen_base);
video: fbdev: neofb: use arch_phys_wc_add() and ioremap_wc() Convert the driver from using the x86 specific MTRR code to the architecture agnostic arch_phys_wc_add(). arch_phys_wc_add() will avoid MTRR if write-combining is available, in order to take advantage of that also ensure the ioremap'd area is requested as write-combining. There are a few motivations for this: a) Take advantage of PAT when available b) Help bury MTRR code away, MTRR is architecture specific and on x86 its replaced by PAT c) Help with the goal of eventually using _PAGE_CACHE_UC over _PAGE_CACHE_UC_MINUS on x86 on ioremap_nocache() (see commit de33c442e titled "x86 PAT: fix performance drop for glx, use UC minus for ioremap(), ioremap_nocache() and pci_mmap_page_range()") The conversion done is expressed by the following Coccinelle SmPL patch, it additionally required manual intervention to address all the #ifdery and removal of redundant things which arch_phys_wc_add() already addresses such as verbose message about when MTRR fails and doing nothing when we didn't get an MTRR. @ mtrr_found @ expression index, base, size; @@ -index = mtrr_add(base, size, MTRR_TYPE_WRCOMB, 1); +index = arch_phys_wc_add(base, size); @ mtrr_rm depends on mtrr_found @ expression mtrr_found.index, mtrr_found.base, mtrr_found.size; @@ -mtrr_del(index, base, size); +arch_phys_wc_del(index); @ mtrr_rm_zero_arg depends on mtrr_found @ expression mtrr_found.index; @@ -mtrr_del(index, 0, 0); +arch_phys_wc_del(index); @ mtrr_rm_fb_info depends on mtrr_found @ struct fb_info *info; expression mtrr_found.index; @@ -mtrr_del(index, info->fix.smem_start, info->fix.smem_len); +arch_phys_wc_del(index); @ ioremap_replace_nocache depends on mtrr_found @ struct fb_info *info; expression base, size; @@ -info->screen_base = ioremap_nocache(base, size); +info->screen_base = ioremap_wc(base, size); @ ioremap_replace_default depends on mtrr_found @ struct fb_info *info; expression base, size; @@ -info->screen_base = ioremap(base, size); +info->screen_base = ioremap_wc(base, size); Generated-by: Coccinelle SmPL Cc: Jean-Christophe Plagniol-Villard <plagnioj@jcrosoft.com> Cc: Tomi Valkeinen <tomi.valkeinen@ti.com> Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Cc: Jingoo Han <jg1.han@samsung.com> Cc: Suresh Siddha <sbsiddha@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Juergen Gross <jgross@suse.com> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Andy Lutomirski <luto@amacapital.net> Cc: Dave Airlie <airlied@redhat.com> Cc: Antonino Daplas <adaplas@gmail.com> Cc: linux-fbdev@vger.kernel.org Cc: linux-kernel@vger.kernel.org Signed-off-by: Luis R. Rodriguez <mcgrof@suse.com> Reviewed-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
2015-04-22 04:16:27 +08:00
par->wc_cookie = arch_phys_wc_add(info->fix.smem_start,
pci_resource_len(dev, 0));
/* Clear framebuffer, it's all white in memory after boot */
memset_io(info->screen_base, 0, info->fix.smem_len);
/* Allocate Cursor drawing pad.
info->fix.smem_len -= PAGE_SIZE;
addr = info->fix.smem_start + info->fix.smem_len;
write_le32(NEOREG_CURSMEMPOS, ((0x000f & (addr >> 10)) << 8) |
((0x0ff0 & (addr >> 10)) >> 4), par);
addr = (unsigned long) info->screen_base + info->fix.smem_len;
info->sprite.addr = (u8 *) addr; */
return 0;
}
static void neo_unmap_video(struct fb_info *info)
{
video: fbdev: neofb: use arch_phys_wc_add() and ioremap_wc() Convert the driver from using the x86 specific MTRR code to the architecture agnostic arch_phys_wc_add(). arch_phys_wc_add() will avoid MTRR if write-combining is available, in order to take advantage of that also ensure the ioremap'd area is requested as write-combining. There are a few motivations for this: a) Take advantage of PAT when available b) Help bury MTRR code away, MTRR is architecture specific and on x86 its replaced by PAT c) Help with the goal of eventually using _PAGE_CACHE_UC over _PAGE_CACHE_UC_MINUS on x86 on ioremap_nocache() (see commit de33c442e titled "x86 PAT: fix performance drop for glx, use UC minus for ioremap(), ioremap_nocache() and pci_mmap_page_range()") The conversion done is expressed by the following Coccinelle SmPL patch, it additionally required manual intervention to address all the #ifdery and removal of redundant things which arch_phys_wc_add() already addresses such as verbose message about when MTRR fails and doing nothing when we didn't get an MTRR. @ mtrr_found @ expression index, base, size; @@ -index = mtrr_add(base, size, MTRR_TYPE_WRCOMB, 1); +index = arch_phys_wc_add(base, size); @ mtrr_rm depends on mtrr_found @ expression mtrr_found.index, mtrr_found.base, mtrr_found.size; @@ -mtrr_del(index, base, size); +arch_phys_wc_del(index); @ mtrr_rm_zero_arg depends on mtrr_found @ expression mtrr_found.index; @@ -mtrr_del(index, 0, 0); +arch_phys_wc_del(index); @ mtrr_rm_fb_info depends on mtrr_found @ struct fb_info *info; expression mtrr_found.index; @@ -mtrr_del(index, info->fix.smem_start, info->fix.smem_len); +arch_phys_wc_del(index); @ ioremap_replace_nocache depends on mtrr_found @ struct fb_info *info; expression base, size; @@ -info->screen_base = ioremap_nocache(base, size); +info->screen_base = ioremap_wc(base, size); @ ioremap_replace_default depends on mtrr_found @ struct fb_info *info; expression base, size; @@ -info->screen_base = ioremap(base, size); +info->screen_base = ioremap_wc(base, size); Generated-by: Coccinelle SmPL Cc: Jean-Christophe Plagniol-Villard <plagnioj@jcrosoft.com> Cc: Tomi Valkeinen <tomi.valkeinen@ti.com> Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Cc: Jingoo Han <jg1.han@samsung.com> Cc: Suresh Siddha <sbsiddha@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Juergen Gross <jgross@suse.com> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Andy Lutomirski <luto@amacapital.net> Cc: Dave Airlie <airlied@redhat.com> Cc: Antonino Daplas <adaplas@gmail.com> Cc: linux-fbdev@vger.kernel.org Cc: linux-kernel@vger.kernel.org Signed-off-by: Luis R. Rodriguez <mcgrof@suse.com> Reviewed-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
2015-04-22 04:16:27 +08:00
struct neofb_par *par = info->par;
video: fbdev: neofb: use arch_phys_wc_add() and ioremap_wc() Convert the driver from using the x86 specific MTRR code to the architecture agnostic arch_phys_wc_add(). arch_phys_wc_add() will avoid MTRR if write-combining is available, in order to take advantage of that also ensure the ioremap'd area is requested as write-combining. There are a few motivations for this: a) Take advantage of PAT when available b) Help bury MTRR code away, MTRR is architecture specific and on x86 its replaced by PAT c) Help with the goal of eventually using _PAGE_CACHE_UC over _PAGE_CACHE_UC_MINUS on x86 on ioremap_nocache() (see commit de33c442e titled "x86 PAT: fix performance drop for glx, use UC minus for ioremap(), ioremap_nocache() and pci_mmap_page_range()") The conversion done is expressed by the following Coccinelle SmPL patch, it additionally required manual intervention to address all the #ifdery and removal of redundant things which arch_phys_wc_add() already addresses such as verbose message about when MTRR fails and doing nothing when we didn't get an MTRR. @ mtrr_found @ expression index, base, size; @@ -index = mtrr_add(base, size, MTRR_TYPE_WRCOMB, 1); +index = arch_phys_wc_add(base, size); @ mtrr_rm depends on mtrr_found @ expression mtrr_found.index, mtrr_found.base, mtrr_found.size; @@ -mtrr_del(index, base, size); +arch_phys_wc_del(index); @ mtrr_rm_zero_arg depends on mtrr_found @ expression mtrr_found.index; @@ -mtrr_del(index, 0, 0); +arch_phys_wc_del(index); @ mtrr_rm_fb_info depends on mtrr_found @ struct fb_info *info; expression mtrr_found.index; @@ -mtrr_del(index, info->fix.smem_start, info->fix.smem_len); +arch_phys_wc_del(index); @ ioremap_replace_nocache depends on mtrr_found @ struct fb_info *info; expression base, size; @@ -info->screen_base = ioremap_nocache(base, size); +info->screen_base = ioremap_wc(base, size); @ ioremap_replace_default depends on mtrr_found @ struct fb_info *info; expression base, size; @@ -info->screen_base = ioremap(base, size); +info->screen_base = ioremap_wc(base, size); Generated-by: Coccinelle SmPL Cc: Jean-Christophe Plagniol-Villard <plagnioj@jcrosoft.com> Cc: Tomi Valkeinen <tomi.valkeinen@ti.com> Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Cc: Jingoo Han <jg1.han@samsung.com> Cc: Suresh Siddha <sbsiddha@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Juergen Gross <jgross@suse.com> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Andy Lutomirski <luto@amacapital.net> Cc: Dave Airlie <airlied@redhat.com> Cc: Antonino Daplas <adaplas@gmail.com> Cc: linux-fbdev@vger.kernel.org Cc: linux-kernel@vger.kernel.org Signed-off-by: Luis R. Rodriguez <mcgrof@suse.com> Reviewed-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
2015-04-22 04:16:27 +08:00
DBG("neo_unmap_video");
video: fbdev: neofb: use arch_phys_wc_add() and ioremap_wc() Convert the driver from using the x86 specific MTRR code to the architecture agnostic arch_phys_wc_add(). arch_phys_wc_add() will avoid MTRR if write-combining is available, in order to take advantage of that also ensure the ioremap'd area is requested as write-combining. There are a few motivations for this: a) Take advantage of PAT when available b) Help bury MTRR code away, MTRR is architecture specific and on x86 its replaced by PAT c) Help with the goal of eventually using _PAGE_CACHE_UC over _PAGE_CACHE_UC_MINUS on x86 on ioremap_nocache() (see commit de33c442e titled "x86 PAT: fix performance drop for glx, use UC minus for ioremap(), ioremap_nocache() and pci_mmap_page_range()") The conversion done is expressed by the following Coccinelle SmPL patch, it additionally required manual intervention to address all the #ifdery and removal of redundant things which arch_phys_wc_add() already addresses such as verbose message about when MTRR fails and doing nothing when we didn't get an MTRR. @ mtrr_found @ expression index, base, size; @@ -index = mtrr_add(base, size, MTRR_TYPE_WRCOMB, 1); +index = arch_phys_wc_add(base, size); @ mtrr_rm depends on mtrr_found @ expression mtrr_found.index, mtrr_found.base, mtrr_found.size; @@ -mtrr_del(index, base, size); +arch_phys_wc_del(index); @ mtrr_rm_zero_arg depends on mtrr_found @ expression mtrr_found.index; @@ -mtrr_del(index, 0, 0); +arch_phys_wc_del(index); @ mtrr_rm_fb_info depends on mtrr_found @ struct fb_info *info; expression mtrr_found.index; @@ -mtrr_del(index, info->fix.smem_start, info->fix.smem_len); +arch_phys_wc_del(index); @ ioremap_replace_nocache depends on mtrr_found @ struct fb_info *info; expression base, size; @@ -info->screen_base = ioremap_nocache(base, size); +info->screen_base = ioremap_wc(base, size); @ ioremap_replace_default depends on mtrr_found @ struct fb_info *info; expression base, size; @@ -info->screen_base = ioremap(base, size); +info->screen_base = ioremap_wc(base, size); Generated-by: Coccinelle SmPL Cc: Jean-Christophe Plagniol-Villard <plagnioj@jcrosoft.com> Cc: Tomi Valkeinen <tomi.valkeinen@ti.com> Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Cc: Jingoo Han <jg1.han@samsung.com> Cc: Suresh Siddha <sbsiddha@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Juergen Gross <jgross@suse.com> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Andy Lutomirski <luto@amacapital.net> Cc: Dave Airlie <airlied@redhat.com> Cc: Antonino Daplas <adaplas@gmail.com> Cc: linux-fbdev@vger.kernel.org Cc: linux-kernel@vger.kernel.org Signed-off-by: Luis R. Rodriguez <mcgrof@suse.com> Reviewed-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
2015-04-22 04:16:27 +08:00
arch_phys_wc_del(par->wc_cookie);
iounmap(info->screen_base);
info->screen_base = NULL;
release_mem_region(info->fix.smem_start,
info->fix.smem_len);
}
static int neo_scan_monitor(struct fb_info *info)
{
struct neofb_par *par = info->par;
unsigned char type, display;
int w;
// Eventually we will have i2c support.
info->monspecs.modedb = kmalloc(sizeof(struct fb_videomode), GFP_KERNEL);
if (!info->monspecs.modedb)
return -ENOMEM;
info->monspecs.modedb_len = 1;
/* Determine the panel type */
vga_wgfx(NULL, 0x09, 0x26);
type = vga_rgfx(NULL, 0x21);
display = vga_rgfx(NULL, 0x20);
if (!par->internal_display && !par->external_display) {
par->internal_display = display & 2 || !(display & 3) ? 1 : 0;
par->external_display = display & 1;
printk (KERN_INFO "Autodetected %s display\n",
par->internal_display && par->external_display ? "simultaneous" :
par->internal_display ? "internal" : "external");
}
/* Determine panel width -- used in NeoValidMode. */
w = vga_rgfx(NULL, 0x20);
vga_wgfx(NULL, 0x09, 0x00);
switch ((w & 0x18) >> 3) {
case 0x00:
// 640x480@60
par->NeoPanelWidth = 640;
par->NeoPanelHeight = 480;
memcpy(info->monspecs.modedb, &vesa_modes[3], sizeof(struct fb_videomode));
break;
case 0x01:
par->NeoPanelWidth = 800;
if (par->libretto) {
par->NeoPanelHeight = 480;
memcpy(info->monspecs.modedb, &mode800x480, sizeof(struct fb_videomode));
} else {
// 800x600@60
par->NeoPanelHeight = 600;
memcpy(info->monspecs.modedb, &vesa_modes[8], sizeof(struct fb_videomode));
}
break;
case 0x02:
// 1024x768@60
par->NeoPanelWidth = 1024;
par->NeoPanelHeight = 768;
memcpy(info->monspecs.modedb, &vesa_modes[13], sizeof(struct fb_videomode));
break;
case 0x03:
/* 1280x1024@60 panel support needs to be added */
#ifdef NOT_DONE
par->NeoPanelWidth = 1280;
par->NeoPanelHeight = 1024;
memcpy(info->monspecs.modedb, &vesa_modes[20], sizeof(struct fb_videomode));
break;
#else
printk(KERN_ERR
"neofb: Only 640x480, 800x600/480 and 1024x768 panels are currently supported\n");
return -1;
#endif
default:
// 640x480@60
par->NeoPanelWidth = 640;
par->NeoPanelHeight = 480;
memcpy(info->monspecs.modedb, &vesa_modes[3], sizeof(struct fb_videomode));
break;
}
printk(KERN_INFO "Panel is a %dx%d %s %s display\n",
par->NeoPanelWidth,
par->NeoPanelHeight,
(type & 0x02) ? "color" : "monochrome",
(type & 0x10) ? "TFT" : "dual scan");
return 0;
}
static int neo_init_hw(struct fb_info *info)
{
struct neofb_par *par = info->par;
int videoRam = 896;
int maxClock = 65000;
int CursorMem = 1024;
int CursorOff = 0x100;
DBG("neo_init_hw");
neoUnlock();
#if 0
printk(KERN_DEBUG "--- Neo extended register dump ---\n");
for (int w = 0; w < 0x85; w++)
printk(KERN_DEBUG "CR %p: %p\n", (void *) w,
(void *) vga_rcrt(NULL, w));
for (int w = 0; w < 0xC7; w++)
printk(KERN_DEBUG "GR %p: %p\n", (void *) w,
(void *) vga_rgfx(NULL, w));
#endif
switch (info->fix.accel) {
case FB_ACCEL_NEOMAGIC_NM2070:
videoRam = 896;
maxClock = 65000;
break;
case FB_ACCEL_NEOMAGIC_NM2090:
case FB_ACCEL_NEOMAGIC_NM2093:
case FB_ACCEL_NEOMAGIC_NM2097:
videoRam = 1152;
maxClock = 80000;
break;
case FB_ACCEL_NEOMAGIC_NM2160:
videoRam = 2048;
maxClock = 90000;
break;
case FB_ACCEL_NEOMAGIC_NM2200:
videoRam = 2560;
maxClock = 110000;
break;
case FB_ACCEL_NEOMAGIC_NM2230:
videoRam = 3008;
maxClock = 110000;
break;
case FB_ACCEL_NEOMAGIC_NM2360:
videoRam = 4096;
maxClock = 110000;
break;
case FB_ACCEL_NEOMAGIC_NM2380:
videoRam = 6144;
maxClock = 110000;
break;
}
switch (info->fix.accel) {
case FB_ACCEL_NEOMAGIC_NM2070:
case FB_ACCEL_NEOMAGIC_NM2090:
case FB_ACCEL_NEOMAGIC_NM2093:
CursorMem = 2048;
CursorOff = 0x100;
break;
case FB_ACCEL_NEOMAGIC_NM2097:
case FB_ACCEL_NEOMAGIC_NM2160:
CursorMem = 1024;
CursorOff = 0x100;
break;
case FB_ACCEL_NEOMAGIC_NM2200:
case FB_ACCEL_NEOMAGIC_NM2230:
case FB_ACCEL_NEOMAGIC_NM2360:
case FB_ACCEL_NEOMAGIC_NM2380:
CursorMem = 1024;
CursorOff = 0x1000;
par->neo2200 = (Neo2200 __iomem *) par->mmio_vbase;
break;
}
/*
info->sprite.size = CursorMem;
info->sprite.scan_align = 1;
info->sprite.buf_align = 1;
info->sprite.flags = FB_PIXMAP_IO;
info->sprite.outbuf = neofb_draw_cursor;
*/
par->maxClock = maxClock;
par->cursorOff = CursorOff;
return videoRam * 1024;
}
static struct fb_info *neo_alloc_fb_info(struct pci_dev *dev,
const struct pci_device_id *id)
{
struct fb_info *info;
struct neofb_par *par;
info = framebuffer_alloc(sizeof(struct neofb_par), &dev->dev);
if (!info)
return NULL;
par = info->par;
info->fix.accel = id->driver_data;
par->pci_burst = !nopciburst;
par->lcd_stretch = !nostretch;
par->libretto = libretto;
par->internal_display = internal;
par->external_display = external;
info->flags = FBINFO_DEFAULT | FBINFO_HWACCEL_YPAN;
switch (info->fix.accel) {
case FB_ACCEL_NEOMAGIC_NM2070:
snprintf(info->fix.id, sizeof(info->fix.id),
"MagicGraph 128");
break;
case FB_ACCEL_NEOMAGIC_NM2090:
snprintf(info->fix.id, sizeof(info->fix.id),
"MagicGraph 128V");
break;
case FB_ACCEL_NEOMAGIC_NM2093:
snprintf(info->fix.id, sizeof(info->fix.id),
"MagicGraph 128ZV");
break;
case FB_ACCEL_NEOMAGIC_NM2097:
snprintf(info->fix.id, sizeof(info->fix.id),
"MagicGraph 128ZV+");
break;
case FB_ACCEL_NEOMAGIC_NM2160:
snprintf(info->fix.id, sizeof(info->fix.id),
"MagicGraph 128XD");
break;
case FB_ACCEL_NEOMAGIC_NM2200:
snprintf(info->fix.id, sizeof(info->fix.id),
"MagicGraph 256AV");
info->flags |= FBINFO_HWACCEL_IMAGEBLIT |
FBINFO_HWACCEL_COPYAREA |
FBINFO_HWACCEL_FILLRECT;
break;
case FB_ACCEL_NEOMAGIC_NM2230:
snprintf(info->fix.id, sizeof(info->fix.id),
"MagicGraph 256AV+");
info->flags |= FBINFO_HWACCEL_IMAGEBLIT |
FBINFO_HWACCEL_COPYAREA |
FBINFO_HWACCEL_FILLRECT;
break;
case FB_ACCEL_NEOMAGIC_NM2360:
snprintf(info->fix.id, sizeof(info->fix.id),
"MagicGraph 256ZX");
info->flags |= FBINFO_HWACCEL_IMAGEBLIT |
FBINFO_HWACCEL_COPYAREA |
FBINFO_HWACCEL_FILLRECT;
break;
case FB_ACCEL_NEOMAGIC_NM2380:
snprintf(info->fix.id, sizeof(info->fix.id),
"MagicGraph 256XL+");
info->flags |= FBINFO_HWACCEL_IMAGEBLIT |
FBINFO_HWACCEL_COPYAREA |
FBINFO_HWACCEL_FILLRECT;
break;
}
info->fix.type = FB_TYPE_PACKED_PIXELS;
info->fix.type_aux = 0;
info->fix.xpanstep = 0;
info->fix.ypanstep = 4;
info->fix.ywrapstep = 0;
info->fix.accel = id->driver_data;
info->fbops = &neofb_ops;
info->pseudo_palette = par->palette;
return info;
}
static void neo_free_fb_info(struct fb_info *info)
{
if (info) {
/*
* Free the colourmap
*/
fb_dealloc_cmap(&info->cmap);
framebuffer_release(info);
}
}
/* --------------------------------------------------------------------- */
static int neofb_probe(struct pci_dev *dev, const struct pci_device_id *id)
{
struct fb_info *info;
u_int h_sync, v_sync;
int video_len, err;
DBG("neofb_probe");
err = pci_enable_device(dev);
if (err)
return err;
err = -ENOMEM;
info = neo_alloc_fb_info(dev, id);
if (!info)
return err;
err = neo_map_mmio(info, dev);
if (err)
goto err_map_mmio;
err = neo_scan_monitor(info);
if (err)
goto err_scan_monitor;
video_len = neo_init_hw(info);
if (video_len < 0) {
err = video_len;
goto err_init_hw;
}
err = neo_map_video(info, dev, video_len);
if (err)
goto err_init_hw;
if (!fb_find_mode(&info->var, info, mode_option, NULL, 0,
info->monspecs.modedb, 16)) {
printk(KERN_ERR "neofb: Unable to find usable video mode.\n");
err = -EINVAL;
goto err_map_video;
}
/*
* Calculate the hsync and vsync frequencies. Note that
* we split the 1e12 constant up so that we can preserve
* the precision and fit the results into 32-bit registers.
* (1953125000 * 512 = 1e12)
*/
h_sync = 1953125000 / info->var.pixclock;
h_sync =
h_sync * 512 / (info->var.xres + info->var.left_margin +
info->var.right_margin + info->var.hsync_len);
v_sync =
h_sync / (info->var.yres + info->var.upper_margin +
info->var.lower_margin + info->var.vsync_len);
printk(KERN_INFO "neofb v" NEOFB_VERSION
": %dkB VRAM, using %dx%d, %d.%03dkHz, %dHz\n",
info->fix.smem_len >> 10, info->var.xres,
info->var.yres, h_sync / 1000, h_sync % 1000, v_sync);
err = fb_alloc_cmap(&info->cmap, 256, 0);
if (err < 0)
goto err_map_video;
err = register_framebuffer(info);
if (err < 0)
goto err_reg_fb;
fb_info(info, "%s frame buffer device\n", info->fix.id);
/*
* Our driver data
*/
pci_set_drvdata(dev, info);
return 0;
err_reg_fb:
fb_dealloc_cmap(&info->cmap);
err_map_video:
neo_unmap_video(info);
err_init_hw:
fb_destroy_modedb(info->monspecs.modedb);
err_scan_monitor:
neo_unmap_mmio(info);
err_map_mmio:
neo_free_fb_info(info);
return err;
}
static void neofb_remove(struct pci_dev *dev)
{
struct fb_info *info = pci_get_drvdata(dev);
DBG("neofb_remove");
if (info) {
/*
* If unregister_framebuffer fails, then
* we will be leaving hooks that could cause
* oopsen laying around.
*/
if (unregister_framebuffer(info))
printk(KERN_WARNING
"neofb: danger danger! Oopsen imminent!\n");
neo_unmap_video(info);
fb_destroy_modedb(info->monspecs.modedb);
neo_unmap_mmio(info);
neo_free_fb_info(info);
}
}
static struct pci_device_id neofb_devices[] = {
{PCI_VENDOR_ID_NEOMAGIC, PCI_CHIP_NM2070,
PCI_ANY_ID, PCI_ANY_ID, 0, 0, FB_ACCEL_NEOMAGIC_NM2070},
{PCI_VENDOR_ID_NEOMAGIC, PCI_CHIP_NM2090,
PCI_ANY_ID, PCI_ANY_ID, 0, 0, FB_ACCEL_NEOMAGIC_NM2090},
{PCI_VENDOR_ID_NEOMAGIC, PCI_CHIP_NM2093,
PCI_ANY_ID, PCI_ANY_ID, 0, 0, FB_ACCEL_NEOMAGIC_NM2093},
{PCI_VENDOR_ID_NEOMAGIC, PCI_CHIP_NM2097,
PCI_ANY_ID, PCI_ANY_ID, 0, 0, FB_ACCEL_NEOMAGIC_NM2097},
{PCI_VENDOR_ID_NEOMAGIC, PCI_CHIP_NM2160,
PCI_ANY_ID, PCI_ANY_ID, 0, 0, FB_ACCEL_NEOMAGIC_NM2160},
{PCI_VENDOR_ID_NEOMAGIC, PCI_CHIP_NM2200,
PCI_ANY_ID, PCI_ANY_ID, 0, 0, FB_ACCEL_NEOMAGIC_NM2200},
{PCI_VENDOR_ID_NEOMAGIC, PCI_CHIP_NM2230,
PCI_ANY_ID, PCI_ANY_ID, 0, 0, FB_ACCEL_NEOMAGIC_NM2230},
{PCI_VENDOR_ID_NEOMAGIC, PCI_CHIP_NM2360,
PCI_ANY_ID, PCI_ANY_ID, 0, 0, FB_ACCEL_NEOMAGIC_NM2360},
{PCI_VENDOR_ID_NEOMAGIC, PCI_CHIP_NM2380,
PCI_ANY_ID, PCI_ANY_ID, 0, 0, FB_ACCEL_NEOMAGIC_NM2380},
{0, 0, 0, 0, 0, 0, 0}
};
MODULE_DEVICE_TABLE(pci, neofb_devices);
static struct pci_driver neofb_driver = {
.name = "neofb",
.id_table = neofb_devices,
.probe = neofb_probe,
.remove = neofb_remove,
};
/* ************************* init in-kernel code ************************** */
#ifndef MODULE
static int __init neofb_setup(char *options)
{
char *this_opt;
DBG("neofb_setup");
if (!options || !*options)
return 0;
while ((this_opt = strsep(&options, ",")) != NULL) {
if (!*this_opt)
continue;
if (!strncmp(this_opt, "internal", 8))
internal = 1;
else if (!strncmp(this_opt, "external", 8))
external = 1;
else if (!strncmp(this_opt, "nostretch", 9))
nostretch = 1;
else if (!strncmp(this_opt, "nopciburst", 10))
nopciburst = 1;
else if (!strncmp(this_opt, "libretto", 8))
libretto = 1;
else
mode_option = this_opt;
}
return 0;
}
#endif /* MODULE */
static int __init neofb_init(void)
{
#ifndef MODULE
char *option = NULL;
if (fb_get_options("neofb", &option))
return -ENODEV;
neofb_setup(option);
#endif
return pci_register_driver(&neofb_driver);
}
module_init(neofb_init);
#ifdef MODULE
static void __exit neofb_exit(void)
{
pci_unregister_driver(&neofb_driver);
}
module_exit(neofb_exit);
#endif /* MODULE */