linux/drivers/video/fbdev/tdfxfb.c

1660 lines
42 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0-only
/*
*
* tdfxfb.c
*
* Author: Hannu Mallat <hmallat@cc.hut.fi>
*
* Copyright © 1999 Hannu Mallat
* All rights reserved
*
* Created : Thu Sep 23 18:17:43 1999, hmallat
* Last modified: Tue Nov 2 21:19:47 1999, hmallat
*
* I2C part copied from the i2c-voodoo3.c driver by:
* Frodo Looijaard <frodol@dds.nl>,
* Philip Edelbrock <phil@netroedge.com>,
* Ralph Metzler <rjkm@thp.uni-koeln.de>, and
* Mark D. Studebaker <mdsxyz123@yahoo.com>
*
* Lots of the information here comes from the Daryll Strauss' Banshee
* patches to the XF86 server, and the rest comes from the 3dfx
* Banshee specification. I'm very much indebted to Daryll for his
* work on the X server.
*
* Voodoo3 support was contributed Harold Oga. Lots of additions
* (proper acceleration, 24 bpp, hardware cursor) and bug fixes by Attila
* Kesmarki. Thanks guys!
*
* Voodoo1 and Voodoo2 support aren't relevant to this driver as they
* behave very differently from the Voodoo3/4/5. For anyone wanting to
* use frame buffer on the Voodoo1/2, see the sstfb driver (which is
* located at http://www.sourceforge.net/projects/sstfb).
*
* While I _am_ grateful to 3Dfx for releasing the specs for Banshee,
* I do wish the next version is a bit more complete. Without the XF86
* patches I couldn't have gotten even this far... for instance, the
* extensions to the VGA register set go completely unmentioned in the
* spec! Also, lots of references are made to the 'SST core', but no
* spec is publicly available, AFAIK.
*
* The structure of this driver comes pretty much from the Permedia
* driver by Ilario Nardinocchi, which in turn is based on skeletonfb.
*
* TODO:
* - multihead support (basically need to support an array of fb_infos)
* - support other architectures (PPC, Alpha); does the fact that the VGA
* core can be accessed only thru I/O (not memory mapped) complicate
* things?
*
* Version history:
*
* 0.1.4 (released 2002-05-28) ported over to new fbdev api by James Simmons
*
* 0.1.3 (released 1999-11-02) added Attila's panning support, code
* reorg, hwcursor address page size alignment
* (for mmapping both frame buffer and regs),
* and my changes to get rid of hardcoded
* VGA i/o register locations (uses PCI
* configuration info now)
* 0.1.2 (released 1999-10-19) added Attila Kesmarki's bug fixes and
* improvements
* 0.1.1 (released 1999-10-07) added Voodoo3 support by Harold Oga.
* 0.1.0 (released 1999-10-06) initial version
*
*/
#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/fb.h>
#include <linux/init.h>
#include <linux/pci.h>
#include <asm/io.h>
#include <video/tdfx.h>
#define DPRINTK(a, b...) pr_debug("fb: %s: " a, __func__ , ## b)
#define BANSHEE_MAX_PIXCLOCK 270000
#define VOODOO3_MAX_PIXCLOCK 300000
#define VOODOO5_MAX_PIXCLOCK 350000
static const struct fb_fix_screeninfo tdfx_fix = {
.type = FB_TYPE_PACKED_PIXELS,
.visual = FB_VISUAL_PSEUDOCOLOR,
.ypanstep = 1,
.ywrapstep = 1,
.accel = FB_ACCEL_3DFX_BANSHEE
};
static const struct fb_var_screeninfo tdfx_var = {
/* "640x480, 8 bpp @ 60 Hz */
.xres = 640,
.yres = 480,
.xres_virtual = 640,
.yres_virtual = 1024,
.bits_per_pixel = 8,
.red = {0, 8, 0},
.blue = {0, 8, 0},
.green = {0, 8, 0},
.activate = FB_ACTIVATE_NOW,
.height = -1,
.width = -1,
.accel_flags = FB_ACCELF_TEXT,
.pixclock = 39722,
.left_margin = 40,
.right_margin = 24,
.upper_margin = 32,
.lower_margin = 11,
.hsync_len = 96,
.vsync_len = 2,
.vmode = FB_VMODE_NONINTERLACED
};
/*
* PCI driver prototypes
*/
static int tdfxfb_probe(struct pci_dev *pdev, const struct pci_device_id *id);
static void tdfxfb_remove(struct pci_dev *pdev);
static const struct pci_device_id tdfxfb_id_table[] = {
{ PCI_VENDOR_ID_3DFX, PCI_DEVICE_ID_3DFX_BANSHEE,
PCI_ANY_ID, PCI_ANY_ID, PCI_BASE_CLASS_DISPLAY << 16,
0xff0000, 0 },
{ PCI_VENDOR_ID_3DFX, PCI_DEVICE_ID_3DFX_VOODOO3,
PCI_ANY_ID, PCI_ANY_ID, PCI_BASE_CLASS_DISPLAY << 16,
0xff0000, 0 },
{ PCI_VENDOR_ID_3DFX, PCI_DEVICE_ID_3DFX_VOODOO5,
PCI_ANY_ID, PCI_ANY_ID, PCI_BASE_CLASS_DISPLAY << 16,
0xff0000, 0 },
{ 0, }
};
static struct pci_driver tdfxfb_driver = {
.name = "tdfxfb",
.id_table = tdfxfb_id_table,
.probe = tdfxfb_probe,
.remove = tdfxfb_remove,
};
MODULE_DEVICE_TABLE(pci, tdfxfb_id_table);
/*
* Driver data
*/
static int nopan;
static int nowrap = 1; /* not implemented (yet) */
static int hwcursor = 1;
static char *mode_option;
static bool nomtrr;
/* -------------------------------------------------------------------------
* Hardware-specific funcions
* ------------------------------------------------------------------------- */
static inline u8 vga_inb(struct tdfx_par *par, u32 reg)
{
return inb(par->iobase + reg - 0x300);
}
static inline void vga_outb(struct tdfx_par *par, u32 reg, u8 val)
{
outb(val, par->iobase + reg - 0x300);
}
static inline void gra_outb(struct tdfx_par *par, u32 idx, u8 val)
{
vga_outb(par, GRA_I, idx);
wmb();
vga_outb(par, GRA_D, val);
wmb();
}
static inline void seq_outb(struct tdfx_par *par, u32 idx, u8 val)
{
vga_outb(par, SEQ_I, idx);
wmb();
vga_outb(par, SEQ_D, val);
wmb();
}
static inline u8 seq_inb(struct tdfx_par *par, u32 idx)
{
vga_outb(par, SEQ_I, idx);
mb();
return vga_inb(par, SEQ_D);
}
static inline void crt_outb(struct tdfx_par *par, u32 idx, u8 val)
{
vga_outb(par, CRT_I, idx);
wmb();
vga_outb(par, CRT_D, val);
wmb();
}
static inline u8 crt_inb(struct tdfx_par *par, u32 idx)
{
vga_outb(par, CRT_I, idx);
mb();
return vga_inb(par, CRT_D);
}
static inline void att_outb(struct tdfx_par *par, u32 idx, u8 val)
{
unsigned char tmp;
tmp = vga_inb(par, IS1_R);
vga_outb(par, ATT_IW, idx);
vga_outb(par, ATT_IW, val);
}
static inline void vga_disable_video(struct tdfx_par *par)
{
unsigned char s;
s = seq_inb(par, 0x01) | 0x20;
seq_outb(par, 0x00, 0x01);
seq_outb(par, 0x01, s);
seq_outb(par, 0x00, 0x03);
}
static inline void vga_enable_video(struct tdfx_par *par)
{
unsigned char s;
s = seq_inb(par, 0x01) & 0xdf;
seq_outb(par, 0x00, 0x01);
seq_outb(par, 0x01, s);
seq_outb(par, 0x00, 0x03);
}
static inline void vga_enable_palette(struct tdfx_par *par)
{
vga_inb(par, IS1_R);
mb();
vga_outb(par, ATT_IW, 0x20);
}
static inline u32 tdfx_inl(struct tdfx_par *par, unsigned int reg)
{
return readl(par->regbase_virt + reg);
}
static inline void tdfx_outl(struct tdfx_par *par, unsigned int reg, u32 val)
{
writel(val, par->regbase_virt + reg);
}
static inline void banshee_make_room(struct tdfx_par *par, int size)
{
/* Note: The Voodoo3's onboard FIFO has 32 slots. This loop
* won't quit if you ask for more. */
while ((tdfx_inl(par, STATUS) & 0x1f) < size - 1)
cpu_relax();
}
static int banshee_wait_idle(struct fb_info *info)
{
struct tdfx_par *par = info->par;
int i = 0;
banshee_make_room(par, 1);
tdfx_outl(par, COMMAND_3D, COMMAND_3D_NOP);
do {
if ((tdfx_inl(par, STATUS) & STATUS_BUSY) == 0)
i++;
} while (i < 3);
return 0;
}
/*
* Set the color of a palette entry in 8bpp mode
*/
static inline void do_setpalentry(struct tdfx_par *par, unsigned regno, u32 c)
{
banshee_make_room(par, 2);
tdfx_outl(par, DACADDR, regno);
/* read after write makes it working */
tdfx_inl(par, DACADDR);
tdfx_outl(par, DACDATA, c);
}
static u32 do_calc_pll(int freq, int *freq_out)
{
int m, n, k, best_m, best_n, best_k, best_error;
int fref = 14318;
best_error = freq;
best_n = best_m = best_k = 0;
for (k = 3; k >= 0; k--) {
for (m = 63; m >= 0; m--) {
/*
* Estimate value of n that produces target frequency
* with current m and k
*/
int n_estimated = ((freq * (m + 2) << k) / fref) - 2;
/* Search neighborhood of estimated n */
for (n = max(0, n_estimated);
n <= min(255, n_estimated + 1);
n++) {
/*
* Calculate PLL freqency with current m, k and
* estimated n
*/
int f = (fref * (n + 2) / (m + 2)) >> k;
int error = abs(f - freq);
/*
* If this is the closest we've come to the
* target frequency then remember n, m and k
*/
if (error < best_error) {
best_error = error;
best_n = n;
best_m = m;
best_k = k;
}
}
}
}
n = best_n;
m = best_m;
k = best_k;
*freq_out = (fref * (n + 2) / (m + 2)) >> k;
return (n << 8) | (m << 2) | k;
}
static void do_write_regs(struct fb_info *info, struct banshee_reg *reg)
{
struct tdfx_par *par = info->par;
int i;
banshee_wait_idle(info);
tdfx_outl(par, MISCINIT1, tdfx_inl(par, MISCINIT1) | 0x01);
crt_outb(par, 0x11, crt_inb(par, 0x11) & 0x7f); /* CRT unprotect */
banshee_make_room(par, 3);
tdfx_outl(par, VGAINIT1, reg->vgainit1 & 0x001FFFFF);
tdfx_outl(par, VIDPROCCFG, reg->vidcfg & ~0x00000001);
#if 0
tdfx_outl(par, PLLCTRL1, reg->mempll);
tdfx_outl(par, PLLCTRL2, reg->gfxpll);
#endif
tdfx_outl(par, PLLCTRL0, reg->vidpll);
vga_outb(par, MISC_W, reg->misc[0x00] | 0x01);
for (i = 0; i < 5; i++)
seq_outb(par, i, reg->seq[i]);
for (i = 0; i < 25; i++)
crt_outb(par, i, reg->crt[i]);
for (i = 0; i < 9; i++)
gra_outb(par, i, reg->gra[i]);
for (i = 0; i < 21; i++)
att_outb(par, i, reg->att[i]);
crt_outb(par, 0x1a, reg->ext[0]);
crt_outb(par, 0x1b, reg->ext[1]);
vga_enable_palette(par);
vga_enable_video(par);
banshee_make_room(par, 9);
tdfx_outl(par, VGAINIT0, reg->vgainit0);
tdfx_outl(par, DACMODE, reg->dacmode);
tdfx_outl(par, VIDDESKSTRIDE, reg->stride);
tdfx_outl(par, HWCURPATADDR, reg->curspataddr);
tdfx_outl(par, VIDSCREENSIZE, reg->screensize);
tdfx_outl(par, VIDDESKSTART, reg->startaddr);
tdfx_outl(par, VIDPROCCFG, reg->vidcfg);
tdfx_outl(par, VGAINIT1, reg->vgainit1);
tdfx_outl(par, MISCINIT0, reg->miscinit0);
banshee_make_room(par, 8);
tdfx_outl(par, SRCBASE, reg->startaddr);
tdfx_outl(par, DSTBASE, reg->startaddr);
tdfx_outl(par, COMMANDEXTRA_2D, 0);
tdfx_outl(par, CLIP0MIN, 0);
tdfx_outl(par, CLIP0MAX, 0x0fff0fff);
tdfx_outl(par, CLIP1MIN, 0);
tdfx_outl(par, CLIP1MAX, 0x0fff0fff);
tdfx_outl(par, SRCXY, 0);
banshee_wait_idle(info);
}
static unsigned long do_lfb_size(struct tdfx_par *par, unsigned short dev_id)
{
u32 draminit0 = tdfx_inl(par, DRAMINIT0);
u32 draminit1 = tdfx_inl(par, DRAMINIT1);
u32 miscinit1;
int num_chips = (draminit0 & DRAMINIT0_SGRAM_NUM) ? 8 : 4;
int chip_size; /* in MB */
int has_sgram = draminit1 & DRAMINIT1_MEM_SDRAM;
if (dev_id < PCI_DEVICE_ID_3DFX_VOODOO5) {
/* Banshee/Voodoo3 */
chip_size = 2;
if (has_sgram && !(draminit0 & DRAMINIT0_SGRAM_TYPE))
chip_size = 1;
} else {
/* Voodoo4/5 */
has_sgram = 0;
chip_size = draminit0 & DRAMINIT0_SGRAM_TYPE_MASK;
chip_size = 1 << (chip_size >> DRAMINIT0_SGRAM_TYPE_SHIFT);
}
/* disable block writes for SDRAM */
miscinit1 = tdfx_inl(par, MISCINIT1);
miscinit1 |= has_sgram ? 0 : MISCINIT1_2DBLOCK_DIS;
miscinit1 |= MISCINIT1_CLUT_INV;
banshee_make_room(par, 1);
tdfx_outl(par, MISCINIT1, miscinit1);
return num_chips * chip_size * 1024l * 1024;
}
/* ------------------------------------------------------------------------- */
static int tdfxfb_check_var(struct fb_var_screeninfo *var, struct fb_info *info)
{
struct tdfx_par *par = info->par;
u32 lpitch;
if (var->bits_per_pixel != 8 && var->bits_per_pixel != 16 &&
var->bits_per_pixel != 24 && var->bits_per_pixel != 32) {
DPRINTK("depth not supported: %u\n", var->bits_per_pixel);
return -EINVAL;
}
if (var->xres != var->xres_virtual)
var->xres_virtual = var->xres;
if (var->yres > var->yres_virtual)
var->yres_virtual = var->yres;
if (var->xoffset) {
DPRINTK("xoffset not supported\n");
return -EINVAL;
}
var->yoffset = 0;
/*
* Banshee doesn't support interlace, but Voodoo4/5 and probably
* Voodoo3 do.
* no direct information about device id now?
* use max_pixclock for this...
*/
if (((var->vmode & FB_VMODE_MASK) == FB_VMODE_INTERLACED) &&
(par->max_pixclock < VOODOO3_MAX_PIXCLOCK)) {
DPRINTK("interlace not supported\n");
return -EINVAL;
}
if (info->monspecs.hfmax && info->monspecs.vfmax &&
info->monspecs.dclkmax && fb_validate_mode(var, info) < 0) {
DPRINTK("mode outside monitor's specs\n");
return -EINVAL;
}
var->xres = (var->xres + 15) & ~15; /* could sometimes be 8 */
lpitch = var->xres * ((var->bits_per_pixel + 7) >> 3);
if (var->xres < 320 || var->xres > 2048) {
DPRINTK("width not supported: %u\n", var->xres);
return -EINVAL;
}
if (var->yres < 200 || var->yres > 2048) {
DPRINTK("height not supported: %u\n", var->yres);
return -EINVAL;
}
if (lpitch * var->yres_virtual > info->fix.smem_len) {
var->yres_virtual = info->fix.smem_len / lpitch;
if (var->yres_virtual < var->yres) {
DPRINTK("no memory for screen (%ux%ux%u)\n",
var->xres, var->yres_virtual,
var->bits_per_pixel);
return -EINVAL;
}
}
if (PICOS2KHZ(var->pixclock) > par->max_pixclock) {
DPRINTK("pixclock too high (%ldKHz)\n",
PICOS2KHZ(var->pixclock));
return -EINVAL;
}
var->transp.offset = 0;
var->transp.length = 0;
switch (var->bits_per_pixel) {
case 8:
var->red.length = 8;
var->red.offset = 0;
var->green = var->red;
var->blue = var->red;
break;
case 16:
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 32:
var->transp.offset = 24;
var->transp.length = 8;
/* fall through */
case 24:
var->red.offset = 16;
var->green.offset = 8;
var->blue.offset = 0;
var->red.length = var->green.length = var->blue.length = 8;
break;
}
var->width = -1;
var->height = -1;
var->accel_flags = FB_ACCELF_TEXT;
DPRINTK("Checking graphics mode at %dx%d depth %d\n",
var->xres, var->yres, var->bits_per_pixel);
return 0;
}
static int tdfxfb_set_par(struct fb_info *info)
{
struct tdfx_par *par = info->par;
u32 hdispend = info->var.xres;
u32 hsyncsta = hdispend + info->var.right_margin;
u32 hsyncend = hsyncsta + info->var.hsync_len;
u32 htotal = hsyncend + info->var.left_margin;
u32 hd, hs, he, ht, hbs, hbe;
u32 vd, vs, ve, vt, vbs, vbe;
struct banshee_reg reg;
int fout, freq;
u32 wd;
u32 cpp = (info->var.bits_per_pixel + 7) >> 3;
memset(&reg, 0, sizeof(reg));
reg.vidcfg = VIDCFG_VIDPROC_ENABLE | VIDCFG_DESK_ENABLE |
VIDCFG_CURS_X11 |
((cpp - 1) << VIDCFG_PIXFMT_SHIFT) |
(cpp != 1 ? VIDCFG_CLUT_BYPASS : 0);
/* PLL settings */
freq = PICOS2KHZ(info->var.pixclock);
reg.vidcfg &= ~VIDCFG_2X;
if (freq > par->max_pixclock / 2) {
freq = freq > par->max_pixclock ? par->max_pixclock : freq;
reg.dacmode |= DACMODE_2X;
reg.vidcfg |= VIDCFG_2X;
hdispend >>= 1;
hsyncsta >>= 1;
hsyncend >>= 1;
htotal >>= 1;
}
wd = (hdispend >> 3) - 1;
hd = wd;
hs = (hsyncsta >> 3) - 1;
he = (hsyncend >> 3) - 1;
ht = (htotal >> 3) - 1;
hbs = hd;
hbe = ht;
if ((info->var.vmode & FB_VMODE_MASK) == FB_VMODE_DOUBLE) {
vd = (info->var.yres << 1) - 1;
vs = vd + (info->var.lower_margin << 1);
ve = vs + (info->var.vsync_len << 1);
vt = ve + (info->var.upper_margin << 1) - 1;
reg.screensize = info->var.xres | (info->var.yres << 13);
reg.vidcfg |= VIDCFG_HALF_MODE;
reg.crt[0x09] = 0x80;
} else {
vd = info->var.yres - 1;
vs = vd + info->var.lower_margin;
ve = vs + info->var.vsync_len;
vt = ve + info->var.upper_margin - 1;
reg.screensize = info->var.xres | (info->var.yres << 12);
reg.vidcfg &= ~VIDCFG_HALF_MODE;
}
vbs = vd;
vbe = vt;
/* this is all pretty standard VGA register stuffing */
reg.misc[0x00] = 0x0f |
(info->var.xres < 400 ? 0xa0 :
info->var.xres < 480 ? 0x60 :
info->var.xres < 768 ? 0xe0 : 0x20);
reg.gra[0x05] = 0x40;
reg.gra[0x06] = 0x05;
reg.gra[0x07] = 0x0f;
reg.gra[0x08] = 0xff;
reg.att[0x00] = 0x00;
reg.att[0x01] = 0x01;
reg.att[0x02] = 0x02;
reg.att[0x03] = 0x03;
reg.att[0x04] = 0x04;
reg.att[0x05] = 0x05;
reg.att[0x06] = 0x06;
reg.att[0x07] = 0x07;
reg.att[0x08] = 0x08;
reg.att[0x09] = 0x09;
reg.att[0x0a] = 0x0a;
reg.att[0x0b] = 0x0b;
reg.att[0x0c] = 0x0c;
reg.att[0x0d] = 0x0d;
reg.att[0x0e] = 0x0e;
reg.att[0x0f] = 0x0f;
reg.att[0x10] = 0x41;
reg.att[0x12] = 0x0f;
reg.seq[0x00] = 0x03;
reg.seq[0x01] = 0x01; /* fixme: clkdiv2? */
reg.seq[0x02] = 0x0f;
reg.seq[0x03] = 0x00;
reg.seq[0x04] = 0x0e;
reg.crt[0x00] = ht - 4;
reg.crt[0x01] = hd;
reg.crt[0x02] = hbs;
reg.crt[0x03] = 0x80 | (hbe & 0x1f);
reg.crt[0x04] = hs;
reg.crt[0x05] = ((hbe & 0x20) << 2) | (he & 0x1f);
reg.crt[0x06] = vt;
reg.crt[0x07] = ((vs & 0x200) >> 2) |
((vd & 0x200) >> 3) |
((vt & 0x200) >> 4) | 0x10 |
((vbs & 0x100) >> 5) |
((vs & 0x100) >> 6) |
((vd & 0x100) >> 7) |
((vt & 0x100) >> 8);
reg.crt[0x09] |= 0x40 | ((vbs & 0x200) >> 4);
reg.crt[0x10] = vs;
reg.crt[0x11] = (ve & 0x0f) | 0x20;
reg.crt[0x12] = vd;
reg.crt[0x13] = wd;
reg.crt[0x15] = vbs;
reg.crt[0x16] = vbe + 1;
reg.crt[0x17] = 0xc3;
reg.crt[0x18] = 0xff;
/* Banshee's nonvga stuff */
reg.ext[0x00] = (((ht & 0x100) >> 8) |
((hd & 0x100) >> 6) |
((hbs & 0x100) >> 4) |
((hbe & 0x40) >> 1) |
((hs & 0x100) >> 2) |
((he & 0x20) << 2));
reg.ext[0x01] = (((vt & 0x400) >> 10) |
((vd & 0x400) >> 8) |
((vbs & 0x400) >> 6) |
((vbe & 0x400) >> 4));
reg.vgainit0 = VGAINIT0_8BIT_DAC |
VGAINIT0_EXT_ENABLE |
VGAINIT0_WAKEUP_3C3 |
VGAINIT0_ALT_READBACK |
VGAINIT0_EXTSHIFTOUT;
reg.vgainit1 = tdfx_inl(par, VGAINIT1) & 0x1fffff;
if (hwcursor)
reg.curspataddr = info->fix.smem_len;
reg.cursloc = 0;
reg.cursc0 = 0;
reg.cursc1 = 0xffffff;
reg.stride = info->var.xres * cpp;
reg.startaddr = info->var.yoffset * reg.stride
+ info->var.xoffset * cpp;
reg.vidpll = do_calc_pll(freq, &fout);
#if 0
reg.mempll = do_calc_pll(..., &fout);
reg.gfxpll = do_calc_pll(..., &fout);
#endif
if ((info->var.vmode & FB_VMODE_MASK) == FB_VMODE_INTERLACED)
reg.vidcfg |= VIDCFG_INTERLACE;
reg.miscinit0 = tdfx_inl(par, MISCINIT0);
#if defined(__BIG_ENDIAN)
switch (info->var.bits_per_pixel) {
case 8:
case 24:
reg.miscinit0 &= ~(1 << 30);
reg.miscinit0 &= ~(1 << 31);
break;
case 16:
reg.miscinit0 |= (1 << 30);
reg.miscinit0 |= (1 << 31);
break;
case 32:
reg.miscinit0 |= (1 << 30);
reg.miscinit0 &= ~(1 << 31);
break;
}
#endif
do_write_regs(info, &reg);
/* Now change fb_fix_screeninfo according to changes in par */
info->fix.line_length = reg.stride;
info->fix.visual = (info->var.bits_per_pixel == 8)
? FB_VISUAL_PSEUDOCOLOR
: FB_VISUAL_TRUECOLOR;
DPRINTK("Graphics mode is now set at %dx%d depth %d\n",
info->var.xres, info->var.yres, info->var.bits_per_pixel);
return 0;
}
/* A handy macro shamelessly pinched from matroxfb */
#define CNVT_TOHW(val, width) ((((val) << (width)) + 0x7FFF - (val)) >> 16)
static int tdfxfb_setcolreg(unsigned regno, unsigned red, unsigned green,
unsigned blue, unsigned transp,
struct fb_info *info)
{
struct tdfx_par *par = info->par;
u32 rgbcol;
if (regno >= info->cmap.len || regno > 255)
return 1;
/* grayscale works only partially under directcolor */
if (info->var.grayscale) {
/* grayscale = 0.30*R + 0.59*G + 0.11*B */
blue = (red * 77 + green * 151 + blue * 28) >> 8;
green = blue;
red = blue;
}
switch (info->fix.visual) {
case FB_VISUAL_PSEUDOCOLOR:
rgbcol = (((u32)red & 0xff00) << 8) |
(((u32)green & 0xff00) << 0) |
(((u32)blue & 0xff00) >> 8);
do_setpalentry(par, regno, rgbcol);
break;
/* Truecolor has no hardware color palettes. */
case FB_VISUAL_TRUECOLOR:
if (regno < 16) {
rgbcol = (CNVT_TOHW(red, info->var.red.length) <<
info->var.red.offset) |
(CNVT_TOHW(green, info->var.green.length) <<
info->var.green.offset) |
(CNVT_TOHW(blue, info->var.blue.length) <<
info->var.blue.offset) |
(CNVT_TOHW(transp, info->var.transp.length) <<
info->var.transp.offset);
par->palette[regno] = rgbcol;
}
break;
default:
DPRINTK("bad depth %u\n", info->var.bits_per_pixel);
break;
}
return 0;
}
/* 0 unblank, 1 blank, 2 no vsync, 3 no hsync, 4 off */
static int tdfxfb_blank(int blank, struct fb_info *info)
{
struct tdfx_par *par = info->par;
int vgablank = 1;
u32 dacmode = tdfx_inl(par, DACMODE);
dacmode &= ~(BIT(1) | BIT(3));
switch (blank) {
case FB_BLANK_UNBLANK: /* Screen: On; HSync: On, VSync: On */
vgablank = 0;
break;
case FB_BLANK_NORMAL: /* Screen: Off; HSync: On, VSync: On */
break;
case FB_BLANK_VSYNC_SUSPEND: /* Screen: Off; HSync: On, VSync: Off */
dacmode |= BIT(3);
break;
case FB_BLANK_HSYNC_SUSPEND: /* Screen: Off; HSync: Off, VSync: On */
dacmode |= BIT(1);
break;
case FB_BLANK_POWERDOWN: /* Screen: Off; HSync: Off, VSync: Off */
dacmode |= BIT(1) | BIT(3);
break;
}
banshee_make_room(par, 1);
tdfx_outl(par, DACMODE, dacmode);
if (vgablank)
vga_disable_video(par);
else
vga_enable_video(par);
return 0;
}
/*
* Set the starting position of the visible screen to var->yoffset
*/
static int tdfxfb_pan_display(struct fb_var_screeninfo *var,
struct fb_info *info)
{
struct tdfx_par *par = info->par;
u32 addr = var->yoffset * info->fix.line_length;
if (nopan || var->xoffset)
return -EINVAL;
banshee_make_room(par, 1);
tdfx_outl(par, VIDDESKSTART, addr);
return 0;
}
#ifdef CONFIG_FB_3DFX_ACCEL
/*
* FillRect 2D command (solidfill or invert (via ROP_XOR))
*/
static void tdfxfb_fillrect(struct fb_info *info,
const struct fb_fillrect *rect)
{
struct tdfx_par *par = info->par;
u32 bpp = info->var.bits_per_pixel;
u32 stride = info->fix.line_length;
u32 fmt = stride | ((bpp + ((bpp == 8) ? 0 : 8)) << 13);
int tdfx_rop;
u32 dx = rect->dx;
u32 dy = rect->dy;
u32 dstbase = 0;
if (rect->rop == ROP_COPY)
tdfx_rop = TDFX_ROP_COPY;
else
tdfx_rop = TDFX_ROP_XOR;
/* assume always rect->height < 4096 */
if (dy + rect->height > 4095) {
dstbase = stride * dy;
dy = 0;
}
/* assume always rect->width < 4096 */
if (dx + rect->width > 4095) {
dstbase += dx * bpp >> 3;
dx = 0;
}
banshee_make_room(par, 6);
tdfx_outl(par, DSTFORMAT, fmt);
if (info->fix.visual == FB_VISUAL_PSEUDOCOLOR) {
tdfx_outl(par, COLORFORE, rect->color);
} else { /* FB_VISUAL_TRUECOLOR */
tdfx_outl(par, COLORFORE, par->palette[rect->color]);
}
tdfx_outl(par, COMMAND_2D, COMMAND_2D_FILLRECT | (tdfx_rop << 24));
tdfx_outl(par, DSTBASE, dstbase);
tdfx_outl(par, DSTSIZE, rect->width | (rect->height << 16));
tdfx_outl(par, LAUNCH_2D, dx | (dy << 16));
}
/*
* Screen-to-Screen BitBlt 2D command (for the bmove fb op.)
*/
static void tdfxfb_copyarea(struct fb_info *info,
const struct fb_copyarea *area)
{
struct tdfx_par *par = info->par;
u32 sx = area->sx, sy = area->sy, dx = area->dx, dy = area->dy;
u32 bpp = info->var.bits_per_pixel;
u32 stride = info->fix.line_length;
u32 blitcmd = COMMAND_2D_S2S_BITBLT | (TDFX_ROP_COPY << 24);
u32 fmt = stride | ((bpp + ((bpp == 8) ? 0 : 8)) << 13);
u32 dstbase = 0;
u32 srcbase = 0;
/* assume always area->height < 4096 */
if (sy + area->height > 4095) {
srcbase = stride * sy;
sy = 0;
}
/* assume always area->width < 4096 */
if (sx + area->width > 4095) {
srcbase += sx * bpp >> 3;
sx = 0;
}
/* assume always area->height < 4096 */
if (dy + area->height > 4095) {
dstbase = stride * dy;
dy = 0;
}
/* assume always area->width < 4096 */
if (dx + area->width > 4095) {
dstbase += dx * bpp >> 3;
dx = 0;
}
if (area->sx <= area->dx) {
/* -X */
blitcmd |= BIT(14);
sx += area->width - 1;
dx += area->width - 1;
}
if (area->sy <= area->dy) {
/* -Y */
blitcmd |= BIT(15);
sy += area->height - 1;
dy += area->height - 1;
}
banshee_make_room(par, 8);
tdfx_outl(par, SRCFORMAT, fmt);
tdfx_outl(par, DSTFORMAT, fmt);
tdfx_outl(par, COMMAND_2D, blitcmd);
tdfx_outl(par, DSTSIZE, area->width | (area->height << 16));
tdfx_outl(par, DSTXY, dx | (dy << 16));
tdfx_outl(par, SRCBASE, srcbase);
tdfx_outl(par, DSTBASE, dstbase);
tdfx_outl(par, LAUNCH_2D, sx | (sy << 16));
}
static void tdfxfb_imageblit(struct fb_info *info, const struct fb_image *image)
{
struct tdfx_par *par = info->par;
int size = image->height * ((image->width * image->depth + 7) >> 3);
int fifo_free;
int i, stride = info->fix.line_length;
u32 bpp = info->var.bits_per_pixel;
u32 dstfmt = stride | ((bpp + ((bpp == 8) ? 0 : 8)) << 13);
u8 *chardata = (u8 *) image->data;
u32 srcfmt;
u32 dx = image->dx;
u32 dy = image->dy;
u32 dstbase = 0;
if (image->depth != 1) {
#ifdef BROKEN_CODE
banshee_make_room(par, 6 + ((size + 3) >> 2));
srcfmt = stride | ((bpp + ((bpp == 8) ? 0 : 8)) << 13) |
0x400000;
#else
cfb_imageblit(info, image);
#endif
return;
}
banshee_make_room(par, 9);
switch (info->fix.visual) {
case FB_VISUAL_PSEUDOCOLOR:
tdfx_outl(par, COLORFORE, image->fg_color);
tdfx_outl(par, COLORBACK, image->bg_color);
break;
case FB_VISUAL_TRUECOLOR:
default:
tdfx_outl(par, COLORFORE,
par->palette[image->fg_color]);
tdfx_outl(par, COLORBACK,
par->palette[image->bg_color]);
}
#ifdef __BIG_ENDIAN
srcfmt = 0x400000 | BIT(20);
#else
srcfmt = 0x400000;
#endif
/* assume always image->height < 4096 */
if (dy + image->height > 4095) {
dstbase = stride * dy;
dy = 0;
}
/* assume always image->width < 4096 */
if (dx + image->width > 4095) {
dstbase += dx * bpp >> 3;
dx = 0;
}
tdfx_outl(par, DSTBASE, dstbase);
tdfx_outl(par, SRCXY, 0);
tdfx_outl(par, DSTXY, dx | (dy << 16));
tdfx_outl(par, COMMAND_2D,
COMMAND_2D_H2S_BITBLT | (TDFX_ROP_COPY << 24));
tdfx_outl(par, SRCFORMAT, srcfmt);
tdfx_outl(par, DSTFORMAT, dstfmt);
tdfx_outl(par, DSTSIZE, image->width | (image->height << 16));
/* A count of how many free FIFO entries we've requested.
* When this goes negative, we need to request more. */
fifo_free = 0;
/* Send four bytes at a time of data */
for (i = (size >> 2); i > 0; i--) {
if (--fifo_free < 0) {
fifo_free = 31;
banshee_make_room(par, fifo_free);
}
tdfx_outl(par, LAUNCH_2D, *(u32 *)chardata);
chardata += 4;
}
/* Send the leftovers now */
banshee_make_room(par, 3);
switch (size % 4) {
case 0:
break;
case 1:
tdfx_outl(par, LAUNCH_2D, *chardata);
break;
case 2:
tdfx_outl(par, LAUNCH_2D, *(u16 *)chardata);
break;
case 3:
tdfx_outl(par, LAUNCH_2D,
*(u16 *)chardata | (chardata[3] << 24));
break;
}
}
#endif /* CONFIG_FB_3DFX_ACCEL */
static int tdfxfb_cursor(struct fb_info *info, struct fb_cursor *cursor)
{
struct tdfx_par *par = info->par;
u32 vidcfg;
if (!hwcursor)
return -EINVAL; /* just to force soft_cursor() call */
/* Too large of a cursor or wrong bpp :-( */
if (cursor->image.width > 64 ||
cursor->image.height > 64 ||
cursor->image.depth > 1)
return -EINVAL;
vidcfg = tdfx_inl(par, VIDPROCCFG);
if (cursor->enable)
tdfx_outl(par, VIDPROCCFG, vidcfg | VIDCFG_HWCURSOR_ENABLE);
else
tdfx_outl(par, VIDPROCCFG, vidcfg & ~VIDCFG_HWCURSOR_ENABLE);
/*
* If the cursor is not be changed this means either we want the
* current cursor state (if enable is set) or we want to query what
* we can do with the cursor (if enable is not set)
*/
if (!cursor->set)
return 0;
/* fix cursor color - XFree86 forgets to restore it properly */
if (cursor->set & FB_CUR_SETCMAP) {
struct fb_cmap cmap = info->cmap;
u32 bg_idx = cursor->image.bg_color;
u32 fg_idx = cursor->image.fg_color;
unsigned long bg_color, fg_color;
fg_color = (((u32)cmap.red[fg_idx] & 0xff00) << 8) |
(((u32)cmap.green[fg_idx] & 0xff00) << 0) |
(((u32)cmap.blue[fg_idx] & 0xff00) >> 8);
bg_color = (((u32)cmap.red[bg_idx] & 0xff00) << 8) |
(((u32)cmap.green[bg_idx] & 0xff00) << 0) |
(((u32)cmap.blue[bg_idx] & 0xff00) >> 8);
banshee_make_room(par, 2);
tdfx_outl(par, HWCURC0, bg_color);
tdfx_outl(par, HWCURC1, fg_color);
}
if (cursor->set & FB_CUR_SETPOS) {
int x = cursor->image.dx;
int y = cursor->image.dy - info->var.yoffset;
x += 63;
y += 63;
banshee_make_room(par, 1);
tdfx_outl(par, HWCURLOC, (y << 16) + x);
}
if (cursor->set & (FB_CUR_SETIMAGE | FB_CUR_SETSHAPE)) {
/*
* Voodoo 3 and above cards use 2 monochrome cursor patterns.
* The reason is so the card can fetch 8 words at a time
* and are stored on chip for use for the next 8 scanlines.
* This reduces the number of times for access to draw the
* cursor for each screen refresh.
* Each pattern is a bitmap of 64 bit wide and 64 bit high
* (total of 8192 bits or 1024 bytes). The two patterns are
* stored in such a way that pattern 0 always resides in the
* lower half (least significant 64 bits) of a 128 bit word
* and pattern 1 the upper half. If you examine the data of
* the cursor image the graphics card uses then from the
* beginning you see line one of pattern 0, line one of
* pattern 1, line two of pattern 0, line two of pattern 1,
* etc etc. The linear stride for the cursor is always 16 bytes
* (128 bits) which is the maximum cursor width times two for
* the two monochrome patterns.
*/
u8 __iomem *cursorbase = info->screen_base + info->fix.smem_len;
u8 *bitmap = (u8 *)cursor->image.data;
u8 *mask = (u8 *)cursor->mask;
int i;
fb_memset(cursorbase, 0, 1024);
for (i = 0; i < cursor->image.height; i++) {
int h = 0;
int j = (cursor->image.width + 7) >> 3;
for (; j > 0; j--) {
u8 data = *mask ^ *bitmap;
if (cursor->rop == ROP_COPY)
data = *mask & *bitmap;
/* Pattern 0. Copy the cursor mask to it */
fb_writeb(*mask, cursorbase + h);
mask++;
/* Pattern 1. Copy the cursor bitmap to it */
fb_writeb(data, cursorbase + h + 8);
bitmap++;
h++;
}
cursorbase += 16;
}
}
return 0;
}
static struct fb_ops tdfxfb_ops = {
.owner = THIS_MODULE,
.fb_check_var = tdfxfb_check_var,
.fb_set_par = tdfxfb_set_par,
.fb_setcolreg = tdfxfb_setcolreg,
.fb_blank = tdfxfb_blank,
.fb_pan_display = tdfxfb_pan_display,
.fb_sync = banshee_wait_idle,
.fb_cursor = tdfxfb_cursor,
#ifdef CONFIG_FB_3DFX_ACCEL
.fb_fillrect = tdfxfb_fillrect,
.fb_copyarea = tdfxfb_copyarea,
.fb_imageblit = tdfxfb_imageblit,
#else
.fb_fillrect = cfb_fillrect,
.fb_copyarea = cfb_copyarea,
.fb_imageblit = cfb_imageblit,
#endif
};
#ifdef CONFIG_FB_3DFX_I2C
/* The voo GPIO registers don't have individual masks for each bit
so we always have to read before writing. */
static void tdfxfb_i2c_setscl(void *data, int val)
{
struct tdfxfb_i2c_chan *chan = data;
struct tdfx_par *par = chan->par;
unsigned int r;
r = tdfx_inl(par, VIDSERPARPORT);
if (val)
r |= I2C_SCL_OUT;
else
r &= ~I2C_SCL_OUT;
tdfx_outl(par, VIDSERPARPORT, r);
tdfx_inl(par, VIDSERPARPORT); /* flush posted write */
}
static void tdfxfb_i2c_setsda(void *data, int val)
{
struct tdfxfb_i2c_chan *chan = data;
struct tdfx_par *par = chan->par;
unsigned int r;
r = tdfx_inl(par, VIDSERPARPORT);
if (val)
r |= I2C_SDA_OUT;
else
r &= ~I2C_SDA_OUT;
tdfx_outl(par, VIDSERPARPORT, r);
tdfx_inl(par, VIDSERPARPORT); /* flush posted write */
}
/* The GPIO pins are open drain, so the pins always remain outputs.
We rely on the i2c-algo-bit routines to set the pins high before
reading the input from other chips. */
static int tdfxfb_i2c_getscl(void *data)
{
struct tdfxfb_i2c_chan *chan = data;
struct tdfx_par *par = chan->par;
return (0 != (tdfx_inl(par, VIDSERPARPORT) & I2C_SCL_IN));
}
static int tdfxfb_i2c_getsda(void *data)
{
struct tdfxfb_i2c_chan *chan = data;
struct tdfx_par *par = chan->par;
return (0 != (tdfx_inl(par, VIDSERPARPORT) & I2C_SDA_IN));
}
static void tdfxfb_ddc_setscl(void *data, int val)
{
struct tdfxfb_i2c_chan *chan = data;
struct tdfx_par *par = chan->par;
unsigned int r;
r = tdfx_inl(par, VIDSERPARPORT);
if (val)
r |= DDC_SCL_OUT;
else
r &= ~DDC_SCL_OUT;
tdfx_outl(par, VIDSERPARPORT, r);
tdfx_inl(par, VIDSERPARPORT); /* flush posted write */
}
static void tdfxfb_ddc_setsda(void *data, int val)
{
struct tdfxfb_i2c_chan *chan = data;
struct tdfx_par *par = chan->par;
unsigned int r;
r = tdfx_inl(par, VIDSERPARPORT);
if (val)
r |= DDC_SDA_OUT;
else
r &= ~DDC_SDA_OUT;
tdfx_outl(par, VIDSERPARPORT, r);
tdfx_inl(par, VIDSERPARPORT); /* flush posted write */
}
static int tdfxfb_ddc_getscl(void *data)
{
struct tdfxfb_i2c_chan *chan = data;
struct tdfx_par *par = chan->par;
return (0 != (tdfx_inl(par, VIDSERPARPORT) & DDC_SCL_IN));
}
static int tdfxfb_ddc_getsda(void *data)
{
struct tdfxfb_i2c_chan *chan = data;
struct tdfx_par *par = chan->par;
return (0 != (tdfx_inl(par, VIDSERPARPORT) & DDC_SDA_IN));
}
static int tdfxfb_setup_ddc_bus(struct tdfxfb_i2c_chan *chan, const char *name,
struct device *dev)
{
int rc;
strlcpy(chan->adapter.name, name, sizeof(chan->adapter.name));
chan->adapter.owner = THIS_MODULE;
chan->adapter.class = I2C_CLASS_DDC;
chan->adapter.algo_data = &chan->algo;
chan->adapter.dev.parent = dev;
chan->algo.setsda = tdfxfb_ddc_setsda;
chan->algo.setscl = tdfxfb_ddc_setscl;
chan->algo.getsda = tdfxfb_ddc_getsda;
chan->algo.getscl = tdfxfb_ddc_getscl;
chan->algo.udelay = 10;
chan->algo.timeout = msecs_to_jiffies(500);
chan->algo.data = chan;
i2c_set_adapdata(&chan->adapter, chan);
rc = i2c_bit_add_bus(&chan->adapter);
if (rc == 0)
DPRINTK("I2C bus %s registered.\n", name);
else
chan->par = NULL;
return rc;
}
static int tdfxfb_setup_i2c_bus(struct tdfxfb_i2c_chan *chan, const char *name,
struct device *dev)
{
int rc;
strlcpy(chan->adapter.name, name, sizeof(chan->adapter.name));
chan->adapter.owner = THIS_MODULE;
chan->adapter.algo_data = &chan->algo;
chan->adapter.dev.parent = dev;
chan->algo.setsda = tdfxfb_i2c_setsda;
chan->algo.setscl = tdfxfb_i2c_setscl;
chan->algo.getsda = tdfxfb_i2c_getsda;
chan->algo.getscl = tdfxfb_i2c_getscl;
chan->algo.udelay = 10;
chan->algo.timeout = msecs_to_jiffies(500);
chan->algo.data = chan;
i2c_set_adapdata(&chan->adapter, chan);
rc = i2c_bit_add_bus(&chan->adapter);
if (rc == 0)
DPRINTK("I2C bus %s registered.\n", name);
else
chan->par = NULL;
return rc;
}
static void tdfxfb_create_i2c_busses(struct fb_info *info)
{
struct tdfx_par *par = info->par;
tdfx_outl(par, VIDINFORMAT, 0x8160);
tdfx_outl(par, VIDSERPARPORT, 0xcffc0020);
par->chan[0].par = par;
par->chan[1].par = par;
tdfxfb_setup_ddc_bus(&par->chan[0], "Voodoo3-DDC", info->dev);
tdfxfb_setup_i2c_bus(&par->chan[1], "Voodoo3-I2C", info->dev);
}
static void tdfxfb_delete_i2c_busses(struct tdfx_par *par)
{
if (par->chan[0].par)
i2c_del_adapter(&par->chan[0].adapter);
par->chan[0].par = NULL;
if (par->chan[1].par)
i2c_del_adapter(&par->chan[1].adapter);
par->chan[1].par = NULL;
}
static int tdfxfb_probe_i2c_connector(struct tdfx_par *par,
struct fb_monspecs *specs)
{
u8 *edid = NULL;
DPRINTK("Probe DDC Bus\n");
if (par->chan[0].par)
edid = fb_ddc_read(&par->chan[0].adapter);
if (edid) {
fb_edid_to_monspecs(edid, specs);
kfree(edid);
return 0;
}
return 1;
}
#endif /* CONFIG_FB_3DFX_I2C */
/**
* tdfxfb_probe - Device Initializiation
*
* @pdev: PCI Device to initialize
* @id: PCI Device ID
*
* Initializes and allocates resources for PCI device @pdev.
*
*/
static int tdfxfb_probe(struct pci_dev *pdev, const struct pci_device_id *id)
{
struct tdfx_par *default_par;
struct fb_info *info;
int err, lpitch;
struct fb_monspecs *specs;
bool found;
err = pci_enable_device(pdev);
if (err) {
printk(KERN_ERR "tdfxfb: Can't enable pdev: %d\n", err);
return err;
}
info = framebuffer_alloc(sizeof(struct tdfx_par), &pdev->dev);
if (!info)
return -ENOMEM;
default_par = info->par;
info->fix = tdfx_fix;
/* Configure the default fb_fix_screeninfo first */
switch (pdev->device) {
case PCI_DEVICE_ID_3DFX_BANSHEE:
strcpy(info->fix.id, "3Dfx Banshee");
default_par->max_pixclock = BANSHEE_MAX_PIXCLOCK;
break;
case PCI_DEVICE_ID_3DFX_VOODOO3:
strcpy(info->fix.id, "3Dfx Voodoo3");
default_par->max_pixclock = VOODOO3_MAX_PIXCLOCK;
break;
case PCI_DEVICE_ID_3DFX_VOODOO5:
strcpy(info->fix.id, "3Dfx Voodoo5");
default_par->max_pixclock = VOODOO5_MAX_PIXCLOCK;
break;
}
info->fix.mmio_start = pci_resource_start(pdev, 0);
info->fix.mmio_len = pci_resource_len(pdev, 0);
if (!request_mem_region(info->fix.mmio_start, info->fix.mmio_len,
"tdfx regbase")) {
printk(KERN_ERR "tdfxfb: Can't reserve regbase\n");
goto out_err;
}
default_par->regbase_virt =
ioremap_nocache(info->fix.mmio_start, info->fix.mmio_len);
if (!default_par->regbase_virt) {
printk(KERN_ERR "fb: Can't remap %s register area.\n",
info->fix.id);
goto out_err_regbase;
}
info->fix.smem_start = pci_resource_start(pdev, 1);
info->fix.smem_len = do_lfb_size(default_par, pdev->device);
if (!info->fix.smem_len) {
printk(KERN_ERR "fb: Can't count %s memory.\n", info->fix.id);
goto out_err_regbase;
}
if (!request_mem_region(info->fix.smem_start,
pci_resource_len(pdev, 1), "tdfx smem")) {
printk(KERN_ERR "tdfxfb: Can't reserve smem\n");
goto out_err_regbase;
}
video: fbdev: tdfxfb: use arch_phys_wc_add() and ioremap_wc() This driver uses the same area for MTRR as for the ioremap(). 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: Rob Clark <robdclark@gmail.com> Cc: Geert Uytterhoeven <geert@linux-m68k.org> 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: Jean-Christophe Plagniol-Villard <plagnioj@jcrosoft.com> Cc: Tomi Valkeinen <tomi.valkeinen@ti.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 04:16:36 +08:00
info->screen_base = ioremap_wc(info->fix.smem_start,
info->fix.smem_len);
if (!info->screen_base) {
printk(KERN_ERR "fb: Can't remap %s framebuffer.\n",
info->fix.id);
goto out_err_screenbase;
}
default_par->iobase = pci_resource_start(pdev, 2);
if (!request_region(pci_resource_start(pdev, 2),
pci_resource_len(pdev, 2), "tdfx iobase")) {
printk(KERN_ERR "tdfxfb: Can't reserve iobase\n");
goto out_err_screenbase;
}
printk(KERN_INFO "fb: %s memory = %dK\n", info->fix.id,
info->fix.smem_len >> 10);
if (!nomtrr)
video: fbdev: tdfxfb: use arch_phys_wc_add() and ioremap_wc() This driver uses the same area for MTRR as for the ioremap(). 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: Rob Clark <robdclark@gmail.com> Cc: Geert Uytterhoeven <geert@linux-m68k.org> 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: Jean-Christophe Plagniol-Villard <plagnioj@jcrosoft.com> Cc: Tomi Valkeinen <tomi.valkeinen@ti.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 04:16:36 +08:00
default_par->wc_cookie= arch_phys_wc_add(info->fix.smem_start,
info->fix.smem_len);
info->fix.ypanstep = nopan ? 0 : 1;
info->fix.ywrapstep = nowrap ? 0 : 1;
info->fbops = &tdfxfb_ops;
info->pseudo_palette = default_par->palette;
info->flags = FBINFO_DEFAULT | FBINFO_HWACCEL_YPAN;
#ifdef CONFIG_FB_3DFX_ACCEL
info->flags |= FBINFO_HWACCEL_FILLRECT |
FBINFO_HWACCEL_COPYAREA |
FBINFO_HWACCEL_IMAGEBLIT |
FBINFO_READS_FAST;
#endif
/* reserve 8192 bits for cursor */
/* the 2.4 driver says PAGE_MASK boundary is not enough for Voodoo4 */
if (hwcursor)
info->fix.smem_len = (info->fix.smem_len - 1024) &
(PAGE_MASK << 1);
specs = &info->monspecs;
found = false;
info->var.bits_per_pixel = 8;
#ifdef CONFIG_FB_3DFX_I2C
tdfxfb_create_i2c_busses(info);
err = tdfxfb_probe_i2c_connector(default_par, specs);
if (!err) {
if (specs->modedb == NULL)
DPRINTK("Unable to get Mode Database\n");
else {
const struct fb_videomode *m;
fb_videomode_to_modelist(specs->modedb,
specs->modedb_len,
&info->modelist);
m = fb_find_best_display(specs, &info->modelist);
if (m) {
fb_videomode_to_var(&info->var, m);
/* fill all other info->var's fields */
if (tdfxfb_check_var(&info->var, info) < 0)
info->var = tdfx_var;
else
found = true;
}
}
}
#endif
if (!mode_option && !found)
mode_option = "640x480@60";
if (mode_option) {
err = fb_find_mode(&info->var, info, mode_option,
specs->modedb, specs->modedb_len,
NULL, info->var.bits_per_pixel);
if (!err || err == 4)
info->var = tdfx_var;
}
if (found) {
fb_destroy_modedb(specs->modedb);
specs->modedb = NULL;
}
/* maximize virtual vertical length */
lpitch = info->var.xres_virtual * ((info->var.bits_per_pixel + 7) >> 3);
info->var.yres_virtual = info->fix.smem_len / lpitch;
if (info->var.yres_virtual < info->var.yres)
goto out_err_iobase;
if (fb_alloc_cmap(&info->cmap, 256, 0) < 0) {
printk(KERN_ERR "tdfxfb: Can't allocate color map\n");
goto out_err_iobase;
}
if (register_framebuffer(info) < 0) {
printk(KERN_ERR "tdfxfb: can't register framebuffer\n");
fb_dealloc_cmap(&info->cmap);
goto out_err_iobase;
}
/*
* Our driver data
*/
pci_set_drvdata(pdev, info);
return 0;
out_err_iobase:
#ifdef CONFIG_FB_3DFX_I2C
tdfxfb_delete_i2c_busses(default_par);
#endif
video: fbdev: tdfxfb: use arch_phys_wc_add() and ioremap_wc() This driver uses the same area for MTRR as for the ioremap(). 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: Rob Clark <robdclark@gmail.com> Cc: Geert Uytterhoeven <geert@linux-m68k.org> 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: Jean-Christophe Plagniol-Villard <plagnioj@jcrosoft.com> Cc: Tomi Valkeinen <tomi.valkeinen@ti.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 04:16:36 +08:00
arch_phys_wc_del(default_par->wc_cookie);
release_region(pci_resource_start(pdev, 2),
pci_resource_len(pdev, 2));
out_err_screenbase:
if (info->screen_base)
iounmap(info->screen_base);
release_mem_region(info->fix.smem_start, pci_resource_len(pdev, 1));
out_err_regbase:
/*
* Cleanup after anything that was remapped/allocated.
*/
if (default_par->regbase_virt)
iounmap(default_par->regbase_virt);
release_mem_region(info->fix.mmio_start, info->fix.mmio_len);
out_err:
framebuffer_release(info);
return -ENXIO;
}
#ifndef MODULE
static void __init tdfxfb_setup(char *options)
{
char *this_opt;
if (!options || !*options)
return;
while ((this_opt = strsep(&options, ",")) != NULL) {
if (!*this_opt)
continue;
if (!strcmp(this_opt, "nopan")) {
nopan = 1;
} else if (!strcmp(this_opt, "nowrap")) {
nowrap = 1;
} else if (!strncmp(this_opt, "hwcursor=", 9)) {
hwcursor = simple_strtoul(this_opt + 9, NULL, 0);
} else if (!strncmp(this_opt, "nomtrr", 6)) {
nomtrr = 1;
} else {
mode_option = this_opt;
}
}
}
#endif
/**
* tdfxfb_remove - Device removal
*
* @pdev: PCI Device to cleanup
*
* Releases all resources allocated during the course of the driver's
* lifetime for the PCI device @pdev.
*
*/
static void tdfxfb_remove(struct pci_dev *pdev)
{
struct fb_info *info = pci_get_drvdata(pdev);
struct tdfx_par *par = info->par;
unregister_framebuffer(info);
#ifdef CONFIG_FB_3DFX_I2C
tdfxfb_delete_i2c_busses(par);
#endif
video: fbdev: tdfxfb: use arch_phys_wc_add() and ioremap_wc() This driver uses the same area for MTRR as for the ioremap(). 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: Rob Clark <robdclark@gmail.com> Cc: Geert Uytterhoeven <geert@linux-m68k.org> 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: Jean-Christophe Plagniol-Villard <plagnioj@jcrosoft.com> Cc: Tomi Valkeinen <tomi.valkeinen@ti.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 04:16:36 +08:00
arch_phys_wc_del(par->wc_cookie);
iounmap(par->regbase_virt);
iounmap(info->screen_base);
/* Clean up after reserved regions */
release_region(pci_resource_start(pdev, 2),
pci_resource_len(pdev, 2));
release_mem_region(pci_resource_start(pdev, 1),
pci_resource_len(pdev, 1));
release_mem_region(pci_resource_start(pdev, 0),
pci_resource_len(pdev, 0));
fb_dealloc_cmap(&info->cmap);
framebuffer_release(info);
}
static int __init tdfxfb_init(void)
{
#ifndef MODULE
char *option = NULL;
if (fb_get_options("tdfxfb", &option))
return -ENODEV;
tdfxfb_setup(option);
#endif
return pci_register_driver(&tdfxfb_driver);
}
static void __exit tdfxfb_exit(void)
{
pci_unregister_driver(&tdfxfb_driver);
}
MODULE_AUTHOR("Hannu Mallat <hmallat@cc.hut.fi>");
MODULE_DESCRIPTION("3Dfx framebuffer device driver");
MODULE_LICENSE("GPL");
module_param(hwcursor, int, 0644);
MODULE_PARM_DESC(hwcursor, "Enable hardware cursor "
"(1=enable, 0=disable, default=1)");
module_param(mode_option, charp, 0);
MODULE_PARM_DESC(mode_option, "Initial video mode e.g. '648x480-8@60'");
module_param(nomtrr, bool, 0);
MODULE_PARM_DESC(nomtrr, "Disable MTRR support (default: enabled)");
module_init(tdfxfb_init);
module_exit(tdfxfb_exit);