/*
* linux/drivers/video/riva/fbdev.c - nVidia RIVA 128/TNT/TNT2 fb driver
*
* Maintained by Ani Joshi <ajoshi@shell.unixbox.com>
*
* Copyright 1999-2000 Jeff Garzik
*
* Contributors:
*
* Ani Joshi: Lots of debugging and cleanup work, really helped
* get the driver going
*
* Ferenc Bakonyi: Bug fixes, cleanup, modularization
*
* Jindrich Makovicka: Accel code help, hw cursor, mtrr
*
* Paul Richards: Bug fixes, updates
*
* Initial template from skeletonfb.c, created 28 Dec 1997 by Geert Uytterhoeven
* Includes riva_hw.c from nVidia, see copyright below.
* KGI code provided the basis for state storage, init, and mode switching.
*
* 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.
*
* Known bugs and issues:
* restoring text mode fails
* doublescan modes are broken
*/
#include <linuxcomp.h>
#include <linux/config.h>
#include <linux/module.h>
#include <linux/kernel.h>
#include <linux/errno.h>
#include <linux/string.h>
#include <linux/mm.h>
#include <linux/tty.h>
#include <linux/slab.h>
#include <linux/delay.h>
#include <linux/fb.h>
#include <linux/init.h>
#include <linux/pci.h>
#ifdef CONFIG_MTRR
#include <asm/mtrr.h>
#endif
#include "rivafb.h"
#include "nvreg.h"
#ifndef CONFIG_PCI /* sanity check */
#error This driver requires PCI support.
#endif
/* version number of this driver */
#define RIVAFB_VERSION "0.9.5b"
/* ------------------------------------------------------------------------- *
*
* various helpful macros and constants
*
* ------------------------------------------------------------------------- */
#define RIVAFBDEBUG
#ifdef RIVAFBDEBUG
#define DPRINTK(fmt, args...) printk(KERN_DEBUG "%s: " fmt, __FUNCTION__ , ## args)
#else
#define DPRINTK(fmt, args...)
#endif
#ifndef RIVA_NDEBUG
#define assert(expr) \
if(!(expr)) { \
printk( "Assertion failed! %s,%s,%s,line=%d\n",\
#expr,__FILE__,__FUNCTION__,__LINE__); \
BUG(); \
}
#else
#define assert(expr)
#endif
#define PFX "rivafb: "
/* macro that allows you to set overflow bits */
#define SetBitField(value,from,to) SetBF(to,GetBF(value,from))
#define SetBit(n) (1<<(n))
#define Set8Bits(value) ((value)&0xff)
/* HW cursor parameters */
#define MAX_CURS 32
/* ------------------------------------------------------------------------- *
*
* prototypes
*
* ------------------------------------------------------------------------- */
static int rivafb_blank
(int blank
, struct fb_info
*info
);
/* ------------------------------------------------------------------------- *
*
* card identification
*
* ------------------------------------------------------------------------- */
enum riva_chips
{
CH_RIVA_128
= 0,
CH_RIVA_TNT
,
CH_RIVA_TNT2
,
CH_RIVA_UTNT2
,
CH_RIVA_VTNT2
,
CH_RIVA_UVTNT2
,
CH_RIVA_ITNT2
,
CH_GEFORCE_SDR
,
CH_GEFORCE_DDR
,
CH_QUADRO
,
CH_GEFORCE2_MX
,
CH_GEFORCE2_MX2
,
CH_GEFORCE2_GO
,
CH_QUADRO2_MXR
,
CH_GEFORCE2_GTS
,
CH_GEFORCE2_GTS2
,
CH_GEFORCE2_ULTRA
,
CH_QUADRO2_PRO
,
CH_TEST
,
CH_GEFORCE4_MX_460
,
CH_GEFORCE4_MX_440
,
CH_GEFORCE4_MX_420
,
CH_GEFORCE4_440_GO
,
CH_GEFORCE4_420_GO
,
CH_GEFORCE4_420_GO_M32
,
CH_QUADRO4_500XGL
,
CH_GEFORCE4_440_GO_M64
,
CH_QUADRO4_200
,
CH_QUADRO4_550XGL
,
CH_QUADRO4_500_GOGL
,
CH_IGEFORCE2
,
CH_GEFORCE3
,
CH_GEFORCE3_1
,
CH_GEFORCE3_2
,
CH_QUADRO_DDC
,
CH_GEFORCE4_TI_4600
,
CH_GEFORCE4_TI_4400
,
CH_GEFORCE4_TI_4200
,
CH_QUADRO4_900XGL
,
CH_QUADRO4_750XGL
,
CH_QUADRO4_700XGL
};
/* directly indexed by riva_chips enum, above */
static struct riva_chip_info
{
const char *name
;
unsigned arch_rev
;
} riva_chip_info
[] __initdata
= {
{ "RIVA-128", NV_ARCH_03
},
{ "RIVA-TNT", NV_ARCH_04
},
{ "RIVA-TNT2", NV_ARCH_04
},
{ "RIVA-UTNT2", NV_ARCH_04
},
{ "RIVA-VTNT2", NV_ARCH_04
},
{ "RIVA-UVTNT2", NV_ARCH_04
},
{ "RIVA-ITNT2", NV_ARCH_04
},
{ "GeForce-SDR", NV_ARCH_10
},
{ "GeForce-DDR", NV_ARCH_10
},
{ "Quadro", NV_ARCH_10
},
{ "GeForce2-MX", NV_ARCH_10
},
{ "GeForce2-MX", NV_ARCH_10
},
{ "GeForce2-GO", NV_ARCH_10
},
{ "Quadro2-MXR", NV_ARCH_10
},
{ "GeForce2-GTS", NV_ARCH_10
},
{ "GeForce2-GTS", NV_ARCH_10
},
{ "GeForce2-ULTRA", NV_ARCH_10
},
{ "Quadro2-PRO", NV_ARCH_10
},
{ "TEST",NV_ARCH_20
},
{ "GeForce4-MX-460", NV_ARCH_20
},
{ "GeForce4-MX-440", NV_ARCH_20
},
{ "GeForce4-MX-420", NV_ARCH_20
},
{ "GeForce4-440-GO", NV_ARCH_20
},
{ "GeForce4-420-GO", NV_ARCH_20
},
{ "GeForce4-420-GO-M32", NV_ARCH_20
},
{ "Quadro4-500-XGL", NV_ARCH_20
},
{ "GeForce4-440-GO-M64", NV_ARCH_20
},
{ "Quadro4-200", NV_ARCH_20
},
{ "Quadro4-550-XGL", NV_ARCH_20
},
{ "Quadro4-500-GOGL", NV_ARCH_20
},
{ "GeForce2", NV_ARCH_20
},
{ "GeForce3", NV_ARCH_20
},
{ "GeForce3 Ti 200", NV_ARCH_20
},
{ "GeForce3 Ti 500", NV_ARCH_20
},
{ "Quadro DDC", NV_ARCH_20
},
{ "GeForce4 Ti 4600", NV_ARCH_20
},
{ "GeForce4 Ti 4400", NV_ARCH_20
},
{ "GeForce4 Ti 4200", NV_ARCH_20
},
{ "Quadro4-900-XGL", NV_ARCH_20
},
{ "Quadro4-750-XGL", NV_ARCH_20
},
{ "Quadro4-700-XGL", NV_ARCH_20
}
};
static struct pci_device_id rivafb_pci_tbl
[] = {
{ PCI_VENDOR_ID_NVIDIA_SGS
, PCI_DEVICE_ID_NVIDIA_SGS_RIVA128
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_RIVA_128
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_TNT
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_RIVA_TNT
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_TNT2
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_RIVA_TNT2
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_UTNT2
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_RIVA_UTNT2
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_VTNT2
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_RIVA_VTNT2
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_UVTNT2
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_RIVA_VTNT2
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_ITNT2
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_RIVA_ITNT2
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_GEFORCE_SDR
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_GEFORCE_SDR
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_GEFORCE_DDR
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_GEFORCE_DDR
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_QUADRO
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_QUADRO
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_GEFORCE2_MX
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_GEFORCE2_MX
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_GEFORCE2_MX2
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_GEFORCE2_MX2
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_GEFORCE2_GO
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_GEFORCE2_GO
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_QUADRO2_MXR
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_QUADRO2_MXR
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_GEFORCE2_GTS
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_GEFORCE2_GTS
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_GEFORCE2_GTS2
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_GEFORCE2_GTS2
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_GEFORCE2_ULTRA
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_GEFORCE2_ULTRA
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_QUADRO2_PRO
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_QUADRO2_PRO
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_GEFORCE4_MX_460
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_GEFORCE4_MX_460
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_GEFORCE4_MX_440
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_GEFORCE4_MX_440
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_GEFORCE4_MX_420
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_GEFORCE4_MX_420
},
{ PCI_VENDOR_ID_NVIDIA
, 0x0182,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_TEST
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_GEFORCE4_440_GO
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_GEFORCE4_440_GO
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_GEFORCE4_420_GO
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_GEFORCE4_420_GO
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_GEFORCE4_420_GO_M32
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_GEFORCE4_420_GO_M32
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_QUADRO4_500XGL
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_QUADRO4_500XGL
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_GEFORCE4_440_GO_M64
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_GEFORCE4_440_GO_M64
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_QUADRO4_200
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_QUADRO4_200
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_QUADRO4_550XGL
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_QUADRO4_550XGL
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_QUADRO4_500_GOGL
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_QUADRO4_500_GOGL
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_IGEFORCE2
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_IGEFORCE2
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_GEFORCE3
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_GEFORCE3
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_GEFORCE3_1
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_GEFORCE3_1
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_GEFORCE3_2
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_GEFORCE3_2
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_QUADRO_DDC
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_QUADRO_DDC
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_GEFORCE4_TI_4600
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_GEFORCE4_TI_4600
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_GEFORCE4_TI_4400
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_GEFORCE4_TI_4400
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_GEFORCE4_TI_4200
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_GEFORCE4_TI_4200
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_QUADRO4_900XGL
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_QUADRO4_900XGL
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_QUADRO4_750XGL
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_QUADRO4_750XGL
},
{ PCI_VENDOR_ID_NVIDIA
, PCI_DEVICE_ID_NVIDIA_QUADRO4_700XGL
,
PCI_ANY_ID
, PCI_ANY_ID
, 0, 0, CH_QUADRO4_700XGL
},
{ 0, } /* terminate list */
};
MODULE_DEVICE_TABLE
(pci
, rivafb_pci_tbl
);
/* ------------------------------------------------------------------------- *
*
* global variables
*
* ------------------------------------------------------------------------- */
/* command line data, set in rivafb_setup() */
static u32 pseudo_palette
[17];
static int flatpanel __initdata
= -1; /* Autodetect later */
static int forceCRTC __initdata
= -1;
#ifdef CONFIG_MTRR
static int nomtrr __initdata
= 0;
#endif
#ifndef MODULE
static char *mode_option __initdata
= NULL
;
#endif
static struct fb_fix_screeninfo rivafb_fix
= {
.
id = "nVidia",
.
type = FB_TYPE_PACKED_PIXELS
,
.
xpanstep = 1,
.
ypanstep = 1,
};
static struct fb_var_screeninfo rivafb_default_var
= {
.
xres = 640,
.
yres = 480,
.
xres_virtual = 640,
.
yres_virtual = 480,
.
bits_per_pixel = 8,
.
red = {0, 8, 0},
.
green = {0, 8, 0},
.
blue = {0, 8, 0},
.
transp = {0, 0, 0},
.
activate = FB_ACTIVATE_NOW
,
.
height = -1,
.
width = -1,
.
accel_flags = FB_ACCELF_TEXT
,
.
pixclock = 39721,
.
left_margin = 40,
.
right_margin = 24,
.
upper_margin = 32,
.
lower_margin = 11,
.
hsync_len = 96,
.
vsync_len = 2,
.
vmode = FB_VMODE_NONINTERLACED
};
/* from GGI */
static const struct riva_regs reg_template
= {
{0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, /* ATTR */
0x08, 0x09, 0x0A, 0x0B, 0x0C, 0x0D, 0x0E, 0x0F,
0x41, 0x01, 0x0F, 0x00, 0x00},
{0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* CRT */
0x00, 0x00, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xE3, /* 0x10 */
0xFF, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x20 */
0x40, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x30 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, /* 0x40 */
},
{0x00, 0x00, 0x00, 0x00, 0x00, 0x40, 0x05, 0x0F, /* GRA */
0xFF},
{0x03, 0x01, 0x0F, 0x00, 0x0E}, /* SEQ */
0xEB /* MISC */
};
/* ------------------------------------------------------------------------- *
*
* MMIO access macros
*
* ------------------------------------------------------------------------- */
static inline void CRTCout
(struct riva_par
*par
, unsigned char index
,
unsigned char val
)
{
VGA_WR08
(par
->riva.
PCIO, 0x3d4, index
);
VGA_WR08
(par
->riva.
PCIO, 0x3d5, val
);
}
static inline unsigned char CRTCin
(struct riva_par
*par
,
unsigned char index
)
{
VGA_WR08
(par
->riva.
PCIO, 0x3d4, index
);
return (VGA_RD08
(par
->riva.
PCIO, 0x3d5));
}
static inline void GRAout
(struct riva_par
*par
, unsigned char index
,
unsigned char val
)
{
VGA_WR08
(par
->riva.
PVIO, 0x3ce, index
);
VGA_WR08
(par
->riva.
PVIO, 0x3cf, val
);
}
static inline unsigned char GRAin
(struct riva_par
*par
,
unsigned char index
)
{
VGA_WR08
(par
->riva.
PVIO, 0x3ce, index
);
return (VGA_RD08
(par
->riva.
PVIO, 0x3cf));
}
static inline void SEQout
(struct riva_par
*par
, unsigned char index
,
unsigned char val
)
{
VGA_WR08
(par
->riva.
PVIO, 0x3c4, index
);
VGA_WR08
(par
->riva.
PVIO, 0x3c5, val
);
}
static inline unsigned char SEQin
(struct riva_par
*par
,
unsigned char index
)
{
VGA_WR08
(par
->riva.
PVIO, 0x3c4, index
);
return (VGA_RD08
(par
->riva.
PVIO, 0x3c5));
}
static inline void ATTRout
(struct riva_par
*par
, unsigned char index
,
unsigned char val
)
{
VGA_WR08
(par
->riva.
PCIO, 0x3c0, index
);
VGA_WR08
(par
->riva.
PCIO, 0x3c0, val
);
}
static inline unsigned char ATTRin
(struct riva_par
*par
,
unsigned char index
)
{
VGA_WR08
(par
->riva.
PCIO, 0x3c0, index
);
return (VGA_RD08
(par
->riva.
PCIO, 0x3c1));
}
static inline void MISCout
(struct riva_par
*par
, unsigned char val
)
{
VGA_WR08
(par
->riva.
PVIO, 0x3c2, val
);
}
static inline unsigned char MISCin
(struct riva_par
*par
)
{
return (VGA_RD08
(par
->riva.
PVIO, 0x3cc));
}
static u8 byte_rev
[256] = {
0x00, 0x80, 0x40, 0xc0, 0x20, 0xa0, 0x60, 0xe0,
0x10, 0x90, 0x50, 0xd0, 0x30, 0xb0, 0x70, 0xf0,
0x08, 0x88, 0x48, 0xc8, 0x28, 0xa8, 0x68, 0xe8,
0x18, 0x98, 0x58, 0xd8, 0x38, 0xb8, 0x78, 0xf8,
0x04, 0x84, 0x44, 0xc4, 0x24, 0xa4, 0x64, 0xe4,
0x14, 0x94, 0x54, 0xd4, 0x34, 0xb4, 0x74, 0xf4,
0x0c, 0x8c, 0x4c, 0xcc, 0x2c, 0xac, 0x6c, 0xec,
0x1c, 0x9c, 0x5c, 0xdc, 0x3c, 0xbc, 0x7c, 0xfc,
0x02, 0x82, 0x42, 0xc2, 0x22, 0xa2, 0x62, 0xe2,
0x12, 0x92, 0x52, 0xd2, 0x32, 0xb2, 0x72, 0xf2,
0x0a, 0x8a, 0x4a, 0xca, 0x2a, 0xaa, 0x6a, 0xea,
0x1a, 0x9a, 0x5a, 0xda, 0x3a, 0xba, 0x7a, 0xfa,
0x06, 0x86, 0x46, 0xc6, 0x26, 0xa6, 0x66, 0xe6,
0x16, 0x96, 0x56, 0xd6, 0x36, 0xb6, 0x76, 0xf6,
0x0e, 0x8e, 0x4e, 0xce, 0x2e, 0xae, 0x6e, 0xee,
0x1e, 0x9e, 0x5e, 0xde, 0x3e, 0xbe, 0x7e, 0xfe,
0x01, 0x81, 0x41, 0xc1, 0x21, 0xa1, 0x61, 0xe1,
0x11, 0x91, 0x51, 0xd1, 0x31, 0xb1, 0x71, 0xf1,
0x09, 0x89, 0x49, 0xc9, 0x29, 0xa9, 0x69, 0xe9,
0x19, 0x99, 0x59, 0xd9, 0x39, 0xb9, 0x79, 0xf9,
0x05, 0x85, 0x45, 0xc5, 0x25, 0xa5, 0x65, 0xe5,
0x15, 0x95, 0x55, 0xd5, 0x35, 0xb5, 0x75, 0xf5,
0x0d, 0x8d, 0x4d, 0xcd, 0x2d, 0xad, 0x6d, 0xed,
0x1d, 0x9d, 0x5d, 0xdd, 0x3d, 0xbd, 0x7d, 0xfd,
0x03, 0x83, 0x43, 0xc3, 0x23, 0xa3, 0x63, 0xe3,
0x13, 0x93, 0x53, 0xd3, 0x33, 0xb3, 0x73, 0xf3,
0x0b, 0x8b, 0x4b, 0xcb, 0x2b, 0xab, 0x6b, 0xeb,
0x1b, 0x9b, 0x5b, 0xdb, 0x3b, 0xbb, 0x7b, 0xfb,
0x07, 0x87, 0x47, 0xc7, 0x27, 0xa7, 0x67, 0xe7,
0x17, 0x97, 0x57, 0xd7, 0x37, 0xb7, 0x77, 0xf7,
0x0f, 0x8f, 0x4f, 0xcf, 0x2f, 0xaf, 0x6f, 0xef,
0x1f, 0x9f, 0x5f, 0xdf, 0x3f, 0xbf, 0x7f, 0xff,
};
static inline void reverse_order
(u32
*l
)
{
u8
*a
= (u8
*)l
;
*a
= byte_rev
[*a
], a
++;
*a
= byte_rev
[*a
], a
++;
*a
= byte_rev
[*a
], a
++;
*a
= byte_rev
[*a
];
}
/* ------------------------------------------------------------------------- *
*
* cursor stuff
*
* ------------------------------------------------------------------------- */
/**
* rivafb_load_cursor_image - load cursor image to hardware
* @data: address to monochrome bitmap (1 = foreground color, 0 = background)
* @par: pointer to private data
* @w: width of cursor image in pixels
* @h: height of cursor image in scanlines
* @bg: background color (ARGB1555) - alpha bit determines opacity
* @fg: foreground color (ARGB1555)
*
* DESCRIPTiON:
* Loads cursor image based on a monochrome source and mask bitmap. The
* image bits determines the color of the pixel, 0 for background, 1 for
* foreground. Only the affected region (as determined by @w and @h
* parameters) will be updated.
*
* CALLED FROM:
* rivafb_cursor()
*/
static void rivafb_load_cursor_image
(struct riva_par
*par
, u8
*data
,
u8
*mask
, u16 bg
, u16 fg
, u32 w
, u32 h
)
{
int i
, j
, k
= 0;
u32 b
, m
, tmp
;
for (i
= 0; i
< h
; i
++) {
b
= *((u32
*)data
)++;
m
= *((u32
*)mask
)++;
reverse_order
(&b
);
for (j
= 0; j
< w
/2; j
++) {
tmp
= 0;
#if defined (__BIG_ENDIAN)
if (m
& (1 << 31)) {
fg
|= 1 << 15;
bg
|= 1 << 15;
}
tmp
= (b
& (1 << 31)) ? fg
<< 16 : bg
<< 16;
b
<<= 1;
m
<<= 1;
if (m
& (1 << 31)) {
fg
|= 1 << 15;
bg
|= 1 << 15;
}
tmp
|= (b
& (1 << 31)) ? fg
: bg
;
b
<<= 1;
m
<<= 1;
#else
if (m
& 1) {
fg
|= 1 << 15;
bg
|= 1 << 15;
}
tmp
= (b
& 1) ? fg
: bg
;
b
>>= 1;
m
>>= 1;
if (m
& 1) {
fg
|= 1 << 15;
bg
|= 1 << 15;
}
tmp
|= (b
& 1) ? fg
<< 16 : bg
<< 16;
b
>>= 1;
m
>>= 1;
#endif
writel
(tmp
, par
->riva.
CURSOR + k
++);
}
k
+= (MAX_CURS
- w
)/2;
}
}
/* ------------------------------------------------------------------------- *
*
* general utility functions
*
* ------------------------------------------------------------------------- */
/**
* riva_wclut - set CLUT entry
* @chip: pointer to RIVA_HW_INST object
* @regnum: register number
* @red: red component
* @green: green component
* @blue: blue component
*
* DESCRIPTION:
* Sets color register @regnum.
*
* CALLED FROM:
* rivafb_setcolreg()
*/
static void riva_wclut
(RIVA_HW_INST
*chip
,
unsigned char regnum
, unsigned char red
,
unsigned char green
, unsigned char blue
)
{
VGA_WR08
(chip
->PDIO
, 0x3c8, regnum
);
VGA_WR08
(chip
->PDIO
, 0x3c9, red
);
VGA_WR08
(chip
->PDIO
, 0x3c9, green
);
VGA_WR08
(chip
->PDIO
, 0x3c9, blue
);
}
/**
* riva_rclut - read fromCLUT register
* @chip: pointer to RIVA_HW_INST object
* @regnum: register number
* @red: red component
* @green: green component
* @blue: blue component
*
* DESCRIPTION:
* Reads red, green, and blue from color register @regnum.
*
* CALLED FROM:
* rivafb_setcolreg()
*/
static void riva_rclut
(RIVA_HW_INST
*chip
,
unsigned char regnum
, unsigned char *red
,
unsigned char *green
, unsigned char *blue
)
{
VGA_WR08
(chip
->PDIO
, 0x3c8, regnum
);
*red
= VGA_RD08
(chip
->PDIO
, 0x3c9);
*green
= VGA_RD08
(chip
->PDIO
, 0x3c9);
*blue
= VGA_RD08
(chip
->PDIO
, 0x3c9);
}
/**
* riva_save_state - saves current chip state
* @par: pointer to riva_par object containing info for current riva board
* @regs: pointer to riva_regs object
*
* DESCRIPTION:
* Saves current chip state to @regs.
*
* CALLED FROM:
* rivafb_probe()
*/
/* from GGI */
static void riva_save_state
(struct riva_par
*par
, struct riva_regs
*regs
)
{
int i
;
par
->riva.
LockUnlock(&par
->riva
, 0);
par
->riva.
UnloadStateExt(&par
->riva
, ®s
->ext
);
regs
->misc_output
= MISCin
(par
);
for (i
= 0; i
< NUM_CRT_REGS
; i
++)
regs
->crtc
[i
] = CRTCin
(par
, i
);
for (i
= 0; i
< NUM_ATC_REGS
; i
++)
regs
->attr
[i
] = ATTRin
(par
, i
);
for (i
= 0; i
< NUM_GRC_REGS
; i
++)
regs
->gra
[i
] = GRAin
(par
, i
);
for (i
= 0; i
< NUM_SEQ_REGS
; i
++)
regs
->seq
[i
] = SEQin
(par
, i
);
}
/**
* riva_load_state - loads current chip state
* @par: pointer to riva_par object containing info for current riva board
* @regs: pointer to riva_regs object
*
* DESCRIPTION:
* Loads chip state from @regs.
*
* CALLED FROM:
* riva_load_video_mode()
* rivafb_probe()
* rivafb_remove()
*/
/* from GGI */
static void riva_load_state
(struct riva_par
*par
, struct riva_regs
*regs
)
{
RIVA_HW_STATE
*state
= ®s
->ext
;
int i
;
CRTCout
(par
, 0x11, 0x00);
par
->riva.
LockUnlock(&par
->riva
, 0);
par
->riva.
LoadStateExt(&par
->riva
, state
);
MISCout
(par
, regs
->misc_output
);
for (i
= 0; i
< NUM_CRT_REGS
; i
++) {
switch (i
) {
case 0x19:
case 0x20 ...
0x40:
break;
default:
CRTCout
(par
, i
, regs
->crtc
[i
]);
}
}
for (i
= 0; i
< NUM_ATC_REGS
; i
++)
ATTRout
(par
, i
, regs
->attr
[i
]);
for (i
= 0; i
< NUM_GRC_REGS
; i
++)
GRAout
(par
, i
, regs
->gra
[i
]);
for (i
= 0; i
< NUM_SEQ_REGS
; i
++)
SEQout
(par
, i
, regs
->seq
[i
]);
}
/**
* riva_load_video_mode - calculate timings
* @info: pointer to fb_info object containing info for current riva board
*
* DESCRIPTION:
* Calculate some timings and then send em off to riva_load_state().
*
* CALLED FROM:
* rivafb_set_par()
*/
static void riva_load_video_mode
(struct fb_info
*info
)
{
int bpp
, width
, hDisplaySize
, hDisplay
, hStart
,
hEnd
, hTotal
, height
, vDisplay
, vStart
, vEnd
, vTotal
, dotClock
;
int hBlankStart
, hBlankEnd
, vBlankStart
, vBlankEnd
;
struct riva_par
*par
= (struct riva_par
*) info
->par
;
struct riva_regs newmode
;
/* time to calculate */
rivafb_blank
(1, info
);
bpp
= info
->var.
bits_per_pixel;
if (bpp
== 16 && info
->var.
green.
length == 5)
bpp
= 15;
width
= info
->var.
xres_virtual;
hDisplaySize
= info
->var.
xres;
hDisplay
= (hDisplaySize
/ 8) - 1;
hStart
= (hDisplaySize
+ info
->var.
right_margin) / 8 - 1;
hEnd
= (hDisplaySize
+ info
->var.
right_margin +
info
->var.
hsync_len) / 8 - 1;
hTotal
= (hDisplaySize
+ info
->var.
right_margin +
info
->var.
hsync_len + info
->var.
left_margin) / 8 - 5;
hBlankStart
= hDisplay
;
hBlankEnd
= hTotal
+ 4;
height
= info
->var.
yres_virtual;
vDisplay
= info
->var.
yres - 1;
vStart
= info
->var.
yres + info
->var.
lower_margin - 1;
vEnd
= info
->var.
yres + info
->var.
lower_margin +
info
->var.
vsync_len - 1;
vTotal
= info
->var.
yres + info
->var.
lower_margin +
info
->var.
vsync_len + info
->var.
upper_margin + 2;
vBlankStart
= vDisplay
;
vBlankEnd
= vTotal
+ 1;
dotClock
= 1000000000 / info
->var.
pixclock;
memcpy(&newmode
, ®_template
, sizeof(struct riva_regs
));
if ((info
->var.
vmode & FB_VMODE_MASK
) == FB_VMODE_INTERLACED
)
vTotal
|= 1;
if (par
->FlatPanel
) {
vStart
= vTotal
- 3;
vEnd
= vTotal
- 2;
vBlankStart
= vStart
;
hStart
= hTotal
- 3;
hEnd
= hTotal
- 2;
hBlankEnd
= hTotal
+ 4;
}
newmode.
crtc[0x0] = Set8Bits
(hTotal
);
newmode.
crtc[0x1] = Set8Bits
(hDisplay
);
newmode.
crtc[0x2] = Set8Bits
(hBlankStart
);
newmode.
crtc[0x3] = SetBitField
(hBlankEnd
, 4: 0, 4:0) | SetBit
(7);
newmode.
crtc[0x4] = Set8Bits
(hStart
);
newmode.
crtc[0x5] = SetBitField
(hBlankEnd
, 5: 5, 7:7)
| SetBitField
(hEnd
, 4: 0, 4:0);
newmode.
crtc[0x6] = SetBitField
(vTotal
, 7: 0, 7:0);
newmode.
crtc[0x7] = SetBitField
(vTotal
, 8: 8, 0:0)
| SetBitField
(vDisplay
, 8: 8, 1:1)
| SetBitField
(vStart
, 8: 8, 2:2)
| SetBitField
(vBlankStart
, 8: 8, 3:3)
| SetBit
(4)
| SetBitField
(vTotal
, 9: 9, 5:5)
| SetBitField
(vDisplay
, 9: 9, 6:6)
| SetBitField
(vStart
, 9: 9, 7:7);
newmode.
crtc[0x9] = SetBitField
(vBlankStart
, 9: 9, 5:5)
| SetBit
(6);
newmode.
crtc[0x10] = Set8Bits
(vStart
);
newmode.
crtc[0x11] = SetBitField
(vEnd
, 3: 0, 3:0)
| SetBit
(5);
newmode.
crtc[0x12] = Set8Bits
(vDisplay
);
newmode.
crtc[0x13] = (width
/ 8) * ((bpp
+ 1) / 8);
newmode.
crtc[0x15] = Set8Bits
(vBlankStart
);
newmode.
crtc[0x16] = Set8Bits
(vBlankEnd
);
newmode.
ext.
screen = SetBitField
(hBlankEnd
,6:6,4:4)
| SetBitField
(vBlankStart
,10:10,3:3)
| SetBitField
(vStart
,10:10,2:2)
| SetBitField
(vDisplay
,10:10,1:1)
| SetBitField
(vTotal
,10:10,0:0);
newmode.
ext.
horiz = SetBitField
(hTotal
,8:8,0:0)
| SetBitField
(hDisplay
,8:8,1:1)
| SetBitField
(hBlankStart
,8:8,2:2)
| SetBitField
(hStart
,8:8,3:3);
newmode.
ext.
extra = SetBitField
(vTotal
,11:11,0:0)
| SetBitField
(vDisplay
,11:11,2:2)
| SetBitField
(vStart
,11:11,4:4)
| SetBitField
(vBlankStart
,11:11,6:6);
if ((info
->var.
vmode & FB_VMODE_MASK
) == FB_VMODE_INTERLACED
) {
int tmp
= (hTotal
>> 1) & ~
1;
newmode.
ext.
interlace = Set8Bits
(tmp
);
newmode.
ext.
horiz |= SetBitField
(tmp
, 8:8,4:4);
} else
newmode.
ext.
interlace = 0xff; /* interlace off */
if (par
->riva.
Architecture >= NV_ARCH_10
)
par
->riva.
CURSOR = (U032
*)(info
->screen_base
+ par
->riva.
CursorStart);
if (info
->var.
sync & FB_SYNC_HOR_HIGH_ACT
)
newmode.
misc_output &= ~
0x40;
else
newmode.
misc_output |= 0x40;
if (info
->var.
sync & FB_SYNC_VERT_HIGH_ACT
)
newmode.
misc_output &= ~
0x80;
else
newmode.
misc_output |= 0x80;
par
->riva.
CalcStateExt(&par
->riva
, &newmode.
ext, bpp
, width
,
hDisplaySize
, height
, dotClock
);
newmode.
ext.
scale = par
->riva.
PRAMDAC[0x00000848/4] & 0xfff000ff;
if (par
->FlatPanel
== 1) {
newmode.
ext.
pixel |= (1 << 7);
newmode.
ext.
scale |= (1 << 8);
}
if (par
->SecondCRTC
) {
newmode.
ext.
head = par
->riva.
PCRTC0[0x00000860/4] & ~
0x00001000;
newmode.
ext.
head2 = par
->riva.
PCRTC0[0x00002860/4] | 0x00001000;
newmode.
ext.
crtcOwner = 3;
newmode.
ext.
pllsel |= 0x20000800;
newmode.
ext.
vpll2 = newmode.
ext.
vpll;
} else if (par
->riva.
twoHeads) {
newmode.
ext.
head = par
->riva.
PCRTC0[0x00000860/4] | 0x00001000;
newmode.
ext.
head2 = par
->riva.
PCRTC0[0x00002860/4] & ~
0x00001000;
newmode.
ext.
crtcOwner = 0;
newmode.
ext.
vpll2 = par
->riva.
PRAMDAC0[0x00000520/4];
}
if (par
->FlatPanel
== 1) {
newmode.
ext.
pixel |= (1 << 7);
newmode.
ext.
scale |= (1 << 8);
}
newmode.
ext.
cursorConfig = 0x02000100;
par
->current_state
= newmode
;
riva_load_state
(par
, &par
->current_state
);
par
->riva.
LockUnlock(&par
->riva
, 0); /* important for HW cursor */
rivafb_blank
(0, info
);
}
/**
* rivafb_do_maximize -
* @info: pointer to fb_info object containing info for current riva board
* @var:
* @nom:
* @den:
*
* DESCRIPTION:
* .
*
* RETURNS:
* -EINVAL on failure, 0 on success
*
*
* CALLED FROM:
* rivafb_check_var()
*/
static int rivafb_do_maximize
(struct fb_info
*info
,
struct fb_var_screeninfo
*var
,
int nom
, int den
)
{
static struct {
int xres
, yres
;
} modes
[] = {
{1600, 1280},
{1280, 1024},
{1024, 768},
{800, 600},
{640, 480},
{-1, -1}
};
int i
;
/* use highest possible virtual resolution */
if (var
->xres_virtual
== -1 && var
->yres_virtual
== -1) {
printk
(KERN_WARNING PFX
"using maximum available virtual resolution\n");
for (i
= 0; modes
[i
].
xres != -1; i
++) {
if (modes
[i
].
xres * nom
/ den
* modes
[i
].
yres <
info
->fix.
smem_len / 2)
break;
}
if (modes
[i
].
xres == -1) {
printk
(KERN_ERR PFX
"could not find a virtual resolution that fits into video memory!!\n");
DPRINTK
("EXIT - EINVAL error\n");
return -EINVAL
;
}
var
->xres_virtual
= modes
[i
].
xres;
var
->yres_virtual
= modes
[i
].
yres;
printk
(KERN_INFO PFX
"virtual resolution set to maximum of %dx%d\n",
var
->xres_virtual
, var
->yres_virtual
);
} else if (var
->xres_virtual
== -1) {
var
->xres_virtual
= (info
->fix.
smem_len * den
/
(nom
* var
->yres_virtual
* 2)) & ~
15;
printk
(KERN_WARNING PFX
"setting virtual X resolution to %d\n", var
->xres_virtual
);
} else if (var
->yres_virtual
== -1) {
var
->xres_virtual
= (var
->xres_virtual
+ 15) & ~
15;
var
->yres_virtual
= info
->fix.
smem_len * den
/
(nom
* var
->xres_virtual
* 2);
printk
(KERN_WARNING PFX
"setting virtual Y resolution to %d\n", var
->yres_virtual
);
} else {
var
->xres_virtual
= (var
->xres_virtual
+ 15) & ~
15;
if (var
->xres_virtual
* nom
/ den
* var
->yres_virtual
> info
->fix.
smem_len) {
printk
(KERN_ERR PFX
"mode %dx%dx%d rejected...resolution too high to fit into video memory!\n",
var
->xres
, var
->yres
, var
->bits_per_pixel
);
DPRINTK
("EXIT - EINVAL error\n");
return -EINVAL
;
}
}
if (var
->xres_virtual
* nom
/ den
>= 8192) {
printk
(KERN_WARNING PFX
"virtual X resolution (%d) is too high, lowering to %d\n",
var
->xres_virtual
, 8192 * den
/ nom
- 16);
var
->xres_virtual
= 8192 * den
/ nom
- 16;
}
if (var
->xres_virtual
< var
->xres
) {
printk
(KERN_ERR PFX
"virtual X resolution (%d) is smaller than real\n", var
->xres_virtual
);
return -EINVAL
;
}
if (var
->yres_virtual
< var
->yres
) {
printk
(KERN_ERR PFX
"virtual Y resolution (%d) is smaller than real\n", var
->yres_virtual
);
return -EINVAL
;
}
return 0;
}
/* acceleration routines */
inline void wait_for_idle
(struct riva_par
*par
)
{
while (par
->riva.
Busy(&par
->riva
));
}
/* set copy ROP, no mask */
static void riva_setup_ROP
(struct riva_par
*par
)
{
RIVA_FIFO_FREE
(par
->riva
, Patt
, 5);
par
->riva.
Patt->Shape
= 0;
par
->riva.
Patt->Color0
= 0xffffffff;
par
->riva.
Patt->Color1
= 0xffffffff;
par
->riva.
Patt->Monochrome
[0] = 0xffffffff;
par
->riva.
Patt->Monochrome
[1] = 0xffffffff;
RIVA_FIFO_FREE
(par
->riva
, Rop
, 1);
par
->riva.
Rop->Rop3
= 0xCC;
}
void riva_setup_accel
(struct riva_par
*par
)
{
RIVA_FIFO_FREE
(par
->riva
, Clip
, 2);
par
->riva.
Clip->TopLeft
= 0x0;
par
->riva.
Clip->WidthHeight
= 0x80008000;
riva_setup_ROP
(par
);
wait_for_idle
(par
);
}
/**
* riva_get_cmap_len - query current color map length
* @var: standard kernel fb changeable data
*
* DESCRIPTION:
* Get current color map length.
*
* RETURNS:
* Length of color map
*
* CALLED FROM:
* rivafb_setcolreg()
*/
static int riva_get_cmap_len
(const struct fb_var_screeninfo
*var
)
{
int rc
= 256; /* reasonable default */
switch (var
->green.
length) {
case 8:
rc
= 256; /* 256 entries (2^8), 8 bpp and RGB8888 */
break;
case 5:
rc
= 32; /* 32 entries (2^5), 16 bpp, RGB555 */
break;
case 6:
rc
= 64; /* 64 entries (2^6), 16 bpp, RGB565 */
break;
default:
/* should not occur */
break;
}
return rc
;
}
/* ------------------------------------------------------------------------- *
*
* framebuffer operations
*
* ------------------------------------------------------------------------- */
static int rivafb_open
(struct fb_info
*info
, int user
)
{
struct riva_par
*par
= (struct riva_par
*) info
->par
;
int cnt
= atomic_read
(&par
->ref_count
);
if (!cnt
) {
memset(&par
->state
, 0, sizeof(struct vgastate
));
par
->state.
flags = VGA_SAVE_MODE
| VGA_SAVE_FONTS
;
/* save the DAC for Riva128 */
if (par
->riva.
Architecture == NV_ARCH_03
)
par
->state.
flags |= VGA_SAVE_CMAP
;
save_vga
(&par
->state
);
RivaGetConfig
(&par
->riva
, par
->Chipset
);
/* vgaHWunlock() + riva unlock (0x7F) */
CRTCout
(par
, 0x11, 0xFF);
par
->riva.
LockUnlock(&par
->riva
, 0);
riva_save_state
(par
, &par
->initial_state
);
}
atomic_inc
(&par
->ref_count
);
return 0;
}
static int rivafb_release
(struct fb_info
*info
, int user
)
{
struct riva_par
*par
= (struct riva_par
*) info
->par
;
int cnt
= atomic_read
(&par
->ref_count
);
if (!cnt
)
return -EINVAL
;
if (cnt
== 1) {
par
->riva.
LockUnlock(&par
->riva
, 0);
par
->riva.
LoadStateExt(&par
->riva
, &par
->initial_state.
ext);
riva_load_state
(par
, &par
->initial_state
);
restore_vga
(&par
->state
);
par
->riva.
LockUnlock(&par
->riva
, 1);
}
atomic_dec
(&par
->ref_count
);
return 0;
}
static int rivafb_check_var
(struct fb_var_screeninfo
*var
, struct fb_info
*info
)
{
int nom
, den
; /* translating from pixels->bytes */
switch (var
->bits_per_pixel
) {
case 1 ...
8:
var
->red.
offset = var
->green.
offset = var
->blue.
offset = 0;
var
->red.
length = var
->green.
length = var
->blue.
length = 8;
var
->bits_per_pixel
= 8;
nom
= den
= 1;
break;
case 9 ...
15:
var
->green.
length = 5;
/* fall through */
case 16:
var
->bits_per_pixel
= 16;
if (var
->green.
length == 5) {
/* 0rrrrrgg gggbbbbb */
var
->red.
offset = 10;
var
->green.
offset = 5;
var
->blue.
offset = 0;
var
->red.
length = 5;
var
->green.
length = 5;
var
->blue.
length = 5;
} else {
/* rrrrrggg gggbbbbb */
var
->red.
offset = 11;
var
->green.
offset = 5;
var
->blue.
offset = 0;
var
->red.
length = 5;
var
->green.
length = 6;
var
->blue.
length = 5;
}
nom
= 2;
den
= 1;
break;
case 17 ...
32:
var
->red.
length = var
->green.
length = var
->blue.
length = 8;
var
->bits_per_pixel
= 32;
var
->red.
offset = 16;
var
->green.
offset = 8;
var
->blue.
offset = 0;
nom
= 4;
den
= 1;
break;
default:
printk
(KERN_ERR PFX
"mode %dx%dx%d rejected...color depth not supported.\n",
var
->xres
, var
->yres
, var
->bits_per_pixel
);
DPRINTK
("EXIT, returning -EINVAL\n");
return -EINVAL
;
}
if (rivafb_do_maximize
(info
, var
, nom
, den
) < 0)
return -EINVAL
;
if (var
->xoffset
< 0)
var
->xoffset
= 0;
if (var
->yoffset
< 0)
var
->yoffset
= 0;
/* truncate xoffset and yoffset to maximum if too high */
if (var
->xoffset
> var
->xres_virtual
- var
->xres
)
var
->xoffset
= var
->xres_virtual
- var
->xres
- 1;
if (var
->yoffset
> var
->yres_virtual
- var
->yres
)
var
->yoffset
= var
->yres_virtual
- var
->yres
- 1;
var
->red.
msb_right =
var
->green.
msb_right =
var
->blue.
msb_right =
var
->transp.
offset = var
->transp.
length = var
->transp.
msb_right = 0;
return 0;
}
static int rivafb_set_par
(struct fb_info
*info
)
{
struct riva_par
*par
= (struct riva_par
*) info
->par
;
riva_load_video_mode
(info
);
riva_setup_accel
(par
);
info
->fix.
line_length = (info
->var.
xres_virtual * (info
->var.
bits_per_pixel >> 3));
info
->fix.
visual = (info
->var.
bits_per_pixel == 8) ?
FB_VISUAL_PSEUDOCOLOR
: FB_VISUAL_DIRECTCOLOR
;
return 0;
}
/**
* rivafb_pan_display
* @var: standard kernel fb changeable data
* @con: TODO
* @info: pointer to fb_info object containing info for current riva board
*
* DESCRIPTION:
* Pan (or wrap, depending on the `vmode' field) the display using the
* `xoffset' and `yoffset' fields of the `var' structure.
* If the values don't fit, return -EINVAL.
*
* This call looks only at xoffset, yoffset and the FB_VMODE_YWRAP flag
*/
static int rivafb_pan_display
(struct fb_var_screeninfo
*var
,
struct fb_info
*info
)
{
struct riva_par
*par
= (struct riva_par
*)info
->par
;
unsigned int base
;
if (var
->xoffset
> (var
->xres_virtual
- var
->xres
))
return -EINVAL
;
if (var
->yoffset
> (var
->yres_virtual
- var
->yres
))
return -EINVAL
;
if (var
->vmode
& FB_VMODE_YWRAP
) {
if (var
->yoffset
< 0
|| var
->yoffset
>= info
->var.
yres_virtual
|| var
->xoffset
) return -EINVAL
;
} else {
if (var
->xoffset
+ info
->var.
xres > info
->var.
xres_virtual ||
var
->yoffset
+ info
->var.
yres > info
->var.
yres_virtual)
return -EINVAL
;
}
base
= var
->yoffset
* info
->fix.
line_length + var
->xoffset
;
par
->riva.
SetStartAddress(&par
->riva
, base
);
info
->var.
xoffset = var
->xoffset
;
info
->var.
yoffset = var
->yoffset
;
if (var
->vmode
& FB_VMODE_YWRAP
)
info
->var.
vmode |= FB_VMODE_YWRAP
;
else
info
->var.
vmode &= ~FB_VMODE_YWRAP
;
return 0;
}
static int rivafb_blank
(int blank
, struct fb_info
*info
)
{
struct riva_par
*par
= (struct riva_par
*)info
->par
;
unsigned char tmp
, vesa
;
tmp
= SEQin
(par
, 0x01) & ~
0x20; /* screen on/off */
vesa
= CRTCin
(par
, 0x1a) & ~
0xc0; /* sync on/off */
if (blank
) {
tmp
|= 0x20;
switch (blank
- 1) {
case VESA_NO_BLANKING
:
break;
case VESA_VSYNC_SUSPEND
:
vesa
|= 0x80;
break;
case VESA_HSYNC_SUSPEND
:
vesa
|= 0x40;
break;
case VESA_POWERDOWN
:
vesa
|= 0xc0;
break;
}
}
SEQout
(par
, 0x01, tmp
);
CRTCout
(par
, 0x1a, vesa
);
return 0;
}
/**
* rivafb_setcolreg
* @regno: register index
* @red: red component
* @green: green component
* @blue: blue component
* @transp: transparency
* @info: pointer to fb_info object containing info for current riva board
*
* DESCRIPTION:
* Set a single color register. The values supplied have a 16 bit
* magnitude.
*
* RETURNS:
* Return != 0 for invalid regno.
*
* CALLED FROM:
* fbcmap.c:fb_set_cmap()
*/
static int rivafb_setcolreg
(unsigned regno
, unsigned red
, unsigned green
,
unsigned blue
, unsigned transp
,
struct fb_info
*info
)
{
struct riva_par
*par
= (struct riva_par
*)info
->par
;
RIVA_HW_INST
*chip
= &par
->riva
;
int i
;
if (regno
>= riva_get_cmap_len
(&info
->var
))
return -EINVAL
;
if (info
->var.
grayscale) {
/* gray = 0.30*R + 0.59*G + 0.11*B */
red
= green
= blue
=
(red
* 77 + green
* 151 + blue
* 28) >> 8;
}
switch (info
->var.
bits_per_pixel) {
case 8:
/* "transparent" stuff is completely ignored. */
riva_wclut
(chip
, regno
, red
>> 8, green
>> 8, blue
>> 8);
break;
case 16:
if (info
->var.
green.
length == 5) {
if (regno
< 16) {
/* 0rrrrrgg gggbbbbb */
((u32
*)info
->pseudo_palette
)[regno
] =
((red
& 0xf800) >> 1) |
((green
& 0xf800) >> 6) |
((blue
& 0xf800) >> 11);
}
for (i
= 0; i
< 8; i
++)
riva_wclut
(chip
, regno
*8+i
, red
>> 8,
green
>> 8, blue
>> 8);
} else {
u8 r
, g
, b
;
if (regno
< 16) {
/* rrrrrggg gggbbbbb */
((u32
*)info
->pseudo_palette
)[regno
] =
((red
& 0xf800) >> 0) |
((green
& 0xf800) >> 5) |
((blue
& 0xf800) >> 11);
}
if (regno
< 32) {
for (i
= 0; i
< 8; i
++) {
riva_wclut
(chip
, regno
*8+i
, red
>> 8,
green
>> 8, blue
>> 8);
}
}
for (i
= 0; i
< 4; i
++) {
riva_rclut
(chip
, regno
*2+i
, &r
, &g
, &b
);
riva_wclut
(chip
, regno
*4+i
, r
, green
>> 8, b
);
}
}
break;
case 32:
if (regno
< 16) {
((u32
*)info
->pseudo_palette
)[regno
] =
((red
& 0xff00) << 8) |
((green
& 0xff00)) | ((blue
& 0xff00) >> 8);
}
riva_wclut
(chip
, regno
, red
>> 8, green
>> 8, blue
>> 8);
break;
default:
/* do nothing */
break;
}
return 0;
}
/**
* rivafb_fillrect - hardware accelerated color fill function
* @info: pointer to fb_info structure
* @rect: pointer to fb_fillrect structure
*
* DESCRIPTION:
* This function fills up a region of framebuffer memory with a solid
* color with a choice of two different ROP's, copy or invert.
*
* CALLED FROM:
* framebuffer hook
*/
static void rivafb_fillrect
(struct fb_info
*info
, const struct fb_fillrect
*rect
)
{
struct riva_par
*par
= (struct riva_par
*) info
->par
;
u_int color
, rop
= 0;
if (info
->var.
bits_per_pixel == 8)
color
= rect
->color
;
else
color
= ((u32
*)info
->pseudo_palette
)[rect
->color
];
switch (rect
->rop
) {
case ROP_XOR
:
rop
= 0x66;
break;
case ROP_COPY
:
default:
rop
= 0xCC;
break;
}
RIVA_FIFO_FREE
(par
->riva
, Rop
, 1);
par
->riva.
Rop->Rop3
= rop
;
RIVA_FIFO_FREE
(par
->riva
, Bitmap
, 1);
par
->riva.
Bitmap->Color1A
= color
;
RIVA_FIFO_FREE
(par
->riva
, Bitmap
, 2);
par
->riva.
Bitmap->UnclippedRectangle
[0].
TopLeft =
(rect
->dx
<< 16) | rect
->dy
;
par
->riva.
Bitmap->UnclippedRectangle
[0].
WidthHeight =
(rect
->width
<< 16) | rect
->height
;
RIVA_FIFO_FREE
(par
->riva
, Rop
, 1);
par
->riva.
Rop->Rop3
= 0xCC; // back to COPY
}
/**
* rivafb_copyarea - hardware accelerated blit function
* @info: pointer to fb_info structure
* @region: pointer to fb_copyarea structure
*
* DESCRIPTION:
* This copies an area of pixels from one location to another
*
* CALLED FROM:
* framebuffer hook
*/
static void rivafb_copyarea
(struct fb_info
*info
, const struct fb_copyarea
*region
)
{
struct riva_par
*par
= (struct riva_par
*) info
->par
;
RIVA_FIFO_FREE
(par
->riva
, Blt
, 3);
par
->riva.
Blt->TopLeftSrc
= (region
->sy
<< 16) | region
->sx
;
par
->riva.
Blt->TopLeftDst
= (region
->dy
<< 16) | region
->dx
;
par
->riva.
Blt->WidthHeight
= (region
->height
<< 16) | region
->width
;
wait_for_idle
(par
);
}
static inline void convert_bgcolor_16
(u32
*col
)
{
*col
= ((*col
& 0x00007C00) << 9)
| ((*col
& 0x000003E0) << 6)
| ((*col
& 0x0000001F) << 3)
| 0xFF000000;
}
/**
* rivafb_imageblit: hardware accelerated color expand function
* @info: pointer to fb_info structure
* @image: pointer to fb_image structure
*
* DESCRIPTION:
* If the source is a monochrome bitmap, the function fills up a a region
* of framebuffer memory with pixels whose color is determined by the bit
* setting of the bitmap, 1 - foreground, 0 - background.
*
* If the source is not a monochrome bitmap, color expansion is not done.
* In this case, it is channeled to a software function.
*
* CALLED FROM:
* framebuffer hook
*/
static void rivafb_imageblit
(struct fb_info
*info
,
const struct fb_image
*image
)
{
struct riva_par
*par
= (struct riva_par
*) info
->par
;
u32 fgx
= 0, bgx
= 0, width
, tmp
;
u8
*cdat
= (u8
*) image
->data
;
volatile u32
*d
;
int i
, size
;
if (image
->depth
!= 1) {
cfb_imageblit
(info
, image
);
return;
}
switch (info
->var.
bits_per_pixel) {
case 8:
fgx
= image
->fg_color
;
bgx
= image
->bg_color
;
break;
case 16:
fgx
= ((u32
*)info
->pseudo_palette
)[image
->fg_color
];
bgx
= ((u32
*)info
->pseudo_palette
)[image
->bg_color
];
if (info
->var.
green.
length == 6)
convert_bgcolor_16
(&bgx
);
break;
case 32:
fgx
= ((u32
*)info
->pseudo_palette
)[image
->fg_color
];
bgx
= ((u32
*)info
->pseudo_palette
)[image
->bg_color
];
break;
}
RIVA_FIFO_FREE
(par
->riva
, Bitmap
, 7);
par
->riva.
Bitmap->ClipE.
TopLeft =
(image
->dy
<< 16) | (image
->dx
& 0xFFFF);
par
->riva.
Bitmap->ClipE.
BottomRight =
(((image
->dy
+ image
->height
) << 16) |
((image
->dx
+ image
->width
) & 0xffff));
par
->riva.
Bitmap->Color0E
= bgx
;
par
->riva.
Bitmap->Color1E
= fgx
;
par
->riva.
Bitmap->WidthHeightInE
=
(image
->height
<< 16) | ((image
->width
+ 31) & ~
31);
par
->riva.
Bitmap->WidthHeightOutE
=
(image
->height
<< 16) | ((image
->width
+ 31) & ~
31);
par
->riva.
Bitmap->PointE
=
(image
->dy
<< 16) | (image
->dx
& 0xFFFF);
d
= &par
->riva.
Bitmap->MonochromeData01E
;
width
= (image
->width
+ 31)/32;
size
= width
* image
->height
;
while (size
>= 16) {
RIVA_FIFO_FREE
(par
->riva
, Bitmap
, 16);
for (i
= 0; i
< 16; i
++) {
tmp
= *((u32
*)cdat
)++;
reverse_order
(&tmp
);
d
[i
] = tmp
;
}
size
-= 16;
}
if (size
) {
RIVA_FIFO_FREE
(par
->riva
, Bitmap
, size
);
for (i
= 0; i
< size
; i
++) {
tmp
= *((u32
*) cdat
)++;
reverse_order
(&tmp
);
d
[i
] = tmp
;
}
}
}
/**
* rivafb_cursor - hardware cursor function
* @info: pointer to info structure
* @cursor: pointer to fbcursor structure
*
* DESCRIPTION:
* A cursor function that supports displaying a cursor image via hardware.
* Within the kernel, copy and invert rops are supported. If exported
* to user space, only the copy rop will be supported.
*
* CALLED FROM
* framebuffer hook
*/
static int rivafb_cursor
(struct fb_info
*info
, struct fb_cursor
*cursor
)
{
struct riva_par
*par
= (struct riva_par
*) info
->par
;
u8 data
[MAX_CURS
* MAX_CURS
/8];
u8 mask
[MAX_CURS
* MAX_CURS
/8];
u16 fg
, bg
;
int i
;
par
->riva.
ShowHideCursor(&par
->riva
, 0);
if (cursor
->set
& FB_CUR_SETPOS
) {
u32 xx
, yy
, temp
;
info
->cursor.
image.
dx = cursor
->image.
dx;
info
->cursor.
image.
dy = cursor
->image.
dy;
yy
= cursor
->image.
dy - info
->var.
yoffset;
xx
= cursor
->image.
dx - info
->var.
xoffset;
temp
= xx
& 0xFFFF;
temp
|= yy
<< 16;
par
->riva.
PRAMDAC[0x0000300/4] = temp
;
}
if (cursor
->set
& FB_CUR_SETSIZE
) {
info
->cursor.
image.
height = cursor
->image.
height;
info
->cursor.
image.
width = cursor
->image.
width;
memset_io
(par
->riva.
CURSOR, 0, MAX_CURS
* MAX_CURS
* 2);
}
if (cursor
->set
& FB_CUR_SETCMAP
) {
info
->cursor.
image.
bg_color = cursor
->image.
bg_color;
info
->cursor.
image.
fg_color = cursor
->image.
fg_color;
}
if (cursor
->set
& (FB_CUR_SETSHAPE
| FB_CUR_SETCMAP
)) {
u32 bg_idx
= info
->cursor.
image.
bg_color;
u32 fg_idx
= info
->cursor.
image.
fg_color;
u32 s_pitch
= (info
->cursor.
image.
width+7) >> 3;
u32 d_pitch
= MAX_CURS
/8;
u8
*dat
= (u8
*) cursor
->image.
data;
u8
*msk
= (u8
*) info
->cursor.
mask;
u8 src
[64];
switch (info
->cursor.
rop) {
case ROP_XOR
:
for (i
= 0; i
< s_pitch
* info
->cursor.
image.
height; i
++)
src
[i
] = dat
[i
] ^ msk
[i
];
break;
case ROP_COPY
:
default:
for (i
= 0; i
< s_pitch
* info
->cursor.
image.
height; i
++)
src
[i
] = dat
[i
] & msk
[i
];
break;
}
move_buf_aligned
(info
, data
, src
, d_pitch
, s_pitch
, info
->cursor.
image.
height);
move_buf_aligned
(info
, mask
, msk
, d_pitch
, s_pitch
, info
->cursor.
image.
height);
bg
= ((info
->cmap.
red[bg_idx
] & 0xf8) << 7) |
((info
->cmap.
green[bg_idx
] & 0xf8) << 2) |
((info
->cmap.
blue[bg_idx
] & 0xf8) >> 3);
fg
= ((info
->cmap.
red[fg_idx
] & 0xf8) << 7) |
((info
->cmap.
green[fg_idx
] & 0xf8) << 2) |
((info
->cmap.
blue[fg_idx
] & 0xf8) >> 3);
par
->riva.
LockUnlock(&par
->riva
, 0);
rivafb_load_cursor_image
(par
, data
, mask
, bg
, fg
,
info
->cursor.
image.
width,
info
->cursor.
image.
height);
}
if (info
->cursor.
enable)
par
->riva.
ShowHideCursor(&par
->riva
, 1);
return 0;
}
static int rivafb_sync
(struct fb_info
*info
)
{
struct riva_par
*par
= (struct riva_par
*)info
->par
;
wait_for_idle
(par
);
return 0;
}
/* ------------------------------------------------------------------------- *
*
* initialization helper functions
*
* ------------------------------------------------------------------------- */
/* kernel interface */
static struct fb_ops riva_fb_ops
= {
.
owner = THIS_MODULE
,
.
fb_open = rivafb_open
,
.
fb_release = rivafb_release
,
.
fb_check_var = rivafb_check_var
,
.
fb_set_par = rivafb_set_par
,
.
fb_setcolreg = rivafb_setcolreg
,
.
fb_pan_display = rivafb_pan_display
,
.
fb_blank = rivafb_blank
,
.
fb_fillrect = rivafb_fillrect
,
.
fb_copyarea = rivafb_copyarea
,
.
fb_imageblit = rivafb_imageblit
,
.
fb_cursor = rivafb_cursor
,
.
fb_sync = rivafb_sync
,
};
static int __devinit riva_set_fbinfo
(struct fb_info
*info
)
{
struct riva_par
*par
= (struct riva_par
*) info
->par
;
unsigned int cmap_len
;
info
->flags
= FBINFO_FLAG_DEFAULT
;
info
->var
= rivafb_default_var
;
info
->fix
= rivafb_fix
;
info
->fbops
= &riva_fb_ops
;
info
->pseudo_palette
= pseudo_palette
;
#ifndef MODULE
if (mode_option
)
fb_find_mode
(&info
->var
, info
, mode_option
,
NULL
, 0, NULL
, 8);
#endif
if (par
->use_default_var
)
/* We will use the modified default var */
info
->var
= rivafb_default_var
;
cmap_len
= riva_get_cmap_len
(&info
->var
);
fb_alloc_cmap
(&info
->cmap
, cmap_len
, 0);
info
->pixmap.
size = 64 * 1024;
info
->pixmap.
buf_align = 4;
info
->pixmap.
scan_align = 4;
info
->pixmap.
flags = FB_PIXMAP_SYSTEM
;
return 0;
}
#ifdef CONFIG_PPC_OF
static int riva_get_EDID_OF
(struct riva_par
*par
, struct pci_dev
*pd
)
{
struct device_node
*dp
;
unsigned char *pedid
= NULL
;
dp
= pci_device_to_OF_node
(pd
);
pedid
= (unsigned char *)get_property
(dp
, "EDID,B", 0);
if (pedid
) {
par
->EDID
= pedid
;
return 1;
} else
return 0;
}
#endif /* CONFIG_PPC_OF */
static int riva_dfp_parse_EDID
(struct riva_par
*par
)
{
unsigned char *block
= par
->EDID
;
if (!block
)
return 0;
/* jump to detailed timing block section */
block
+= 54;
par
->clock = (block
[0] + (block
[1] << 8));
par
->panel_xres
= (block
[2] + ((block
[4] & 0xf0) << 4));
par
->hblank
= (block
[3] + ((block
[4] & 0x0f) << 8));
par
->panel_yres
= (block
[5] + ((block
[7] & 0xf0) << 4));
par
->vblank
= (block
[6] + ((block
[7] & 0x0f) << 8));
par
->hOver_plus
= (block
[8] + ((block
[11] & 0xc0) << 2));
par
->hSync_width
= (block
[9] + ((block
[11] & 0x30) << 4));
par
->vOver_plus
= ((block
[10] >> 4) + ((block
[11] & 0x0c) << 2));
par
->vSync_width
= ((block
[10] & 0x0f) + ((block
[11] & 0x03) << 4));
par
->interlaced
= ((block
[17] & 0x80) >> 7);
par
->synct
= ((block
[17] & 0x18) >> 3);
par
->misc
= ((block
[17] & 0x06) >> 1);
par
->hAct_high
= par
->vAct_high
= 0;
if (par
->synct
== 3) {
if (par
->misc
& 2)
par
->hAct_high
= 1;
if (par
->misc
& 1)
par
->vAct_high
= 1;
}
printk
(KERN_INFO PFX
"detected DFP panel size from EDID: %dx%d\n",
par
->panel_xres
, par
->panel_yres
);
par
->got_dfpinfo
= 1;
return 1;
}
static void riva_update_default_var
(struct fb_info
*info
)
{
struct fb_var_screeninfo
*var
= &rivafb_default_var
;
struct riva_par
*par
= (struct riva_par
*) info
->par
;
var
->xres
= par
->panel_xres
;
var
->yres
= par
->panel_yres
;
var
->xres_virtual
= par
->panel_xres
;
var
->yres_virtual
= par
->panel_yres
;
var
->xoffset
= var
->yoffset
= 0;
var
->bits_per_pixel
= 8;
var
->pixclock
= 100000000 / par
->clock;
var
->left_margin
= (par
->hblank
- par
->hOver_plus
- par
->hSync_width
);
var
->right_margin
= par
->hOver_plus
;
var
->upper_margin
= (par
->vblank
- par
->vOver_plus
- par
->vSync_width
);
var
->lower_margin
= par
->vOver_plus
;
var
->hsync_len
= par
->hSync_width
;
var
->vsync_len
= par
->vSync_width
;
var
->sync
= 0;
if (par
->synct
== 3) {
if (par
->hAct_high
)
var
->sync
|= FB_SYNC_HOR_HIGH_ACT
;
if (par
->vAct_high
)
var
->sync
|= FB_SYNC_VERT_HIGH_ACT
;
}
var
->vmode
= 0;
if (par
->interlaced
)
var
->vmode
|= FB_VMODE_INTERLACED
;
var
->accel_flags
|= FB_ACCELF_TEXT
;
par
->use_default_var
= 1;
}
static void riva_get_EDID
(struct fb_info
*info
, struct pci_dev
*pdev
)
{
#ifdef CONFIG_PPC_OF
if (!riva_get_EDID_OF
(info
, pdev
))
printk
("rivafb: could not retrieve EDID from OF\n");
#else
/* XXX use other methods later */
#endif
}
static void riva_get_dfpinfo
(struct fb_info
*info
)
{
struct riva_par
*par
= (struct riva_par
*) info
->par
;
if (riva_dfp_parse_EDID
(par
))
riva_update_default_var
(info
);
/* if user specified flatpanel, we respect that */
if (par
->got_dfpinfo
== 1)
par
->FlatPanel
= 1;
}
/* ------------------------------------------------------------------------- *
*
* PCI bus
*
* ------------------------------------------------------------------------- */
static int __devinit rivafb_probe
(struct pci_dev
*pd
,
const struct pci_device_id
*ent
)
{
struct riva_chip_info
*rci
= &riva_chip_info
[ent
->driver_data
];
struct riva_par
*default_par
;
struct fb_info
*info
;
assert(pd
!= NULL
);
assert(rci
!= NULL
);
info
= kmalloc
(sizeof(struct fb_info
), GFP_KERNEL
);
if (!info
)
goto err_out
;
default_par
= kmalloc
(sizeof(struct riva_par
), GFP_KERNEL
);
if (!default_par
)
goto err_out_kfree
;
memset(info
, 0, sizeof(struct fb_info
));
memset(default_par
, 0, sizeof(struct riva_par
));
info
->pixmap.
addr = kmalloc
(64 * 1024, GFP_KERNEL
);
if (info
->pixmap.
addr == NULL
)
goto err_out_kfree1
;
memset(info
->pixmap.
addr, 0, 64 * 1024);
strcat(rivafb_fix.
id, rci
->name
);
default_par
->riva.
Architecture = rci
->arch_rev
;
default_par
->Chipset
= (pd
->vendor
<< 16) | pd
->device
;
printk
(KERN_INFO PFX
"nVidia device/chipset %X\n",default_par
->Chipset
);
default_par
->FlatPanel
= flatpanel
;
if (flatpanel
== 1)
printk
(KERN_INFO PFX
"flatpanel support enabled\n");
default_par
->forceCRTC
= forceCRTC
;
rivafb_fix.
mmio_len = pci_resource_len
(pd
, 0);
rivafb_fix.
smem_len = pci_resource_len
(pd
, 1);
{
/* enable IO and mem if not already done */
unsigned short cmd
;
pci_read_config_word
(pd
, PCI_COMMAND
, &cmd
);
cmd
|= (PCI_COMMAND_IO
| PCI_COMMAND_MEMORY
);
pci_write_config_word
(pd
, PCI_COMMAND
, cmd
);
}
rivafb_fix.
mmio_start = pci_resource_start
(pd
, 0);
rivafb_fix.
smem_start = pci_resource_start
(pd
, 1);
if (!request_mem_region
(rivafb_fix.
mmio_start,
rivafb_fix.
mmio_len, "rivafb")) {
printk
(KERN_ERR PFX
"cannot reserve MMIO region\n");
goto err_out_kfree2
;
}
default_par
->ctrl_base
= ioremap
(rivafb_fix.
mmio_start,
rivafb_fix.
mmio_len);
if (!default_par
->ctrl_base
) {
printk
(KERN_ERR PFX
"cannot ioremap MMIO base\n");
goto err_out_free_base0
;
}
info
->par
= default_par
;
riva_get_EDID
(info
, pd
);
riva_get_dfpinfo
(info
);
switch (default_par
->riva.
Architecture) {
case NV_ARCH_03
:
/* Riva128's PRAMIN is in the "framebuffer" space
* Since these cards were never made with more than 8 megabytes
* we can safely allocate this separately.
*/
if (!request_mem_region
(rivafb_fix.
smem_start + 0x00C00000,
0x00008000, "rivafb")) {
printk
(KERN_ERR PFX
"cannot reserve PRAMIN region\n");
goto err_out_iounmap_ctrl
;
}
default_par
->riva.
PRAMIN = ioremap
(rivafb_fix.
smem_start + 0x00C00000, 0x00008000);
if (!default_par
->riva.
PRAMIN) {
printk
(KERN_ERR PFX
"cannot ioremap PRAMIN region\n");
goto err_out_free_nv3_pramin
;
}
rivafb_fix.
accel = FB_ACCEL_NV3
;
break;
case NV_ARCH_04
:
case NV_ARCH_10
:
case NV_ARCH_20
:
default_par
->riva.
PCRTC0 = (unsigned *)(default_par
->ctrl_base
+ 0x00600000);
default_par
->riva.
PRAMIN = (unsigned *)(default_par
->ctrl_base
+ 0x00710000);
rivafb_fix.
accel = FB_ACCEL_NV4
;
break;
}
riva_common_setup
(default_par
);
if (default_par
->riva.
Architecture == NV_ARCH_03
) {
default_par
->riva.
PCRTC = default_par
->riva.
PCRTC0 = default_par
->riva.
PGRAPH;
}
rivafb_fix.
smem_len = riva_get_memlen
(default_par
) * 1024;
default_par
->dclk_max
= riva_get_maxdclk
(default_par
) * 1000;
if (!request_mem_region
(rivafb_fix.
smem_start,
rivafb_fix.
smem_len, "rivafb")) {
printk
(KERN_ERR PFX
"cannot reserve FB region\n");
goto err_out_iounmap_nv3_pramin
;
}
info
->screen_base
= ioremap
(rivafb_fix.
smem_start,
rivafb_fix.
smem_len);
if (!info
->screen_base
) {
printk
(KERN_ERR PFX
"cannot ioremap FB base\n");
goto err_out_free_base1
;
}
#ifdef CONFIG_MTRR
if (!nomtrr
) {
default_par
->mtrr.
vram = mtrr_add
(rivafb_fix.
smem_start,
rivafb_fix.
smem_len,
MTRR_TYPE_WRCOMB
, 1);
if (default_par
->mtrr.
vram < 0) {
printk
(KERN_ERR PFX
"unable to setup MTRR\n");
} else {
default_par
->mtrr.
vram_valid = 1;
/* let there be speed */
printk
(KERN_INFO PFX
"RIVA MTRR set to ON\n");
}
}
#endif /* CONFIG_MTRR */
if (riva_set_fbinfo
(info
) < 0) {
printk
(KERN_ERR PFX
"error setting initial video mode\n");
goto err_out_iounmap_fb
;
}
if (register_framebuffer
(info
) < 0) {
printk
(KERN_ERR PFX
"error registering riva framebuffer\n");
goto err_out_iounmap_fb
;
}
pci_set_drvdata
(pd
, info
);
printk
(KERN_INFO PFX
"PCI nVidia NV%x framebuffer ver %s (%s, %dMB @ 0x%x)\n",
default_par
->riva.
Architecture,
RIVAFB_VERSION
,
info
->fix.
id,
info
->fix.
smem_len / (1024 * 1024),
info
->fix.
smem_start);
return 0;
err_out_iounmap_fb
:
iounmap
(info
->screen_base
);
err_out_free_base1
:
release_mem_region
(rivafb_fix.
smem_start, rivafb_fix.
smem_len);
err_out_iounmap_nv3_pramin
:
if (default_par
->riva.
Architecture == NV_ARCH_03
)
iounmap
((caddr_t
)default_par
->riva.
PRAMIN);
err_out_free_nv3_pramin
:
if (default_par
->riva.
Architecture == NV_ARCH_03
)
release_mem_region
(rivafb_fix.
smem_start + 0x00C00000, 0x00008000);
err_out_iounmap_ctrl
:
iounmap
(default_par
->ctrl_base
);
err_out_free_base0
:
release_mem_region
(rivafb_fix.
mmio_start, rivafb_fix.
mmio_len);
err_out_kfree2
:
kfree
(info
->pixmap.
addr);
err_out_kfree1
:
kfree
(default_par
);
err_out_kfree
:
kfree
(info
);
err_out
:
return -ENODEV
;
}
static void __exit rivafb_remove
(struct pci_dev
*pd
)
{
struct fb_info
*info
= pci_get_drvdata
(pd
);
struct riva_par
*par
= (struct riva_par
*) info
->par
;
if (!info
)
return;
unregister_framebuffer
(info
);
#ifdef CONFIG_MTRR
if (par
->mtrr.
vram_valid)
mtrr_del
(par
->mtrr.
vram, info
->fix.
smem_start,
info
->fix.
smem_len);
#endif /* CONFIG_MTRR */
iounmap
(par
->ctrl_base
);
iounmap
(info
->screen_base
);
release_mem_region
(info
->fix.
mmio_start,
info
->fix.
mmio_len);
release_mem_region
(info
->fix.
smem_start,
info
->fix.
smem_len);
if (par
->riva.
Architecture == NV_ARCH_03
) {
iounmap
((caddr_t
)par
->riva.
PRAMIN);
release_mem_region
(info
->fix.
smem_start + 0x00C00000, 0x00008000);
}
kfree
(info
->pixmap.
addr);
kfree
(par
);
kfree
(info
);
pci_set_drvdata
(pd
, NULL
);
}
/* ------------------------------------------------------------------------- *
*
* initialization
*
* ------------------------------------------------------------------------- */
#ifndef MODULE
int __init rivafb_setup
(char *options
)
{
char *this_opt
;
if (!options
|| !*options
)
return 0;
while ((this_opt
= strsep
(&options
, ",")) != NULL
) {
if (!strncmp(this_opt
, "forceCRTC", 9)) {
char *p
;
p
= this_opt
+ 9;
if (!*p
|| !*(++p
)) continue;
forceCRTC
= *p
- '0';
if (forceCRTC
< 0 || forceCRTC
> 1)
forceCRTC
= -1;
} else if (!strncmp(this_opt
, "flatpanel", 9)) {
flatpanel
= 1;
#ifdef CONFIG_MTRR
} else if (!strncmp(this_opt
, "nomtrr", 6)) {
nomtrr
= 1;
#endif
} else
mode_option
= this_opt
;
}
return 0;
}
#endif /* !MODULE */
static struct pci_driver rivafb_driver
= {
.
name = "rivafb",
.
id_table = rivafb_pci_tbl
,
.
probe = rivafb_probe
,
.
remove = __exit_p
(rivafb_remove
),
};
/* ------------------------------------------------------------------------- *
*
* modularization
*
* ------------------------------------------------------------------------- */
int __init rivafb_init
(void)
{
if (pci_register_driver
(&rivafb_driver
) > 0)
return 0;
pci_unregister_driver
(&rivafb_driver
);
return -ENODEV
;
}
#ifdef MODULE
static void __exit rivafb_exit
(void)
{
pci_unregister_driver
(&rivafb_driver
);
}
module_init
(rivafb_init
);
module_exit
(rivafb_exit
);
MODULE_PARM
(flatpanel
, "i");
MODULE_PARM_DESC
(flatpanel
, "Enables experimental flat panel support for some chipsets. (0 or 1=enabled) (default=0)");
MODULE_PARM
(forceCRTC
, "i");
MODULE_PARM_DESC
(forceCRTC
, "Forces usage of a particular CRTC in case autodetection fails. (0 or 1) (default=autodetect)");
#ifdef CONFIG_MTRR
MODULE_PARM
(nomtrr
, "i");
MODULE_PARM_DESC
(nomtrr
, "Disables MTRR support (0 or 1=disabled) (default=0)");
#endif
#endif /* MODULE */
MODULE_AUTHOR
("Ani Joshi, maintainer");
MODULE_DESCRIPTION
("Framebuffer driver for nVidia Riva 128, TNT, TNT2, and the GeForce series");
MODULE_LICENSE
("GPL");