/***************************************************************************\ |* *| |* Copyright 2003 NVIDIA, Corporation. All rights reserved. *| |* *| |* NOTICE TO USER: The source code is copyrighted under U.S. and *| |* international laws. Users and possessors of this source code are *| |* hereby granted a nonexclusive, royalty-free copyright license to *| |* use this code in individual and commercial software. *| |* *| |* Any use of this source code must include, in the user documenta- *| |* tion and internal comments to the code, notices to the end user *| |* as follows: *| |* *| |* Copyright 2003 NVIDIA, Corporation. All rights reserved. *| |* *| |* NVIDIA, CORPORATION MAKES NO REPRESENTATION ABOUT THE SUITABILITY *| |* OF THIS SOURCE CODE FOR ANY PURPOSE. IT IS PROVIDED "AS IS" *| |* WITHOUT EXPRESS OR IMPLIED WARRANTY OF ANY KIND. NVIDIA, CORPOR- *| |* ATION DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOURCE CODE, *| |* INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY, NONINFRINGE- *| |* MENT, AND FITNESS FOR A PARTICULAR PURPOSE. IN NO EVENT SHALL *| |* NVIDIA, CORPORATION BE LIABLE FOR ANY SPECIAL, INDIRECT, INCI- *| |* DENTAL, OR CONSEQUENTIAL DAMAGES, OR ANY DAMAGES WHATSOEVER RE- *| |* SULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION *| |* OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF *| |* OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOURCE CODE. *| |* *| |* U.S. Government End Users. This source code is a "commercial *| |* item," as that term is defined at 48 C.F.R. 2.101 (OCT 1995), *| |* consisting of "commercial computer software" and "commercial *| |* computer software documentation," as such terms are used in *| |* 48 C.F.R. 12.212 (SEPT 1995) and is provided to the U.S. Govern- *| |* ment only as a commercial end item. Consistent with 48 C.F.R. *| |* 12.212 and 48 C.F.R. 227.7202-1 through 227.7202-4 (JUNE 1995), *| |* all U.S. Government End Users acquire the source code with only *| |* those rights set forth herein. *| |* *| \***************************************************************************/ /* * GPL Licensing Note - According to Mark Vojkovich, author of the Xorg/ * XFree86 'nv' driver, this source code is provided under MIT-style licensing * where the source code is provided "as is" without warranty of any kind. * The only usage restriction is for the copyright notices to be retained * whenever code is used. * * Antonino Daplas <adaplas@pol.net> 2005-03-11 */ #include <video/vga.h> #include <linux/delay.h> #include <linux/pci.h> #include <linux/slab.h> #include "nv_type.h" #include "nv_local.h" #include "nv_proto.h" /* * Override VGA I/O routines. */ void NVWriteCrtc(struct nvidia_par *par, u8 index, u8 value) { VGA_WR08(par->PCIO, par->IOBase + 0x04, index); VGA_WR08(par->PCIO, par->IOBase + 0x05, value); } u8 NVReadCrtc(struct nvidia_par *par, u8 index) { VGA_WR08(par->PCIO, par->IOBase + 0x04, index); return (VGA_RD08(par->PCIO, par->IOBase + 0x05)); } void NVWriteGr(struct nvidia_par *par, u8 index, u8 value) { VGA_WR08(par->PVIO, VGA_GFX_I, index); VGA_WR08(par->PVIO, VGA_GFX_D, value); } u8 NVReadGr(struct nvidia_par *par, u8 index) { VGA_WR08(par->PVIO, VGA_GFX_I, index); return (VGA_RD08(par->PVIO, VGA_GFX_D)); } void NVWriteSeq(struct nvidia_par *par, u8 index, u8 value) { VGA_WR08(par->PVIO, VGA_SEQ_I, index); VGA_WR08(par->PVIO, VGA_SEQ_D, value); } u8 NVReadSeq(struct nvidia_par *par, u8 index) { VGA_WR08(par->PVIO, VGA_SEQ_I, index); return (VGA_RD08(par->PVIO, VGA_SEQ_D)); } void NVWriteAttr(struct nvidia_par *par, u8 index, u8 value) { volatile u8 tmp; tmp = VGA_RD08(par->PCIO, par->IOBase + 0x0a); if (par->paletteEnabled) index &= ~0x20; else index |= 0x20; VGA_WR08(par->PCIO, VGA_ATT_IW, index); VGA_WR08(par->PCIO, VGA_ATT_W, value); } u8 NVReadAttr(struct nvidia_par *par, u8 index) { volatile u8 tmp; tmp = VGA_RD08(par->PCIO, par->IOBase + 0x0a); if (par->paletteEnabled) index &= ~0x20; else index |= 0x20; VGA_WR08(par->PCIO, VGA_ATT_IW, index); return (VGA_RD08(par->PCIO, VGA_ATT_R)); } void NVWriteMiscOut(struct nvidia_par *par, u8 value) { VGA_WR08(par->PVIO, VGA_MIS_W, value); } u8 NVReadMiscOut(struct nvidia_par *par) { return (VGA_RD08(par->PVIO, VGA_MIS_R)); } #if 0 void NVEnablePalette(struct nvidia_par *par) { volatile u8 tmp; tmp = VGA_RD08(par->PCIO, par->IOBase + 0x0a); VGA_WR08(par->PCIO, VGA_ATT_IW, 0x00); par->paletteEnabled = 1; } void NVDisablePalette(struct nvidia_par *par) { volatile u8 tmp; tmp = VGA_RD08(par->PCIO, par->IOBase + 0x0a); VGA_WR08(par->PCIO, VGA_ATT_IW, 0x20); par->paletteEnabled = 0; } #endif /* 0 */ void NVWriteDacMask(struct nvidia_par *par, u8 value) { VGA_WR08(par->PDIO, VGA_PEL_MSK, value); } #if 0 u8 NVReadDacMask(struct nvidia_par *par) { return (VGA_RD08(par->PDIO, VGA_PEL_MSK)); } #endif /* 0 */ void NVWriteDacReadAddr(struct nvidia_par *par, u8 value) { VGA_WR08(par->PDIO, VGA_PEL_IR, value); } void NVWriteDacWriteAddr(struct nvidia_par *par, u8 value) { VGA_WR08(par->PDIO, VGA_PEL_IW, value); } void NVWriteDacData(struct nvidia_par *par, u8 value) { VGA_WR08(par->PDIO, VGA_PEL_D, value); } u8 NVReadDacData(struct nvidia_par *par) { return (VGA_RD08(par->PDIO, VGA_PEL_D)); } static int NVIsConnected(struct nvidia_par *par, int output) { volatile u32 __iomem *PRAMDAC = par->PRAMDAC0; u32 reg52C, reg608, dac0_reg608 = 0; int present; if (output) { dac0_reg608 = NV_RD32(PRAMDAC, 0x0608); PRAMDAC += 0x800; } reg52C = NV_RD32(PRAMDAC, 0x052C); reg608 = NV_RD32(PRAMDAC, 0x0608); NV_WR32(PRAMDAC, 0x0608, reg608 & ~0x00010000); NV_WR32(PRAMDAC, 0x052C, reg52C & 0x0000FEEE); msleep(1); NV_WR32(PRAMDAC, 0x052C, NV_RD32(PRAMDAC, 0x052C) | 1); NV_WR32(par->PRAMDAC0, 0x0610, 0x94050140); NV_WR32(par->PRAMDAC0, 0x0608, NV_RD32(par->PRAMDAC0, 0x0608) | 0x00001000); msleep(1); present = (NV_RD32(PRAMDAC, 0x0608) & (1 << 28)) ? 1 : 0; if (present) printk("nvidiafb: CRTC%i analog found\n", output); else printk("nvidiafb: CRTC%i analog not found\n", output); if (output) NV_WR32(par->PRAMDAC0, 0x0608, dac0_reg608); NV_WR32(PRAMDAC, 0x052C, reg52C); NV_WR32(PRAMDAC, 0x0608, reg608); return present; } static void NVSelectHeadRegisters(struct nvidia_par *par, int head) { if (head) { par->PCIO = par->PCIO0 + 0x2000; par->PCRTC = par->PCRTC0 + 0x800; par->PRAMDAC = par->PRAMDAC0 + 0x800; par->PDIO = par->PDIO0 + 0x2000; } else { par->PCIO = par->PCIO0; par->PCRTC = par->PCRTC0; par->PRAMDAC = par->PRAMDAC0; par->PDIO = par->PDIO0; } } static void nv4GetConfig(struct nvidia_par *par) { if (NV_RD32(par->PFB, 0x0000) & 0x00000100) { par->RamAmountKBytes = ((NV_RD32(par->PFB, 0x0000) >> 12) & 0x0F) * 1024 * 2 + 1024 * 2; } else { switch (NV_RD32(par->PFB, 0x0000) & 0x00000003) { case 0: par->RamAmountKBytes = 1024 * 32; break; case 1: par->RamAmountKBytes = 1024 * 4; break; case 2: par->RamAmountKBytes = 1024 * 8; break; case 3: default: par->RamAmountKBytes = 1024 * 16; break; } } par->CrystalFreqKHz = (NV_RD32(par->PEXTDEV, 0x0000) & 0x00000040) ? 14318 : 13500; par->CURSOR = &par->PRAMIN[0x1E00]; par->MinVClockFreqKHz = 12000; par->MaxVClockFreqKHz = 350000; } static void nv10GetConfig(struct nvidia_par *par) { struct pci_dev *dev; u32 implementation = par->Chipset & 0x0ff0; #ifdef __BIG_ENDIAN /* turn on big endian register access */ if (!(NV_RD32(par->PMC, 0x0004) & 0x01000001)) { NV_WR32(par->PMC, 0x0004, 0x01000001); mb(); } #endif dev = pci_get_bus_and_slot(0, 1); if ((par->Chipset & 0xffff) == 0x01a0) { u32 amt; pci_read_config_dword(dev, 0x7c, &amt); par->RamAmountKBytes = (((amt >> 6) & 31) + 1) * 1024; } else if ((par->Chipset & 0xffff) == 0x01f0) { u32 amt; pci_read_config_dword(dev, 0x84, &amt); par->RamAmountKBytes = (((amt >> 4) & 127) + 1) * 1024; } else { par->RamAmountKBytes = (NV_RD32(par->PFB, 0x020C) & 0xFFF00000) >> 10; } pci_dev_put(dev); par->CrystalFreqKHz = (NV_RD32(par->PEXTDEV, 0x0000) & (1 << 6)) ? 14318 : 13500; if (par->twoHeads && (implementation != 0x0110)) { if (NV_RD32(par->PEXTDEV, 0x0000) & (1 << 22)) par->CrystalFreqKHz = 27000; } par->CURSOR = NULL; /* can't set this here */ par->MinVClockFreqKHz = 12000; par->MaxVClockFreqKHz = par->twoStagePLL ? 400000 : 350000; } int NVCommonSetup(struct fb_info *info) { struct nvidia_par *par = info->par; struct fb_var_screeninfo *var; u16 implementation = par->Chipset & 0x0ff0; u8 *edidA = NULL, *edidB = NULL; struct fb_monspecs *monitorA, *monitorB; struct fb_monspecs *monA = NULL, *monB = NULL; int mobile = 0; int tvA = 0; int tvB = 0; int FlatPanel = -1; /* really means the CRTC is slaved */ int Television = 0; int err = 0; var = kzalloc(sizeof(struct fb_var_screeninfo), GFP_KERNEL); monitorA = kzalloc(sizeof(struct fb_monspecs), GFP_KERNEL); monitorB = kzalloc(sizeof(struct fb_monspecs), GFP_KERNEL); if (!var || !monitorA || !monitorB) { err = -ENOMEM; goto done; } par->PRAMIN = par->REGS + (0x00710000 / 4); par->PCRTC0 = par->REGS + (0x00600000 / 4); par->PRAMDAC0 = par->REGS + (0x00680000 / 4); par->PFB = par->REGS + (0x00100000 / 4); par->PFIFO = par->REGS + (0x00002000 / 4); par->PGRAPH = par->REGS + (0x00400000 / 4); par->PEXTDEV = par->REGS + (0x00101000 / 4); par->PTIMER = par->REGS + (0x00009000 / 4); par->PMC = par->REGS + (0x00000000 / 4); par->FIFO = par->REGS + (0x00800000 / 4); /* 8 bit registers */ par->PCIO0 = (u8 __iomem *) par->REGS + 0x00601000; par->PDIO0 = (u8 __iomem *) par->REGS + 0x00681000; par->PVIO = (u8 __iomem *) par->REGS + 0x000C0000; par->twoHeads = (par->Architecture >= NV_ARCH_10) && (implementation != 0x0100) && (implementation != 0x0150) && (implementation != 0x01A0) && (implementation != 0x0200); par->fpScaler = (par->FpScale && par->twoHeads && (implementation != 0x0110)); par->twoStagePLL = (implementation == 0x0310) || (implementation == 0x0340) || (par->Architecture >= NV_ARCH_40); par->WaitVSyncPossible = (par->Architecture >= NV_ARCH_10) && (implementation != 0x0100); par->BlendingPossible = ((par->Chipset & 0xffff) != 0x0020); /* look for known laptop chips */ switch (par->Chipset & 0xffff) { case 0x0112: case 0x0174: case 0x0175: case 0x0176: case 0x0177: case 0x0179: case 0x017C: case 0x017D: case 0x0186: case 0x0187: case 0x018D: case 0x01D7: case 0x0228: case 0x0286: case 0x028C: case 0x0316: case 0x0317: case 0x031A: case 0x031B: case 0x031C: case 0x031D: case 0x031E: case 0x031F: case 0x0324: case 0x0325: case 0x0328: case 0x0329: case 0x032C: case 0x032D: case 0x0347: case 0x0348: case 0x0349: case 0x034B: case 0x034C: case 0x0160: case 0x0166: case 0x0169: case 0x016B: case 0x016C: case 0x016D: case 0x00C8: case 0x00CC: case 0x0144: case 0x0146: case 0x0147: case 0x0148: case 0x0098: case 0x0099: mobile = 1; break; default: break; } if (par->Architecture == NV_ARCH_04) nv4GetConfig(par); else nv10GetConfig(par); NVSelectHeadRegisters(par, 0); NVLockUnlock(par, 0); par->IOBase = (NVReadMiscOut(par) & 0x01) ? 0x3d0 : 0x3b0; par->Television = 0; nvidia_create_i2c_busses(par); if (!par->twoHeads) { par->CRTCnumber = 0; if (nvidia_probe_i2c_connector(info, 1, &edidA)) nvidia_probe_of_connector(info, 1, &edidA); if (edidA && !fb_parse_edid(edidA, var)) { printk("nvidiafb: EDID found from BUS1\n"); monA = monitorA; fb_edid_to_monspecs(edidA, monA); FlatPanel = (monA->input & FB_DISP_DDI) ? 1 : 0; /* NV4 doesn't support FlatPanels */ if ((par->Chipset & 0x0fff) <= 0x0020) FlatPanel = 0; } else { VGA_WR08(par->PCIO, 0x03D4, 0x28); if (VGA_RD08(par->PCIO, 0x03D5) & 0x80) { VGA_WR08(par->PCIO, 0x03D4, 0x33); if (!(VGA_RD08(par->PCIO, 0x03D5) & 0x01)) Television = 1; FlatPanel = 1; } else { FlatPanel = 0; } printk("nvidiafb: HW is currently programmed for %s\n", FlatPanel ? (Television ? "TV" : "DFP") : "CRT"); } if (par->FlatPanel == -1) { par->FlatPanel = FlatPanel; par->Television = Television; } else { printk("nvidiafb: Forcing display type to %s as " "specified\n", par->FlatPanel ? "DFP" : "CRT"); } } else { u8 outputAfromCRTC, outputBfromCRTC; int CRTCnumber = -1; u8 slaved_on_A, slaved_on_B; int analog_on_A, analog_on_B; u32 oldhead; u8 cr44; if (implementation != 0x0110) { if (NV_RD32(par->PRAMDAC0, 0x0000052C) & 0x100) outputAfromCRTC = 1; else outputAfromCRTC = 0; if (NV_RD32(par->PRAMDAC0, 0x0000252C) & 0x100) outputBfromCRTC = 1; else outputBfromCRTC = 0; analog_on_A = NVIsConnected(par, 0); analog_on_B = NVIsConnected(par, 1); } else { outputAfromCRTC = 0; outputBfromCRTC = 1; analog_on_A = 0; analog_on_B = 0; } VGA_WR08(par->PCIO, 0x03D4, 0x44); cr44 = VGA_RD08(par->PCIO, 0x03D5); VGA_WR08(par->PCIO, 0x03D5, 3); NVSelectHeadRegisters(par, 1); NVLockUnlock(par, 0); VGA_WR08(par->PCIO, 0x03D4, 0x28); slaved_on_B = VGA_RD08(par->PCIO, 0x03D5) & 0x80; if (slaved_on_B) { VGA_WR08(par->PCIO, 0x03D4, 0x33); tvB = !(VGA_RD08(par->PCIO, 0x03D5) & 0x01); } VGA_WR08(par->PCIO, 0x03D4, 0x44); VGA_WR08(par->PCIO, 0x03D5, 0); NVSelectHeadRegisters(par, 0); NVLockUnlock(par, 0); VGA_WR08(par->PCIO, 0x03D4, 0x28); slaved_on_A = VGA_RD08(par->PCIO, 0x03D5) & 0x80; if (slaved_on_A) { VGA_WR08(par->PCIO, 0x03D4, 0x33); tvA = !(VGA_RD08(par->PCIO, 0x03D5) & 0x01); } oldhead = NV_RD32(par->PCRTC0, 0x00000860); NV_WR32(par->PCRTC0, 0x00000860, oldhead | 0x00000010); if (nvidia_probe_i2c_connector(info, 1, &edidA)) nvidia_probe_of_connector(info, 1, &edidA); if (edidA && !fb_parse_edid(edidA, var)) { printk("nvidiafb: EDID found from BUS1\n"); monA = monitorA; fb_edid_to_monspecs(edidA, monA); } if (nvidia_probe_i2c_connector(info, 2, &edidB)) nvidia_probe_of_connector(info, 2, &edidB); if (edidB && !fb_parse_edid(edidB, var)) { printk("nvidiafb: EDID found from BUS2\n"); monB = monitorB; fb_edid_to_monspecs(edidB, monB); } if (slaved_on_A && !tvA) { CRTCnumber = 0; FlatPanel = 1; printk("nvidiafb: CRTC 0 is currently programmed for " "DFP\n"); } else if (slaved_on_B && !tvB) { CRTCnumber = 1; FlatPanel = 1; printk("nvidiafb: CRTC 1 is currently programmed " "for DFP\n"); } else if (analog_on_A) { CRTCnumber = outputAfromCRTC; FlatPanel = 0; printk("nvidiafb: CRTC %i appears to have a " "CRT attached\n", CRTCnumber); } else if (analog_on_B) { CRTCnumber = outputBfromCRTC; FlatPanel = 0; printk("nvidiafb: CRTC %i appears to have a " "CRT attached\n", CRTCnumber); } else if (slaved_on_A) { CRTCnumber = 0; FlatPanel = 1; Television = 1; printk("nvidiafb: CRTC 0 is currently programmed " "for TV\n"); } else if (slaved_on_B) { CRTCnumber = 1; FlatPanel = 1; Television = 1; printk("nvidiafb: CRTC 1 is currently programmed for " "TV\n"); } else if (monA) { FlatPanel = (monA->input & FB_DISP_DDI) ? 1 : 0; } else if (monB) { FlatPanel = (monB->input & FB_DISP_DDI) ? 1 : 0; } if (par->FlatPanel == -1) { if (FlatPanel != -1) { par->FlatPanel = FlatPanel; par->Television = Television; } else { printk("nvidiafb: Unable to detect display " "type...\n"); if (mobile) { printk("...On a laptop, assuming " "DFP\n"); par->FlatPanel = 1; } else { printk("...Using default of CRT\n"); par->FlatPanel = 0; } } } else { printk("nvidiafb: Forcing display type to %s as " "specified\n", par->FlatPanel ? "DFP" : "CRT"); } if (par->CRTCnumber == -1) { if (CRTCnumber != -1) par->CRTCnumber = CRTCnumber; else { printk("nvidiafb: Unable to detect which " "CRTCNumber...\n"); if (par->FlatPanel) par->CRTCnumber = 1; else par->CRTCnumber = 0; printk("...Defaulting to CRTCNumber %i\n", par->CRTCnumber); } } else { printk("nvidiafb: Forcing CRTCNumber %i as " "specified\n", par->CRTCnumber); } if (monA) { if (((monA->input & FB_DISP_DDI) && par->FlatPanel) || ((!(monA->input & FB_DISP_DDI)) && !par->FlatPanel)) { if (monB) { fb_destroy_modedb(monB->modedb); monB = NULL; } } else { fb_destroy_modedb(monA->modedb); monA = NULL; } } if (monB) { if (((monB->input & FB_DISP_DDI) && !par->FlatPanel) || ((!(monB->input & FB_DISP_DDI)) && par->FlatPanel)) { fb_destroy_modedb(monB->modedb); monB = NULL; } else monA = monB; } if (implementation == 0x0110) cr44 = par->CRTCnumber * 0x3; NV_WR32(par->PCRTC0, 0x00000860, oldhead); VGA_WR08(par->PCIO, 0x03D4, 0x44); VGA_WR08(par->PCIO, 0x03D5, cr44); NVSelectHeadRegisters(par, par->CRTCnumber); } printk("nvidiafb: Using %s on CRTC %i\n", par->FlatPanel ? (par->Television ? "TV" : "DFP") : "CRT", par->CRTCnumber); if (par->FlatPanel && !par->Television) { par->fpWidth = NV_RD32(par->PRAMDAC, 0x0820) + 1; par->fpHeight = NV_RD32(par->PRAMDAC, 0x0800) + 1; par->fpSyncs = NV_RD32(par->PRAMDAC, 0x0848) & 0x30000033; printk("nvidiafb: Panel size is %i x %i\n", par->fpWidth, par->fpHeight); } if (monA) info->monspecs = *monA; if (!par->FlatPanel || !par->twoHeads) par->FPDither = 0; par->LVDS = 0; if (par->FlatPanel && par->twoHeads) { NV_WR32(par->PRAMDAC0, 0x08B0, 0x00010004); if (NV_RD32(par->PRAMDAC0, 0x08b4) & 1) par->LVDS = 1; printk("nvidiafb: Panel is %s\n", par->LVDS ? "LVDS" : "TMDS"); } kfree(edidA); kfree(edidB); done: kfree(var); kfree(monitorA); kfree(monitorB); return err; }