New ATI Framebuffer Xcode project to start from

0xdeadbeef · March 20, 2009

And I'm a little confused here about whether just running the card BIOS got it POSTed or you have to additionaly running ATOM init functions to get job done?

Without int10, there is also another problem of accessing the video BIOS as it's not located at 0xC0000 (already used by primary card) and you got no idea where it will be (maybe not in system memory at all). Without access to the BIOS image, there is no way to call ATOM init functions.

Maybe you can get all that info by calling IOACPIPlatformDevice methods:

_ROM (Get ROM data)

_GPD (Get POST Device)

_SPD (Set POST Device)

See pages 576..578 of the ACPI Spec

Although if this can be done from a display driver I have no idea. I do know that ATY_* framebuffers call IOACPIPlatformDevice->_DSS (Device Set State)

np_ · March 20, 2009

here is more info

http://www.x.org/docs/AMD/

http://www.x.org/wiki/radeonhd

what you need to do is new iokit driver who loads before any ati***.kext ( something like natit )

mapping ATI mem base IO ( check docs for correct registers, they are almost same for any ati card )

and you can read/write anywhere in card mem space , that include to read EDID , card bios ..etc

getting correct info you can push it into ioregs without no problem at all

just note - avoid to hard code EDID info data into ioregs or any data copied from real mac's ioregs, it will cause in most situation black screen or even put video card in funny moods

netkas · March 20, 2009

dong, the bios you can always get from ioreg, for r6xx+ cards its there under name ATY,bin_image

and here is what radeonhd x86 does (from rhd_atombios.c)

if (unposted) {

/* run AsicInit */

if (!rhdAtomASICInit(handle))

xf86DrvMsg(scrnIndex, X_WARNING,

"%s: AsicInit failed. Won't be able to obtain in VRAM "

"FB scratch space\n",__func__);

static Bool

rhdAtomASICInit(atomBiosHandlePtr handle)

{

ASIC_INIT_PS_ALLOCATION asicInit;

AtomBiosArgRec data;

RHDFUNC(handle);

RHDAtomBiosFunc(handle->scrnIndex, handle,

GET_DEFAULT_ENGINE_CLOCK,

&data);

asicInit.sASICInitClocks.ulDefaultEngineClock = data.val / 10;/*in 10 Khz*/

RHDAtomBiosFunc(handle->scrnIndex, handle,

GET_DEFAULT_MEMORY_CLOCK,

&data);

asicInit.sASICInitClocks.ulDefaultMemoryClock = data.val / 10;/*in 10 Khz*/

data.exec.dataSpace = NULL;

data.exec.index = GetIndexIntoMasterTable(COMMAND, ASIC_Init);

data.exec.pspace = &asicInit;

xf86DrvMsg(handle->scrnIndex, X_INFO, "Calling ASIC Init\n");

atomDebugPrintPspace(handle, &data, sizeof(asicInit));

if (RHDAtomBiosFunc(handle->scrnIndex, handle,

ATOMBIOS_EXEC, &data) == ATOM_SUCCESS) {

xf86DrvMsg(handle->scrnIndex, X_INFO, "ASIC_INIT Successful\n");

return TRUE;

}

xf86DrvMsg(handle->scrnIndex, X_INFO, "ASIC_INIT Failed\n");

return FALSE;

}

netkas · March 20, 2009

ah well, framebuffer already has _atom_ASIC_Init function, it called on wakeup.

dong · March 20, 2009

netkas, I checked your natit code, you're still getting BIOS ROM from the legacy address 0xC0000 if the ROM is not present in plist file. It won't work for a secondary ATI card.

So, do you want to modify the native driver some way to make it execute atom_ASIC_Init at initialization time?

Or a tiny driver is still needed to do this job?

netkas · March 20, 2009

i get it from 0xc0000 only if binimage wasnt found in plist, that;s answer :blink:

well. i modified motmot that way, and it's attitude changed when card run in vga mode, not vesa, but still isnt perfect.

dong · March 21, 2009

Netkas, I put rhdAtomInit related stuff together. It's compiled ok but not tested yet. If it does not do its job or causes KP for some reason, don't blame me.

In the probe function, I assume BIOS can be read from IOPCIDevice's "ATY,bin_image". You have to make sure Natit load before this and set that to IOPCIDevice.

Now if BIOS can not be read from "ATY,bin_image", the BIOS at legacy address will be used instead. Adding this is because I don't have a "ATY,bin_image" to debug the code.

Find it in thanoulas's post in this topic if you want to download.

netkas · March 22, 2009

seems like you forgot to add some sources to it

kld(): Undefined symbols:

_DelayMicroseconds

_DelayMilliseconds

_ReadIndReg32

_ReadPCIReg16

_ReadPCIReg32

_ReadPCIReg8

_ReadReg32

_ReadSysIOReg16

_ReadSysIOReg32

_ReadSysIOReg8

_WriteIndReg32

_WritePCIReg16

_WritePCIReg32

_WritePCIReg8

_WriteReg32

_WriteSysIOReg16

_WriteSysIOReg32

_WriteSysIOReg8

I attached corrected info.plist, the default one doesnt work.

http://rapidshare.de/files/46265563/Info.plist.html

dong · March 22, 2009

My bad, didn't even test it with kextload. The undefined symbols is caused by atombios compiling macros. All the sources are already there. To solve this,

In CD_hw_services.h, add a line at the beginning, i.e., add

#define ENABLE_ALL_SERVICE_FUNCTIONS

right after

#include "CD_Structs.h"

You may also add the line at the beginning of hwserv_drv.c, all the function implementations are there.

The ENABLE_ALL_SERVICE_FUNCTIONS definition will enable them.

The whole project in my last post is also updated.

netkas · March 22, 2009

panics in atomSaveRegisters function

dong · March 22, 2009

panics in atomSaveRegisters function

Don't know exactly what caused it, possibly the memory allocation/deallocation stuff. And I do not see how useful can this function be, it seems only backup the registers each time when there is a writing but I do not find out where they are used to restore the values. So a simple suggestion to avoid this panic is to comment out all occasions of atomSaveRegisters in the code.

netkas · March 23, 2009

Sorry, i was a bit lazy to analize backtrace last time so made wrong assumption, so getting rid of saveregisters functions made no difference

panic was in __RHDRegWrite

│ 1950 ! mov edx, eax

│ 1952 ! mov eax, [ebp+offset_10]

│ 1955 ! mov [edx], eax

│ 1957 ! leave

│ 1958 ! ret

EIP - 1955

edx - a030 , eax - 0x02

dong · March 23, 2009

Just a quick response, I suddenly remember that I forgot to add a "device->setMemoryEnable(true)" in ATY_Init.cpp. I will try to debug further by myself, but you may give another try.

Update: after adding that, it now works for me, though the screen goes into a strange state. Here is the log:

dong/project/ATY_Init/build/Debug; USER=root; COMMAND=/sbin/kextload -t ATY_Init.kext
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadATIRegister(8) = 97
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: atomSaveRegisters
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister(8,16)
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadATIRegister© = 1ca
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadATIRegister(8) = 16
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: atomSaveRegisters
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister(8,96)
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: atomSaveRegisters
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister(c,1ca)
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadATIRegister(8) = 96
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: atomSaveRegisters
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister(8,9c)
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: atomSaveRegisters
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister(c,1a)
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadATIRegister(8) = 9c
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: atomSaveRegisters
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister(8,27)
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadATIRegister© = 24a
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadATIRegister(8) = 27
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: atomSaveRegisters
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister(8,a7)
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: atomSaveRegisters
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister(c,24a)
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadATIRegister(8) = a7
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: atomSaveRegisters
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister(8,28)
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadATIRegister© = 28a
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadATIRegister(8) = 28
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: atomSaveRegisters
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister(8,a8)
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: atomSaveRegisters
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister(c,28a)
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadPLL
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadPLL(26) = 17
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWritePLL
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWritePLL(26,13)
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: atomSaveRegisters
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadPLL
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadPLL(1e) = 17
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWritePLL
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWritePLL(1e,13)
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: atomSaveRegisters
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadPLL
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadPLL(1d) = 17
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWritePLL
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWritePLL(1d,13)
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: atomSaveRegisters
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: atomSaveRegisters
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister(574,1)
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadATIRegister(570) = 0
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: atomSaveRegisters
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister(570,0)
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: atomSaveRegisters
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister(580,c000000)
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: atomSaveRegisters
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister(4cc,1f000000)
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadPLL
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadPLL( = 100fac8f
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWritePLL
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWritePLL(b,100fac8f)
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: atomSaveRegisters
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadPLL
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadPLL( = 100fac8f
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWritePLL
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWritePLL(b,100fac13)
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: atomSaveRegisters
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadATIRegister(8) = 8b
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: atomSaveRegisters
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister(8,
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadATIRegister© = 100fac13
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadATIRegister(8) = b
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: atomSaveRegisters
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister(8,8b)
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: atomSaveRegisters
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister(c,100fac8f)
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadPLL
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadPLL( = 100fac8f
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWritePLL
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWritePLL(b,100fac8f)
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: atomSaveRegisters
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReleaseMemory
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReadATIRegister(10) = 40000
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: atomSaveRegisters
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailWriteATIRegister(10,40000)
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: CailReleaseMemory
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: ParseTable said: CD_SUCCESS
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: ASIC_INIT Successful
Mar 23 11:49:49 myLenovo kernel[0]: ATY_Init: Card initialize completed

I do not comment out atomSaveRegisters yet, but it's better to remove them since the memory allocated in it is not freed in the project.

LeoMark · March 25, 2009

Hi, netkas and dong, congratulations for the project! I am a user ati mobility hd2400, I have followed the work of you and I'm anxious for news, some hope for the LVDS connector? sorry for english, I'm Brazilian

netkas · March 28, 2009

latest still panics for me O.o

LeoMark · March 31, 2009

Hey guys ... do you think the project will not have a solution? the project will continue or is this the end of the ATI Mobility ...

dong · March 31, 2009

latest still panics for me O.o

I don't have a later ATI card to test the code. The _RHDReadMC and _RHDWriteMC are different for my x1400 compared to later cards, maybe panic here?

thanoulas · April 5, 2009

I don't have a later ATI card to test the code. The _RHDReadMC and _RHDWriteMC are different for my x1400 compared to later cards, maybe panic here?

Hi there. This is my first post in these forums and i'm really interested in the development of the ATY_HD Framebuffer. I'm a developer myself, so i could lend a hand if you'd like to.

Anyway, i've tested out the ATY_Init code in my hackbook with a mobility HD3650 (M86) and I can verify that it panicks.

Specifically, kernel panicks in rhd_init.c, function:

void _RHDRegWrite(RHDPtr rhdPtr, CARD16 offset, CARD32 value) {
 *(volatile CARD32 *)((CARD8 *)(rhdPtr->IObase) + offset)) = value;
}

CARD16 offset was 544c, CARD32 value was 8

_RHDRegRead previously returned 0 for the offset 544c

Hope it helps

super_engine · April 5, 2009

Hi

I am one of hundrends laptop users with an ATI Mobility card (X1350)

Please can attach a compiled file in this thread. So we, no developers, can test it ???

thank you in advance.

dong · April 6, 2009

Hi there. This is my first post in these forums and i'm really interested in the development of the ATY_HD Framebuffer. I'm a developer myself, so i could lend a hand if you'd like to.

Anyway, i've tested out the ATY_Init code in my hackbook with a mobility HD3650 (M86) and I can verify that it panicks.

Specifically, kernel panicks in rhd_init.c, function:
void _RHDRegWrite(RHDPtr rhdPtr, CARD16 offset, CARD32 value) {
  *(volatile CARD32 *)((CARD8 *)(rhdPtr->IObase) + offset)) = value;
}
CARD16 offset was 544c, CARD32 value was 8

_RHDRegRead previously returned 0 for the offset 544c

Hope it helps

Thanks for your interest and effort.

A roughly search of 0x544C did not give any results from radeonHD or ati open source. I then looked at the binary code of ATY_MegaLodon. It turns out the register address is visited in the procedures of PCIE_Gen1_Enable and PCIE_Gen2_Enable. Below is the disemblyed code of PCIE_Gen2_Enable:

OSStatus PCIE_Gen2_Enable(DriverGlobal *aDriverRecPtr) {
UInt32 value, value1;
pciepw32(aDriverRecPtr, 0xA2, pciepr32(aDriverRecPtr, 0xA2) & ~(1 << 13));
value = pciepr32(aDriverRecPtr, 0xA2);
if (value & (1 << 9))
	pciepw32(aDriverRecPtr, 0xA2, ((((value | (1 << 10)) & ~7) | ((value & 0x70) >> 4)) | (1 << 8)) & ~(1 << 7));
else
	pciepw32(aDriverRecPtr, 0xA2, value | (1 << 13));
value = pciepr32(aDriverRecPtr, 0xA4);
if ((value & (1 << 23)) && (value & (1 << 24))) {
	regw32(aDriverRecPtr, 0x5488, regr32(aDriverRecPtr, 5488) & ~(1 << 25));
	regw32(aDriverRecPtr, 0x548C, regr32(aDriverRecPtr, 548C) & ~(1 << 28));
	regw32(aDriverRecPtr, 0x5404, regr32(aDriverRecPtr, 5404) | (1 << 25));
	regw32(aDriverRecPtr, 0x5484, regr32(aDriverRecPtr, 5484) & ~0xF;
	regw32(aDriverRecPtr, 0x5410, regr32(aDriverRecPtr, 5410) | (1 << 23));
	regw32(aDriverRecPtr, 0x548C, regr32(aDriverRecPtr, 548C) & ~0x1E);
	value = (value | 1) & ~2;
	value1 = pciepr32(aDriverRecPtr, 0xA1);
	if (value1 & (1 << 6)) pciepw32(aDriverRecPtr, 0xA1, value1 & ~0x40);
	pciepw32(aDriverRecPtr, 0xA4, ((((value & ~0x300) | 0x300) & ~0x3C000) & ~0x40) | 0x20);
	regw32(aDriverRecPtr, 0x541C, regr32(aDriverRecPtr, 541C) | (1 << 3));
	regw32(aDriverRecPtr, 0x544C, 8);
	regw32(aDriverRecPtr, 0x4088, (regr32(aDriverRecPtr, 4088) & ~0xF) | 2);
	regw32(aDriverRecPtr, 0x544C, 0);
}
return noErr;
}

However, I could not find out why it could cause a kernel panic. Hope someone else could give an answer.

thanoulas · April 6, 2009

Thanks for your interest and effort.

A roughly search of 0x544C did not give any results from radeonHD or ati open source. I then looked at the binary code of ATY_MegaLodon. It turns out the register address is visited in the procedures of PCIE_Gen1_Enable and PCIE_Gen2_Enable. Below is the disemblyed code of PCIE_Gen2_Enable:

OSStatus PCIE_Gen2_Enable(DriverGlobal *aDriverRecPtr) {
 UInt32 value, value1;
 pciepw32(aDriverRecPtr, 0xA2, pciepr32(aDriverRecPtr, 0xA2) & ~(1 << 13));
 value = pciepr32(aDriverRecPtr, 0xA2);
 if (value & (1 << 9))
	 pciepw32(aDriverRecPtr, 0xA2, ((((value | (1 << 10)) & ~7) | ((value & 0x70) >> 4)) | (1 << 8)) & ~(1 << 7));
 else
	 pciepw32(aDriverRecPtr, 0xA2, value | (1 << 13));
 value = pciepr32(aDriverRecPtr, 0xA4);
 if ((value & (1 << 23)) && (value & (1 << 24))) {
	 regw32(aDriverRecPtr, 0x5488, regr32(aDriverRecPtr, 5488) & ~(1 << 25));
	 regw32(aDriverRecPtr, 0x548C, regr32(aDriverRecPtr, 548C) & ~(1 << 28));
	 regw32(aDriverRecPtr, 0x5404, regr32(aDriverRecPtr, 5404) | (1 << 25));
	 regw32(aDriverRecPtr, 0x5484, regr32(aDriverRecPtr, 5484) & ~0xF;
	 regw32(aDriverRecPtr, 0x5410, regr32(aDriverRecPtr, 5410) | (1 << 23));
	 regw32(aDriverRecPtr, 0x548C, regr32(aDriverRecPtr, 548C) & ~0x1E);
	 value = (value | 1) & ~2;
	 value1 = pciepr32(aDriverRecPtr, 0xA1);
	 if (value1 & (1 << 6)) pciepw32(aDriverRecPtr, 0xA1, value1 & ~0x40);
	 pciepw32(aDriverRecPtr, 0xA4, ((((value & ~0x300) | 0x300) & ~0x3C000) & ~0x40) | 0x20);
	 regw32(aDriverRecPtr, 0x541C, regr32(aDriverRecPtr, 541C) | (1 << 3));
	 regw32(aDriverRecPtr, 0x544C, 8);
	 regw32(aDriverRecPtr, 0x4088, (regr32(aDriverRecPtr, 4088) & ~0xF) | 2);
	 regw32(aDriverRecPtr, 0x544C, 0);
 }
 return noErr;
}

However, I could not find out why it could cause a kernel panic. Hope someone else could give an answer.

Well I do have another hackbook that I can use to debug the kext with the kernel debug kit. I'll give it a try and tell you what happens.

Another interesting (maybe?) thing is that I skipped the 544c offset and now it panics at the exact same address that is next in the Megalodon fb, 0x4088.

ridgeline · April 8, 2009

oh for the love of OSX thank you guys for not giving up on this project

Slice · April 16, 2009

Hi there. This is my first post in these forums and i'm really interested in the development of the ATY_HD Framebuffer. I'm a developer myself, so i could lend a hand if you'd like to.

Anyway, i've tested out the ATY_Init code in my hackbook with a mobility HD3650 (M86) and I can verify that it panicks.

Specifically, kernel panicks in rhd_init.c, function:
void _RHDRegWrite(RHDPtr rhdPtr, CARD16 offset, CARD32 value) {
 *(volatile CARD32 *)((CARD8 *)(rhdPtr->IObase) + offset)) = value;
}
CARD16 offset was 544c, CARD32 value was 8

_RHDRegRead previously returned 0 for the offset 544c

Hope it helps

Bad offset can't cause kernel panic!

Kernel panic can be caused by bad rhdPtr. Check!

dong · April 16, 2009

Bad offset can't cause kernel panic!
Kernel panic can be caused by bad rhdPtr. Check!

If it was the case, somehow IOmap was read wrong for those cards. People can try change two lines in ATY_Init.cpp from

IOmap = device->mapDeviceMemoryWithIndex(RHD_MMIO_BAR);

to

IOmap = device->mapDeviceMemoryWithRegister(kIOPCIConfigBaseAddress2);

and from

FBmap = device->mapDeviceMemoryWithIndex(RHD_FB_BAR);

to

FBmap = device->mapDeviceMemoryWithRegister(kIOPCIConfigBaseAddress0);

"mapDeviceMemoryWithRegister" is used by native drivers and I think it should found the correct map.

thanoulas · April 16, 2009

If it was the case, somehow IOmap was read wrong for those cards. People can try change two lines in ATY_Init.cpp from
IOmap = device->mapDeviceMemoryWithIndex(RHD_MMIO_BAR);
to
IOmap = device->mapDeviceMemoryWithRegister(kIOPCIConfigBaseAddress2);
and from
FBmap = device->mapDeviceMemoryWithIndex(RHD_FB_BAR);
to
FBmap = device->mapDeviceMemoryWithRegister(kIOPCIConfigBaseAddress0);
"mapDeviceMemoryWithRegister" is used by native drivers and I think it should found the correct map.

I'll check that and report back ASAP

New ATI Framebuffer Xcode project to start from

Recommended Posts

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites