Clover test and patches for Polaris GPU

fantomas · September 5, 2016

THIS TOPIC IS THE RESULT OF VARIUS SPLITTED POST FROM CLOVER GENERAL DISCUSSION

THE MAIN GOAL OF THAT IS KEEP TRACK OF THE RESULT IN THE TEST AND MAYBE FIND POSSIBLE SOLUTION ON THE BOOT HANG WITH POLARIS CARDS WITHOUT THE HELP OF PRIMARY iGPU

ALSO TO KEEP THE CLOVER GENERAL DISCUSSION TOPIC CLEAN...

Other related topic:

- AMD Polaris IDs on 10.12 Sierra

- 10.12 Sierra AMD Framebuffer dump script

ErmaC (November, 12 2016)

Hi guys

A question from a noob: it's about Clover support for Polaris graphics cards

https://sourceforge.net/p/cloverefiboot/code/HEAD/tree/rEFIt_UEFI/Platform/ati.c (line 277)

  //Polaris
  { 0x67C0,	0x00000000, CHIP_FAMILY_HAWAII,	"AMD Radeon Polaris 10",        kNull	},
  { 0x67DF,	0x00000000, CHIP_FAMILY_HAWAII,	"AMD Radeon RX480",             kNull	},
  { 0x67E0,	0x00000000, CHIP_FAMILY_HAWAII,	"AMD Radeon RX460",             kNull	},
  { 0x67EF,	0x00000000, CHIP_FAMILY_HAWAII,	"AMD Radeon RX460",             kNull	},
  { 0x67FF,	0x00000000, CHIP_FAMILY_HAWAII,	"AMD Radeon Polaris 11",

Shouldn't we have CHIP_FAMILY_POLARIS instead of CHIP_FAMILY_HAWAII???

Or maybe CHIP_FAMILY_BAFFIN (for Polaris 11) and CHIP_FAMILY_ELLESMERE (for Polaris 10)???

Also, are you sure about RX460 with 0x67E0??? According to AMD and Linux Kernel Driver DataBase, there's no RX 460 with that device.

Here, all known Polaris graphics cards device-ids

vendor: 1002 ("Advanced Micro Devices, Inc. [AMD/ATI]"), device: 67c0 ("Ellesmere [Polaris10]")
vendor: 1002 ("Advanced Micro Devices, Inc. [AMD/ATI]"), device: 67c1
vendor: 1002 ("Advanced Micro Devices, Inc. [AMD/ATI]"), device: 67c2
vendor: 1002 ("Advanced Micro Devices, Inc. [AMD/ATI]"), device: 67c4
vendor: 1002 ("Advanced Micro Devices, Inc. [AMD/ATI]"), device: 67c7
vendor: 1002 ("Advanced Micro Devices, Inc. [AMD/ATI]"), device: 67c8
vendor: 1002 ("Advanced Micro Devices, Inc. [AMD/ATI]"), device: 67c9
vendor: 1002 ("Advanced Micro Devices, Inc. [AMD/ATI]"), device: 67ca
vendor: 1002 ("Advanced Micro Devices, Inc. [AMD/ATI]"), device: 67cc
vendor: 1002 ("Advanced Micro Devices, Inc. [AMD/ATI]"), device: 67cf
vendor: 1002 ("Advanced Micro Devices, Inc. [AMD/ATI]"), device: 67df ("Ellesmere [Radeon RX 470/480]")
vendor: 1002 ("Advanced Micro Devices, Inc. [AMD/ATI]"), device: 67e0 ("Baffin [Polaris11]")
vendor: 1002 ("Advanced Micro Devices, Inc. [AMD/ATI]"), device: 67e1 ("Baffin [Polaris11]")
vendor: 1002 ("Advanced Micro Devices, Inc. [AMD/ATI]"), device: 67e3
vendor: 1002 ("Advanced Micro Devices, Inc. [AMD/ATI]"), device: 67e7
vendor: 1002 ("Advanced Micro Devices, Inc. [AMD/ATI]"), device: 67e8 ("Baffin [Polaris11]")
vendor: 1002 ("Advanced Micro Devices, Inc. [AMD/ATI]"), device: 67e9 ("Baffin [Polaris11]")
vendor: 1002 ("Advanced Micro Devices, Inc. [AMD/ATI]"), device: 67eb ("Baffin [Polaris11]")
vendor: 1002 ("Advanced Micro Devices, Inc. [AMD/ATI]"), device: 67ef ("Baffin [Radeon RX 460]")
vendor: 1002 ("Advanced Micro Devices, Inc. [AMD/ATI]"), device: 67ff ("Baffin [Polaris11]")

Edited November 12, 2016 by ErmaC
Split from Clover general discussion

Ciro82 · September 5, 2016

Also, are you sure about RX460 with 0x67E0??? According to AMD and Linux Kernel Driver DataBase, there's no RX 460 with that device.

From Linux:

+ /* Polaris11 */

+ {0x1002, 0x67E0, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS11},

+ {0x1002, 0x67E1, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS11},

+ {0x1002, 0x67E8, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS11},

+ {0x1002, 0x67E9, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS11},

+ {0x1002, 0x67EB, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS11},

+ {0x1002, 0x67FF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS11},

+ /* Polaris10 */

+ {0x1002, 0x67C0, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS10},

+ {0x1002, 0x67DF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS10},

fantomas · September 5, 2016

From Linux:

+ /* Polaris11 */

+ {0x1002, 0x67E0, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS11},

+ {0x1002, 0x67E1, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS11},

+ {0x1002, 0x67E8, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS11},

+ {0x1002, 0x67E9, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS11},

+ {0x1002, 0x67EB, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS11},

+ {0x1002, 0x67FF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS11},

+ /* Polaris10 */

+ {0x1002, 0x67C0, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS10},

+ {0x1002, 0x67DF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS10},

Yes but it does not show the model name

EDIT:

CHIP_FAMILY_POLARIS10

CHIP_FAMILY_POLARIS11

it sounds good (instead of CHIP_FAMILY_HAWAII).

Ciro82 · September 5, 2016

Yes but it does not show the model name

Yeah, your list is more accurate

Maybe Mobile models?

fantomas · September 5, 2016

Yeah, your list is more accurate

Maybe Mobile models?

Probably

Needy · September 6, 2016

Yes but it does not show the model name

EDIT:

CHIP_FAMILY_POLARIS10

CHIP_FAMILY_POLARIS11

it sounds good (instead of CHIP_FAMILY_HAWAII).

Hello there,

We'll look into your concern.

Best regards.

Slice · September 6, 2016

From Linux:

+ /* Polaris11 */

+ {0x1002, 0x67E0, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS11},

+ {0x1002, 0x67E1, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS11},

+ {0x1002, 0x67E8, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS11},

+ {0x1002, 0x67E9, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS11},

+ {0x1002, 0x67EB, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS11},

+ {0x1002, 0x67FF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS11},

+ /* Polaris10 */

+ {0x1002, 0x67C0, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS10},

+ {0x1002, 0x67DF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS10},

Thanks, I will add this to Clover sources when I'll return home.

But I have to notice that it will not influence on macOS, just naming.

For modern cards in Sierra Inject->ATI=false is the best solution.

Hi,

Is it possible with clover to solve this ssdtprgen script warning?

Warning: 'system-type' may be set improperly (1 instead of 3)

I have also had a cpu warning that I solved in clover adding type parameter for my cpu.

Any advice?

system-type calculated from ModelName in Smbios section of your config.plist.

Set all to default and be happy.

PS. I don't use ssdtprgen

fantomas · November 4, 2016

@Slice

@Slice can you add the 9500controller in ATI.h and ATI.c with related Framebuffers? Please

Kext /System/Library/Extensions/AMD9500Controller.kext/Contents/MacOS/AMD9500Controller
Personality: Exmoor
ConnectorInfo count in decimal: 6
Disk offset in decimal 1892144
0000000 02 00 00 00 00 01 00 00 00 01 01 51 00 00 00 00 22 05 02 04 00 00 00 00
0000024 02 00 00 00 00 01 00 00 00 01 02 61 01 00 00 00 12 04 01 03 00 00 00 00
0000048 00 04 00 00 04 03 00 00 00 01 03 43 00 00 00 00 11 02 03 01 00 00 00 00
0000072 00 04 00 00 00 01 00 00 00 01 04 31 00 00 00 00 21 03 05 05 00 00 00 00
0000096 00 04 00 00 04 03 00 00 00 01 05 23 00 00 00 00 10 00 04 02 00 00 00 00
0000120 00 04 00 00 00 01 00 00 00 01 06 11 00 00 00 00 20 01 05 05 00 00 00 00
0000144
Personality: Berbice
ConnectorInfo count in decimal: 5
Disk offset in decimal 1892288
0000000 02 00 00 00 00 01 00 00 39 05 01 08 00 00 00 00 20 01 01 01 00 00 00 00
0000024 00 04 00 00 00 01 00 00 00 01 02 43 00 00 00 00 10 00 02 02 00 00 00 00
0000048 00 04 00 00 04 03 00 00 00 01 03 13 00 00 00 00 21 03 03 03 00 00 00 00
0000072 00 04 00 00 04 03 00 00 00 01 04 53 00 00 00 00 11 02 04 04 00 00 00 00
0000096 00 04 00 00 04 03 00 00 00 01 05 33 00 00 00 00 12 04 05 05 00 00 00 00
0000120
Personality: Elqui
ConnectorInfo count in decimal: 5
Disk offset in decimal 1892416
0000000 02 00 00 00 00 01 00 00 00 01 01 41 00 00 00 00 11 02 02 04 00 00 00 00
0000024 02 00 00 00 00 01 00 00 00 01 02 31 01 00 00 00 21 03 01 03 00 00 00 00
0000048 00 04 00 00 04 03 00 00 00 01 03 03 00 00 00 00 20 01 03 01 00 00 00 00
0000072 00 04 00 00 04 03 00 00 00 01 04 03 00 00 00 00 12 04 04 02 00 00 00 00
0000096 00 04 00 00 04 03 00 00 00 01 05 01 00 00 00 00 10 00 05 05 00 00 00 00
0000120
Personality: Acre
ConnectorInfo count in decimal: 3
Disk offset in decimal 1892640
0000000 00 04 00 00 04 03 00 00 00 01 01 01 00 00 00 00 11 02 02 01 00 00 00 00
0000024 00 08 00 00 04 02 00 00 00 01 02 00 00 00 00 00 21 03 05 04 00 00 00 00
0000048 04 00 00 00 04 02 00 00 00 01 03 00 00 00 00 00 00 00 03 05 00 00 00 00
0000072
Personality: Caroni
ConnectorInfo count in decimal: 5
Disk offset in decimal 1892832
0000000 02 00 00 00 00 01 00 00 00 01 01 41 00 00 00 00 11 02 02 04 00 00 00 00
0000024 02 00 00 00 00 01 00 00 00 01 02 31 01 00 00 00 21 03 01 03 00 00 00 00
0000048 00 04 00 00 04 03 00 00 00 01 03 03 00 00 00 00 20 01 03 01 00 00 00 00
0000072 00 04 00 00 04 03 00 00 00 01 04 03 00 00 00 00 12 04 04 02 00 00 00 00
0000096 00 04 00 00 04 03 00 00 00 01 05 01 00 00 00 00 10 00 05 05 00 00 00 00
0000120
Personality: Florin
ConnectorInfo count in decimal: 6
Disk offset in decimal 1893056
0000000 02 00 00 00 00 01 00 00 00 01 01 41 00 00 00 00 11 02 02 04 00 00 00 00
0000024 02 00 00 00 00 01 00 00 00 01 02 31 01 00 00 00 21 03 01 03 00 00 00 00
0000048 00 04 00 00 04 03 00 00 00 01 03 13 00 00 00 00 20 01 03 01 00 00 00 00
0000072 00 04 00 00 04 03 00 00 00 01 04 21 00 00 00 00 10 00 05 05 00 00 00 00
0000096 00 04 00 00 04 03 00 00 00 01 05 53 00 00 00 00 22 05 04 02 00 00 00 00
0000120 00 04 00 00 04 03 00 00 00 01 06 61 00 00 00 00 12 04 06 06 00 00 00 00
0000144
Personality: Guariba
ConnectorInfo count in decimal: 6
Disk offset in decimal 1893232
0000000 00 04 00 00 04 03 00 00 00 01 01 01 00 00 00 00 12 04 06 01 00 00 00 00
0000024 00 04 00 00 04 03 00 00 00 01 02 01 00 00 00 00 22 05 04 03 00 00 00 00
0000048 00 04 00 00 04 03 00 00 00 01 03 01 00 00 00 00 11 02 01 02 00 00 00 00
0000072 00 08 00 00 04 02 00 00 00 01 04 00 00 00 00 00 21 03 05 04 00 00 00 00
0000096 04 00 00 00 04 02 00 00 00 01 05 00 00 00 00 00 00 00 03 06 00 00 00 00
0000120 00 04 00 00 00 01 00 00 00 01 06 01 00 00 00 00 20 01 02 05 00 00 00 00
0000144
Personality: Dayman
ConnectorInfo count in decimal: 6
Disk offset in decimal 1893376
0000000 00 04 00 00 04 03 00 00 00 01 01 01 00 00 00 00 12 04 06 01 00 00 00 00
0000024 00 04 00 00 04 03 00 00 00 01 02 01 00 00 00 00 22 05 04 03 00 00 00 00
0000048 00 04 00 00 04 03 00 00 00 01 03 01 00 00 00 00 11 02 01 02 00 00 00 00
0000072 00 08 00 00 04 02 00 00 00 01 04 00 00 00 00 00 21 03 05 04 00 00 00 00
0000096 04 00 00 00 04 02 00 00 00 01 05 00 00 00 00 00 00 00 03 06 00 00 00 00
0000120 00 04 00 00 00 01 00 00 00 01 06 01 00 00 00 00 20 01 02 05 00 00 00 00
0000144
Personality: Baladi
ConnectorInfo count in decimal: 6
Disk offset in decimal 1893552
0000000 00 04 00 00 04 03 00 00 00 01 03 00 00 00 00 00 12 04 03 03 00 00 00 00
0000024 00 04 00 00 04 03 00 00 00 01 01 00 00 00 00 00 11 02 01 01 00 00 00 00
0000048 00 04 00 00 04 03 00 00 00 01 02 00 00 00 00 00 21 03 02 02 00 00 00 00
0000072 00 04 00 00 04 03 00 00 00 01 04 00 00 00 00 00 22 05 04 04 00 00 00 00
0000096 00 04 00 00 04 03 00 00 00 01 05 00 00 00 00 00 10 00 05 05 00 00 00 00
0000120 00 04 00 00 04 03 00 00 00 01 06 00 00 00 00 00 20 01 06 06 00 00 00 00
0000144
Personality: OPM
ConnectorInfo count in decimal: 5
Disk offset in decimal 1893696
0000000 00 04 00 00 04 03 00 00 00 00 00 01 00 00 00 00 11 02 02 02 00 00 00 00
0000024 00 04 00 00 04 03 00 00 00 00 00 01 00 00 00 00 21 03 03 03 00 00 00 00
0000048 00 04 00 00 04 03 00 00 00 00 00 01 00 00 00 00 12 04 01 01 00 00 00 00
0000072 00 04 00 00 04 03 00 00 00 00 00 01 00 00 00 00 10 00 04 04 00 00 00 00
0000096 00 04 00 00 04 03 00 00 00 00 00 01 00 00 00 00 20 01 05 05 00 00 00 00
0000120

ErmaC · November 4, 2016

@Slice

All the collected info are in this source... anyway...far from get it working

- ati.c

- ati.h

-> REMOVED update and info here

ErmaC

Mork vom Ork · November 5, 2016

I know zero developers trying to understand a problem with some AMD CPU and some chipsets for AMD CPU.

I have no reports and logs about this.

Clover has detection for AMD CPU and uses MSR that is known to work with some AMD CPU. That's all I know about.

Slice,

i think, i might know where the problem is - see my Bootlog here:

5:426  0:000  Framebuffer @0xB0000000  MMIO @0xC0000000    I/O Port @0x0000E000 ROM Addr @0xDE740000
5:426  0:000  ATI card POSTed,
5:426  0:000  Set VRAM for Cedar+ = -1Mb        //<--- the value for VRAM readout is NOT correct
5:426  0:000  ATI: get_vram_size returned 0xFFF00000 //<---- but here the Value is correct, it is 4.096 dec.
5:426  0:000  ATI Radeon EVERGREEN family
5:426  0:000  Users config name Acre
5:426  0:000  Use # of ports setting from config.plist: 3
5:426  0:000  (AtiPorts) Nr of ports set to: 3
5:426  0:000  ATI Oland AMD Radeon RX460 -1MB (Acre) [1002:67EF] (subsys [174B:E344]):: PciRoot(0x0)\Pci(0x1,0x0)\Pci(0x0,0x0)
5:426  0:000  Intel Intel HD Graphics 530 [8086:1912] :: PciRoot(0x0)\Pci(0x2,0x0)
5:426  0:000  Intel GFX revision  =0x6
5:426  0:000   RCBA disabled; cannot use it
5:426  0:000  stringlength = 4500
5:426  0:000  CurrentMode: Width=3840 Height=2160
5:426  0:000  Beginning FSInjection
5:426  0:000  Blocking kext caches

might have something to do with this code-snippet within "ATI.c" file here:

//  size in MB on evergreen
//  XXX watch for overflow!!!

    card->vram_size = ((UINT64)REG32(card->mmio, R600_CONFIG_MEMSIZE)) << 20;  //<--- this is the part to readout cards VRAM-size, if no Value for VRAM was set within config.plist
    DBG("Set VRAM for Cedar+ = %dMb\n", (INTN)RShiftU64(card->vram_size, 20));
    } else if (chip_family >= CHIP_FAMILY_R600) {
            card->vram_size = REG32(card->mmio, R600_CONFIG_MEMSIZE);
    } else {
      card->vram_size = REG32(card->mmio, RADEON_CONFIG_MEMSIZE);
      if (card->vram_size == 0) {
        card->vram_size = REG32(card->mmio, RADEON_CONFIG_APER_SIZE);

//  Slice - previously I successfully made Radeon9000 working
//  by writing this register
    
          WRITEREG32(card->mmio, RADEON_CONFIG_MEMSIZE, (UINT32)card->vram_size);
      }
    }
  }
  gSettings.VRAM = card->vram_size;
  DBG("ATI: get_vram_size returned 0x%x\n", card->vram_size);  //<--- and this is the part for correct VRAM size in Bootlog
}

tested it just with an installed RADEON RX460.

i might could test this part also with an installed R9 280x, to see, if VRAM size gets read out correctly with such an installed card <--- if this would be helpfull.

If card->vram_size really gets just "-1MB" that might be the problem, why system does not get loaded after verbose boot, cause OS X > 10.9 needs at least more memory than just a minimum of 128MB VRAM.

But please get noticed, that this is just some thoughts, cause i am NO coder - but i can read and compare ;-) You might give it a try... your choice.

EDIT#1:

This is Bootlog with R9 280x installed as main GFX:

7:230  0:000  Framebuffer @0xB0000000  MMIO @0xDE900000    I/O Port @0x0000E000 ROM Addr @0xDE940000
7:230  0:000  ATI card POSTed,
7:230  0:000  Set VRAM for Cedar+ = 3072Mb  //<--- see here, this time VRAM size was correct detected
7:230  0:000  ATI: get_vram_size returned 0xC0000000  //<--- also here correct VRAM size value
7:230  0:000  ATI Radeon EVERGREEN family
7:230  0:000  Users config name Junsai
7:230  0:000  Use # of ports setting from config.plist: 5
7:230  0:000  (AtiPorts) Nr of ports set to: 5
7:230  0:000  ATI Tahiti AMD Radeon HD 7970X/8970/R9 280X 3072MB (Junsai) [1002:6798] (subsys [174B:3000]):: PciRoot(0x0)\Pci(0x1,0x0)\Pci(0x0,0x0)
7:230  0:000   RCBA disabled; cannot use it

so i think, the code has to be modified in some way to get the correct VRAM size from newer cards (like cards with TONGA chipset or newer).

EDIT#2:

this is Bootlog output when using

		<key>VRAM</key>
		<integer>4096</integer>

within config.plist to determine VRAM size:

12:046  0:000  Framebuffer @0xB0000000  MMIO @0xC0000000    I/O Port @0x0000E000 ROM Addr @0xDE740000
12:046  0:000  ATI card POSTed,
12:046  0:000  Set VRAM from config = 4096Mb  // <--- correct, cause provided by config.plist
12:046  0:000  ATI: get_vram_size returned 0x0  // <--- incorrect, cause code will be ignored (through if...then...else), if VRAM size was set by config.plist
12:046  0:000  ATI Radeon EVERGREEN family
12:046  0:000  Users config name Acre
12:046  0:000  Use # of ports setting from config.plist: 3
12:046  0:000  (AtiPorts) Nr of ports set to: 3
12:046  0:000  ATI Oland AMD Radeon RX460 4096MB (Acre) [1002:67EF] (subsys [174B:E344]):: PciRoot(0x0)\Pci(0x1,0x0)\Pci(0x0,0x0)
12:046  0:000  Intel Intel HD Graphics 530 [8086:1912] :: PciRoot(0x0)\Pci(0x2,0x0)
12:046  0:000  Intel GFX revision  =0x6

Attached you will find the "ati.c" and "ati.h" files, which i i have expanded to contain BAFFIN and ELLESMERE code snippets:

expanded "ati.h" with:

    /* Arctic Islands */    
    CHIP_FAMILY_ELLESMERE, /* Polaris 10 */
    CHIP_FAMILY_BAFFIN,   /* Polaris 11 */
    CHIP_FAMILY_LAST

and

    /* AMD9500Controller */
    kBarbice,
    kElqui,
    kAcre,
    kCaroni,
    kFlorin,
    kGuariba,
    kDayman,
    kCfgEnd

and "ati.c" with:

    /* AMD9500Controller */
    {"Exmoor",    6},
    {"Barbice",   5},
    {"Elqui",     5},
    {"Acre",      3},
    {"Caroni",    5},
    {"Florin",    6},
    {"Guariba",   6},
    {"Dayman",    6},
    {"Baladi",    6},
    {"OPM",       5},

and (in parts)

//  ELLESMERE Polaris 10

    { 0x67C0,    0x00000000, CHIP_FAMILY_ELLESMERE,    "AMD Radeon Polaris 10",    kNull    },
    { 0x67DF,    0x00000000, CHIP_FAMILY_ELLESMERE,    "AMD Radeon RX480",         kNull    },

//  BAFFIN Polaris 11

    { 0x67E0,    0x00000000, CHIP_FAMILY_BAFFIN,    "AMD Radeon Polaris 11",        kNull    },
    { 0x67E1,    0x00000000, CHIP_FAMILY_BAFFIN,    "AMD Radeon Polaris 11",        kNull    },
    { 0x67E8,    0x00000000, CHIP_FAMILY_BAFFIN,    "AMD Radeon Polaris 11",        kNull    },
    { 0x67E9,    0x00000000, CHIP_FAMILY_BAFFIN,    "AMD Radeon Polaris 11",        kNull    },
    { 0x67EB,    0x00000000, CHIP_FAMILY_BAFFIN,    "AMD Radeon Polaris 11",        kNull    },
    { 0x67EF,    0x00000000, CHIP_FAMILY_BAFFIN,    "AMD Radeon RX460",             kAcre    },
    { 0x67FF,    0x00000000, CHIP_FAMILY_BAFFIN,    "AMD Radeon Polaris 11",        kNull    },

also i "cleaned" it a little bit to have a better overview:

Bildschirmfoto2016-11-05um18.01.17.png

Bildschirmfoto2016-11-05um18.03.31.png

Bildschirmfoto2016-11-05um18.03.54.png

and just to be sure: i did NOT change any existing part* or removed any outcommented parts.

* except this line:

    { 0x67EF,    0x00000000, CHIP_FAMILY_BAFFIN,    "AMD Radeon RX460",             kAcre    },  //changed "kNULL" to "kAcre" as standard FB for RX460

ati_files_clover.zip

Mork vom Ork · November 6, 2016

About VRAM size problem above <--- i fixed it by changing one line:

//  size in MB on evergreen
//  XXX watch for overflow!!!


//  card->vram_size = ((UINT64)REG32(card->mmio, R600_CONFIG_MEMSIZE)) << 20;   // <--- changed THIS line
    card->vram_size = IoRead32(R600_CONFIG_MEMSIZE) << 20;                      // <--- into THIS line
    DBG("Set VRAM for Cedar+ = %dMb\n", (INTN)RShiftU64(card->vram_size, 20));
    } else if (chip_family <= CHIP_FAMILY_R600) {
            card->vram_size = REG32(card->mmio, R600_CONFIG_MEMSIZE);
    } else {
      card->vram_size = REG32(card->mmio, RADEON_CONFIG_MEMSIZE);
      if (card->vram_size == 0) {
        card->vram_size = REG32(card->mmio, RADEON_CONFIG_APER_SIZE);

//  Slice - previously I successfully made Radeon9000 working
//  by writing this register

and now my Bootlog looks like this:

2:831  0:000  Framebuffer @0xB0000000  MMIO @0xC0000000    I/O Port @0x0000E000 ROM Addr @0xDE740000
2:831  0:000  ATI card POSTed,
2:831  0:000  Set VRAM for Cedar+ = 4095Mb             // <--- correct detected VRAM size here
2:831  0:000  ATI: get_vram_size returned 0xFFF00000   // <--- and here
2:831  0:000  Found generic VBIOS ROM file (1002_67EF.rom)
2:833  0:001  Loaded ROM len=262144
2:833  0:000  Calculated ROM len=58368
2:833  0:000  VideoBIOS read from file
2:833  0:000  ATI Radeon EVERGREEN family
2:833  0:000  Users config name Acre
2:833  0:000  Use # of ports setting from config.plist: 3
2:833  0:000  (AtiPorts) Nr of ports set to: 3
2:833  0:000  ATI Oland AMD Radeon RX460 4095MB (Acre) [1002:67EF] (subsys [174B:E344]):: PciRoot(0x0)\Pci(0x1,0x0)\Pci(0x0,0x0)

I know, it may NOT be the best way to fix it, but ONE way, and for me it works. Maybe someone of the PROs find a better way. Anyway... this change just fixed the VRAM size problem,

NOT the blacksreen after verbose boot with RADEON RX460.

Sorry, was hoping, too , that it may also fix the "blackscreen" problem.

Pavo · November 6, 2016

Slice I have a real Mac Pro 5,1 and able to do any test you may need to fix this gpu issue that everyone seems to be having, I have 7970, 290X and RX 480 to test if needed. Just let me know what you need

ErmaC · November 6, 2016

About VRAM size problem above

Sorry, was hoping, too , that it may also fix the "blackscreen" problem.

Hi Mork von Ork

just to be sure that I understand what you are try to do... (I try to help)

(PS I don't have this hardware... I mean ATi/AMD to test myself...)

So...

I use the latest Clover 3905 sources with the ATI changes that I post few "post" ago with you tiny change of today...

... with respect of your change made yesterday there is some duplicated in the card_configs...

can you test this? (pls the logs are very useful)

-> ~~removed~~... new one in the next post

also... one things... are you using a FAKE id? (maybe I lost where you mention it...)

but looking at you output I see this

2:833  0:000  ATI Oland AMD Radeon RX460 4095MB (Acre) [1002:67EF] (subsys [174B:E344]):: PciRoot(0x0)\Pci(0x1,0x0)\Pci(0x0,0x0)

Oland?????

the oland card we have in the source are this:

 
  // OLAND
  
  // Oland: R7-240, 250  - Southand Island
  { 0x6600, 0x00000000, CHIP_FAMILY_OLAND, "AMD Radeon HD 8600/8700M", kNull       }, // Mobile
  { 0x6601, 0x00000000, CHIP_FAMILY_OLAND, "AMD Radeon HD 8500/8700M", kNull       }, // Mobile
  // { 0x6602, 0x00000000, CHIP_FAMILY_OLAND, "AMD Radeon",           kNull       }, // Mobile
  // { 0x6603, 0x00000000, CHIP_FAMILY_OLAND, "AMD Radeon",           kNull       }, // Mobile
  { 0x6604, 0x00000000, CHIP_FAMILY_OLAND, "AMD Radeon R7 M265",       kNull       }, // Mobile
  { 0x6605, 0x00000000, CHIP_FAMILY_OLAND, "AMD Radeon R7 M260",       kNull       }, // Mobile
  { 0x6606, 0x00000000, CHIP_FAMILY_OLAND, "AMD Radeon HD 8790M",      kNull       }, // Mobile
  { 0x6607, 0x00000000, CHIP_FAMILY_OLAND, "AMD Radeon R5 M240",       kNull       }, // Mobile
  { 0x6608, 0x00000000, CHIP_FAMILY_OLAND, "AMD FirePro W2100",        kNull       },
  { 0x6610, 0x00000000, CHIP_FAMILY_OLAND, "AMD Radeon R7 250",        kFutomaki },
  { 0x6611, 0x00000000, CHIP_FAMILY_OLAND, "AMD Radeon R7 340 Series", kNull       },
  { 0x6613, 0x00000000, CHIP_FAMILY_OLAND, "AMD Radeon R7 240",        kFutomaki },
  // { 0x6620, 0x00000000, CHIP_FAMILY_OLAND, "AMD Radeon",           kNull       }, // Mobile
  // { 0x6621, 0x00000000, CHIP_FAMILY_OLAND, "AMD Radeon",           kNull       }, // Mobile
  // { 0x6623, 0x00000000, CHIP_FAMILY_OLAND, "AMD Radeon",           kNull       }, // Mobile
  // { 0x6631, 0x00000000, CHIP_FAMILY_OLAND, "AMD Radeon",           kNull       },

ErmaC

Mork vom Ork · November 6, 2016

@ErmaC

same effect: no initialization of RX460 w/o helper card (IGPU in my case). Still boots to BlackScreen after verbose boot.

but now ELLESMERE gets detected correctly in BootLog:

4:806  0:000  Framebuffer @0xB0000000  MMIO @0xC0000000    I/O Port @0x0000E000 ROM Addr @0xDE740000
4:806  0:000  ATI card POSTed,
4:806  0:000  Set VRAM for Cedar+ =4095Mb
4:806  0:000  ATI: get_vram_size returned 0xFFF00000
4:806  0:000  ATI Radeon EVERGREEN family
4:806  0:000  Users config name Acre
4:806  0:000   use N ports setting from config.plist: 3
4:806  0:000  (AtiPorts) Nr of ports set to: 3
4:806  0:000  ATI Ellesmere AMD Radeon RX460 4095MB (Acre) [1002:67EF] (subsys [174B:E344]):: PciRoot(0x0)\Pci(0x1,0x0)\Pci(0x0,0x0)

Have to check one of my files: ati.c or ati.h for that OLAND output.

No FakeID here - i completly let CLOVER detect device-id, vendor-id and subsystem-id:

Bildschirmfoto2016-11-06um19.02.55.png

see "Arbitrary" is outcommented, so that old "inject"-Method should work.

ErmaC · November 6, 2016

4:806  0:000  Users config name Acre
4:806  0:000   use N ports setting from config.plist: 3
4:806  0:000  (AtiPorts) Nr of ports set to: 3
4:806  0:000  ATI Ellesmere AMD Radeon RX460 4095MB (Acre) [1002:67EF] (subsys [174B:E344]):: PciRoot(0x0)\Pci(0x1,0x0)\Pci(0x0,0x0)

pls try also removing the port number in config.plist

because ithe "logic" automatically set the number of port to 3 (acre =3)

 { 0x67EF, 0x00000000, CHIP_FAMILY_BAFFIN, "AMD Radeon RX460",             kAcre },

your device have already set the correct number of port (so no need to set it in the config)

{"Acre", 3},

but now ELLESMERE gets detected correctly in BootLog:

I made some error...

yep ELLESMERE is now detect but your card is a BAFFIN...

I will check the list probably I forgot some CHIP_FAMILY...

EDIT:

YES I forgot to add a family chip...

this one is correct. (I mean the chip family for your card)

-> ~~removed~~ new one posted

ErmaC

gengik84 · November 6, 2016

@Ermac

Bios setting: primary PCIe

With FakeID

Black screen again... :wallbash:

Test RX-480

ErmaC · November 6, 2016

@Ermac

Bios setting: primary PCIe

With FakeID

Black screen again...

Test RX-480

@gengik84

I only try to fix the memory size and chip family (+ port number)

set the primary GFX is another story for now...

In the same way as Mork vom Ork set your gfx card in a way where you are able to boot and post the logs

ErmaC

Mork vom Ork · November 6, 2016

@ErmaC:

removed the entry for "VideoPorts" in config.plist. <--- but CLOVER still defaults to 6 ports for the RX460 (which only have 3 ports: DP, HDMI and DVI).

All other values have been detected as expected so far, even "CHIP_FAMILY_BAFFIN" :thumbsup_anim:

About the Portdetection ithink we have to look at this part of the code:

//  use cfg_name on radeon_cards, to retrive the default name from card_configs,
        card->cfg_name = card_configs[card->info->cfg_name].name;
        n_ports = card_configs[card->info->cfg_name].ports;        // <--- # of ports declared here (declared through  detection?)
        
//  which means one of the fb's or kNull
        DBG("Framebuffer set to device's default: %a\n", card->cfg_name);
    DBG(" N ports defaults to %d\n", n_ports);
  }
 
  if (gSettings.VideoPorts != 0) {                                  // <--- part, when # of Ports provided by config.plist
    n_ports = gSettings.VideoPorts;
    DBG("Use # of ports setting from config.plist: %d\n", n_ports);
  }
 
    if (n_ports > 0) {                                              // <--- part, when # of Ports NOT provided by config.plist, then try to detect through CLOVER
        card->ports = (UINT8)n_ports; // use it.
        DBG("(AtiPorts) Nr of ports set to: %d\n", card->ports);
  } else {
    
//      if (card->cfg_name > 0) // do we want 0 ports if fb is kNull or mistyped ?
//      else, match cfg_name with card_configs list and retrive default nr of ports.
        for (i = 0; i < kCfgEnd; i++) {
            if (AsciiStrCmp(card->cfg_name, card_configs[i].name) == 0) {
                card->ports = card_configs[i].ports; // default
      }
    }
        
        DBG("Nr of ports set to framebuffer's default: %d\n", card->ports);
    }
 
  if (card->ports == 0) {
    card->ports = 2; //real minimum
    DBG("Nr of ports set to min: %d\n", card->ports);
  }

fantomas · November 6, 2016

but CLOVER still defaults to 6 ports for the RX460 (which only have 3 ports: DP, HDMI and DVI).

remove also "FB Name" entry if it is not empty.

ErmaC · November 6, 2016

removed the entry for "VideoPorts" in config.plist.

All other values have been detected as expected so far, even "CHIP_FAMILY_BAFFIN"

remove also "FB Name" entry if it is not empty.

of course this also is important no 'missed' value have to interfere...

Can you post the log with this?

-> CLOVER_3905_ATI_V4.zip

setting in config and w/o setting in config..

pls

ErmaC

Mork vom Ork · November 6, 2016

next problem here:

even i have set ACRE to be RX460's default Framebuffer

    { 0x67EF,    0x00000000, CHIP_FAMILY_BAFFIN,    "AMD Radeon RX460",             kAcre    },

CLOVER allways sets LABRADOR as default framebuffer if i don't provide a Framebuffer via Config.plist. That's why my default # of Ports allways get set to "6" instead of "3" ("6" is default value for LABRADOR)

@ErmaC

testing now your files with values provided by config.plist and w/o providing values via config.plist. I'll be back in couple of minutes to report...

EDIT#1:

WOW, i am impressed, all values correct w/o providing them via config.plist (CLOVER detected them all correct):

20:107  0:000  Framebuffer @0xB0000000  MMIO @0xC0000000    I/O Port @0x0000E000 ROM Addr @0xDE740000
20:107  0:000  ATI card POSTed,
20:107  0:000  Set VRAM for Baffin =4095Mb
20:107  0:000  ATI: get_vram_size returned 0xFFF00000
20:107  0:000  ATI Radeon EVERGREEN family
20:107  0:000  Framebuffer set to device's default: Acre
20:107  0:000   N ports defaults to 3
20:107  0:000  (AtiPorts) Nr of ports set to: 3
20:107  0:000  ATI Baffin AMD Radeon RX460 4095MB (Acre) [1002:67EF] (subsys [174B:E344]):: PciRoot(0x0)\Pci(0x1,0x0)\Pci(0x0,0x0)

CONGRATULATIONS!

Mork vom Ork · November 6, 2016

Got good news and got bad news!

Good News: we don't need to provide Framebuffername and # of VideoPorts anymore through config.plist for RX460 now, cause CLOVER detects them itself.

Bad News: still need Helpergpu like IGPU to get RX460 working. W/o HelperGPU i still get BlackScreen after verbose boot. But i think, we might get closer to solve this finally.

@ErmaC:

would you please provide me with your "ati.c" and "ati.h" files - so i can update my versions of the file? Would make me happy...

ErmaC · November 6, 2016

@ErmaC:

would you please provide me with your "ati.c" and "ati.h" files - so i can update my versions of the file? Would make me happy...

Hi...

No problem I will post it in the patch section of clover....

just let me write some info about what is what.... so then the changes will be avaiable for all..

ErmaC

EDIT: done

Mork vom Ork · November 7, 2016

With all the above applied patches and some cosmetical corrections, my bootlog for Radeon RX460 now looks like this:

4:883  0:000  Framebuffer @0xB0000000 • MMIO @0xC0000000 • I/O Port @0x0000E000 • ROM Addr @0xDE740000
4:883  0:000  ATI card POSTed
4:883  0:000  Set VRAM for Cedar+ to 4095Mb
4:883  0:000  ATI get_vram_size returned: 0xFFF00000
4:883  0:000  ATI Radeon EVERGREEN family
4:883  0:000  Devices default Framebuffer: Acre
4:883  0:000  Default Nr. of ports: 3
4:883  0:000  (AtiPorts) Nr. of ports: 3
4:883  0:000  ATI CHIP FAMILY: Baffin • CARD NAME: AMD Radeon RX460 • VRAM: 4095MB • FRAMEBUFFER: Acre • VENDOR-/DEVICE-ID: 1002:67EF • SUBSYSTEM-ID: 174B:E344 • PciRoot(0x0)\Pci(0x1,0x0)\Pci(0x0,0x0)

by using these settings within my config.plist:

Bildschirmfoto2016-11-07um01.20.50.png

But remember: still have to use IGPU to get RX460 working.

vusun123 · November 7, 2016

But remember: still have to use IGPU to get RX460 working.

Was it a Clover bug or the entire lineup of the new AMD cards are all like that ? I'm planning to get a RX470 soon, and if this problem can't be fixed in the near future i might refrain from getting it.

Also can Clover identify Haswell-E/Broadwell-E CPUs properly ? I have a setup at work running Sierra with Xeon E5-2683v3 and System Report shows it as Unknown ( even though many applications still see the name correctly ), which preventing me from installing Intel Power Gadget to check for speedstep. Tried FakeCPUID and manually injecting CPU type, doesn't work I'm an idiot for not researching enough, PikeRAlpha has a kernel patch to resolve this matter

Clover test and patches for Polaris GPU

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