Bug 98523

Summary: Can't run without "nomodeset" on MacPro6,1 with two AMD R9 280X Tahiti video cards
Product: DRI Reporter: Heigh <Heigh.Leigh171>
Component: DRM/RadeonAssignee: Default DRI bug account <dri-devel>
Status: RESOLVED NOTOURBUG QA Contact:
Severity: major    
Priority: medium    
Version: unspecified   
Hardware: Other   
OS: Linux (All)   
Whiteboard:
i915 platform: i915 features:
Attachments:
Description Flags
Journals from Ubuntu and KaOS and hardware info none

Description Heigh 2016-11-01 03:31:28 UTC 
Created attachment 127655 [details]
Journals from Ubuntu and KaOS and hardware info

Dear Radeon driver developers:

I have a rare configuration of 2016 MacPro6,1 with two AMD Radeon R9 280X Tahiti video cards, and three 4K monitors connected via DisplayPort / Thunderbolt ports.

The system installs successfully, but both LiveCD (for installation) and installed system require "nomodeset" option, otherwise boot process hangs with a black screen.

With "nomodeset" system detects and works only on one monitor, connected via HDMI. All monitors connected via DisplayPorts remain blank.

I tried different versions of Ubuntu (14.04.4, 14.04.5, 15.10, 16.04, 16.04.1, 16.10), all have this problem, and also several versions of Linux Mint, with same result. I also tried newer upstream kernels (4.6, 4.7.3, 4.8) on Ubuntu 16.04.1, all with the same result. I tried to build recent radeon driver from git, with no success.

But I found one distro of Linux, called KaOS (https://kaosx.us/) that works OK on my hardware. Soon after butting, it successfully detects and turns On all three monitors connected via Display Ports, and then presents normal wide KDE desktop on all three monitors.

I compared journal entries from Ubuntu and KaOS boots, and it seems that
KaOS successfully detects and activates both AMD cards, and then continues to boot normally. I have checked, KaOS uses radeon driver: 
lspci -nnk | grep -i vga -A3 | grep 'in use' gives: Kernel driver in use: radeon

Ubuntu, on the other hand (all versions and kernels listed above), with the same radeon driver detects only the first card, and fails on the second card, which causes immediate shutdown of radeon driver, and boot process hangs after that.

Here's the relevant messages from the journals. Full journal logs are available in the attachments. In both cases I start kernel with options "debug ignore_loglevel drm.debug=1", and only till runlevel 1 to avoid excessive not relevant output. Difference in behaviour of radeon driver in KaOS and Ubuntu is visible very early on boot process.

KaOS:
=====

First card:
-------------
    [drm] radeon kernel modesetting enabled.
    [drm] initializing kernel modesetting (TAHITI 0x1002:0x6798 0x106B:0x0128 0x00).
    [drm] register mmio base: 0xA0700000
    [drm] register mmio size: 262144
    [drm] ACPI VFCT contains a BIOS for 02:00.0 1002:6798, size 65536
    [drm:radeon_get_bios] ATOMBIOS detected
    ATOM BIOS: Tahiti
    [drm] Loading tahiti Microcode
    [drm] Initialized radeon 2.45.0 20080528 for 0000:02:00.0 on minor 0
    fb: switching to radeondrmfb from EFI VGA
    .....

Second card:
----------------
    [drm] ACPI VFCT contains a BIOS for 02:00.0 1002:6798, size 65536
    [drm] ACPI VFCT table is not for this card
    radeon 0000:06:00.0: Invalid PCI ROM header signature: expecting 0xaa55, got 0x0000
    radeon 0000:06:00.0: Invalid PCI ROM header signature: expecting 0xaa55, got 0x0000
    [drm:radeon_get_bios] ATOMBIOS detected
    ATOM BIOS: Tahiti
    [drm] Loading tahiti Microcode
    [drm] radeon: dpm initialized
    [drm] Initialized radeon 2.45.0 20080528 for 0000:06:00.0 on minor 1


(Notice these "Invalid PCI ROM header signature" errors,
but the card initialized successfully despite them,
and KaOS developers confirmed, that these errors are normal for ATI cards,
and are harmless)


Ubuntu:
=======

First card:
-------------
    [drm] radeon kernel modesetting enabled.
    [drm] initializing kernel modesetting (TAHITI 0x1002:0x6798 0x106B:0x0128).
    [drm] register mmio base: 0xA0700000
    [drm] register mmio size: 262144
    [drm] ACPI VFCT contains a BIOS for 02:00.0 1002:6798, size 65536
    [drm:radeon_get_bios] ATOMBIOS detected
    ATOM BIOS: Tahiti
    [drm] Loading tahiti Microcode
    [drm] radeon: dpm initialized
    [drm] Initialized radeon 2.43.0 20080528 for 0000:02:00.0 on minor 0
    fb: switching to radeondrmfb from EFI VGA

Second card:
-----------------
    [drm] initializing kernel modesetting (TAHITI 0x1002:0x6798 0x106B:0x0127).
    [drm] register mmio base: 0xA0600000
    [drm] register mmio size: 262144
    [drm] ACPI VFCT contains a BIOS for 02:00.0 1002:6798, size 65536
    [drm] ACPI VFCT table is not for this card
    radeon 0000:06:00.0: Invalid ROM contents
    radeon 0000:06:00.0: Invalid ROM contents
    [drm:radeon_get_bios [radeon]] *ERROR* Unable to locate a BIOS ROM
    radeon 0000:06:00.0: Fatal error during GPU init
    [drm] radeon: finishing device.
    radeon: probe of 0000:06:00.0 failed with error -22

Notice same errors "Invalid ROM contents" (slightly different wording,
but I have checked in sources that they mean the same - unexpected ROM signature) After these errors radeon/drm driver fails with *ERROR* Unable to locate a BIOS ROM, and then "finishing device."

KaOS uses newer kernel 4.7.3-1, but I tried 4.6, 4.7.3, 4.8 on Ubuntu, without success.

Another thing I noticed from journals: on KaOS udev rules are loaded before kernel modesetting, but on Ubuntu - after modesetting. Can this be a reason?


Graphics cards info:
---------------------------

    lspci | grep VGA
    02:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Tahiti XT [Radeon HD 7970/8970 OEM / R9 280X]
    06:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Tahiti XT [Radeon HD 7970/8970 OEM / R9 280X]

Full "lspci -vvv" and "lshw" outputs are also attached for completeness.

I really need to make this workstation work under Linux, don't want
to settle for MacOS.

If somebody competent could have a look at this and give some suggestions,
I'd appreciate it a lot. 

Some workaround/hack to make it work somehow until the problem is fixed, would be great as well.

Thanks,
Heigh
Comment 1 Heigh 2016-11-08 11:31:09 UTC 
Hi Michel,

Thank you for looking into this, and for assigning the bug.

In the meantime, I have found a workaround or a solution.

The system needs to be booted with systemd-boot bootloader
(or perhaps any other supporting UEFI boot)

When I boot Ubuntu with standard, installed during installation
grub2, radeon driver would't detect external displays connected via
Display Ports.

But when I boot system using systemd-boot installed by Arch Linux (or KaOS),
it detects displays normally, and works correctly, supporting
all external displays as expected.

I think this is related to some aspects of the hardware not
exposed in BIOS boot mode, and exposed in EFI/UEFI boot mode.

Preferred solution would be working radeon driver in BIOS mode
if possible, or, if that is not possible, some meaningful message 
in the system journal explaining that certain features 
(e.g. external monitors via Display Port) will not be supported 
unless the system is booted in EFI/UEFI mode.

Thank you,
Heigh
Comment 2 Alex Deucher 2016-11-08 13:59:26 UTC 
The problem is Macs expose the vbios for some cards in a proprietary way.  IIRC, the vbios is only available via EFI prior to the OS loading.  The bootloader needs to snag the vbios and then make it available to the OS when it loads.  I'm not really a Mac expert so I don't remember all the details.  There's no way to work around it in the driver.

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.