[BUG] 3.14-rc6 problems with an R7 260X

Mon Mar 10 08:47:03 PDT 2014

Hi,

I recently added a R7 260X to my system.  While the card works with 3.13 its supposed work much better with 14-rc.
This is not the case.  My system is unstable without radeon.dpm=0 which was the default in .13.  Here are some extracts
from the logs of the latest fun with dpm enabled:

with 3.14-rc6 (with an up to date arch, stable X and mesa-git (10.2) mesa 10.1 and 10.0 also show very similar problems.

Mar 10 10:45:04 localhost kernel: [    0.000000] Command line: BOOT_IMAGE=/vmlinuz-linux-mainline root=/dev/sda3 ro console=tty0 verbose

Mar 10 10:45:04 localhost kernel: [    3.258402] [drm] radeon kernel modesetting enabled.
Mar 10 10:45:04 localhost kernel: [    3.258421] checking generic (e0000000 300000) vs hw (e0000000 10000000)
Mar 10 10:45:04 localhost kernel: [    3.258422] fb: conflicting fb hw usage radeondrmfb vs VESA VGA - removing generic driver
Mar 10 10:45:04 localhost kernel: [    3.270529] Console: switching to colour dummy device 80x25
Mar 10 10:45:04 localhost kernel: [    3.270757] [drm] initializing kernel modesetting (BONAIRE 0x1002:0x6658 0x174B:0xE253).
Mar 10 10:45:04 localhost kernel: [    3.270781] [drm] register mmio base: 0xF0800000
Mar 10 10:45:04 localhost kernel: [    3.270782] [drm] register mmio size: 262144
Mar 10 10:45:04 localhost kernel: [    3.270785] [drm] doorbell mmio base: 0xF0000000
Mar 10 10:45:04 localhost kernel: [    3.270786] [drm] doorbell mmio size: 8388608
Mar 10 10:45:04 localhost kernel: [    3.270837] ATOM BIOS: Bonaire
Mar 10 10:45:04 localhost kernel: [    3.270884] radeon 0000:01:00.0: VRAM: 2048M 0x0000000000000000 - 0x000000007FFFFFFF (2048M used)
Mar 10 10:45:04 localhost kernel: [    3.270884] radeon 0000:01:00.0: GTT: 1024M 0x0000000080000000 - 0x00000000BFFFFFFF
Mar 10 10:45:04 localhost kernel: [    3.270885] [drm] Detected VRAM RAM=2048M, BAR=256M
Mar 10 10:45:04 localhost kernel: [    3.270886] [drm] RAM width 128bits DDR
Mar 10 10:45:04 localhost kernel: [    3.270925] [TTM] Zone  kernel: Available graphics memory: 8145248 kiB
Mar 10 10:45:04 localhost kernel: [    3.270926] [TTM] Zone   dma32: Available graphics memory: 2097152 kiB
Mar 10 10:45:04 localhost kernel: [    3.270927] [TTM] Initializing pool allocator
Mar 10 10:45:04 localhost kernel: [    3.270929] [TTM] Initializing DMA pool allocator
Mar 10 10:45:04 localhost kernel: [    3.270940] [drm] radeon: 2048M of VRAM memory ready
Mar 10 10:45:04 localhost kernel: [    3.270940] [drm] radeon: 1024M of GTT memory ready.
Mar 10 10:45:04 localhost kernel: [    3.270948] [drm] Loading BONAIRE Microcode
Mar 10 10:45:04 localhost kernel: [    3.276841] nct6775: Found NCT6776D/F or compatible chip at 0x2e:0x290
Mar 10 10:45:04 localhost kernel: [    3.280263] [drm] Internal thermal controller with fan control
Mar 10 10:45:04 localhost kernel: [    3.280294] [drm] probing gen 2 caps for device 8086:c01 = 261ad03/e
Mar 10 10:45:04 localhost kernel: [    3.286710] [drm] radeon: dpm initialized
Mar 10 10:45:04 localhost kernel: [    3.288706] [drm] GART: num cpu pages 262144, num gpu pages 262144
Mar 10 10:45:04 localhost kernel: [    3.289112] [drm] probing gen 2 caps for device 8086:c01 = 261ad03/e
Mar 10 10:45:04 localhost kernel: [    3.289114] [drm] PCIE gen 3 link speeds already enabled
Mar 10 10:45:04 localhost kernel: [    3.299762] [drm] PCIE GART of 1024M enabled (table at 0x0000000000277000).

from xorg's log

[    55.320] (II) RADEON(0): Creating default Display subsection in Screen section
        "Default Screen Section" for depth/fbbpp 24/32
[    55.320] (==) RADEON(0): Depth 24, (--) framebuffer bpp 32
[    55.320] (II) RADEON(0): Pixel depth = 24 bits stored in 4 bytes (32 bpp pixmaps)
[    55.320] (==) RADEON(0): Default visual is TrueColor
[    55.320] (==) RADEON(0): RGB weight 888
[    55.320] (II) RADEON(0): Using 8 bits per RGB (8 bit DAC)
[    55.320] (--) RADEON(0): Chipset: "BONAIRE" (ChipID = 0x6658)
[    55.320] (II) Loading sub module "dri2"
[    55.320] (II) LoadModule: "dri2"
[    55.320] (II) Module "dri2" already built-in
[    55.320] (II) Loading sub module "glamoregl"
[    55.320] (II) LoadModule: "glamoregl"
[    55.320] (II) Loading /usr/lib/xorg/modules/libglamoregl.so
[    55.320] (II) Module glamoregl: vendor="X.Org Foundation"
[    55.320]    compiled for 1.15.0, module version = 0.6.0
[    55.320]    ABI class: X.Org ANSI C Emulation, version 0.4
[    55.320] (II) glamor: OpenGL accelerated X.org driver based.
[    55.832] (II) glamor: EGL version 1.4 (DRI2):
[    55.845] (II) RADEON(0): glamor detected, initialising EGL layer.
[    55.845] (II) RADEON(0): KMS Color Tiling: disabled
[    55.845] (II) RADEON(0): KMS Color Tiling 2D: disabled
[    55.845] (II) RADEON(0): KMS Pageflipping: enabled
[    55.845] (II) RADEON(0): SwapBuffers wait for vsync: enabled
[    55.856] (II) RADEON(0): Output DisplayPort-1 has no monitor section
[    55.858] (II) RADEON(0): Output HDMI-3 has no monitor section
[    55.861] (II) RADEON(0): Output DVI-0 has no monitor section
[    55.891] (II) RADEON(0): Output DVI-1 has no monitor section
[    55.903] (II) RADEON(0): EDID for output DisplayPort-1
[    55.905] (II) RADEON(0): EDID for output HDMI-3
[    55.907] (II) RADEON(0): EDID for output DVI-0
[    55.938] (II) RADEON(0): EDID for output DVI-1
[    55.938] (II) RADEON(0): Manufacturer: SAM  Model: 2b5  Serial#: 1213542964
[    55.938] (II) RADEON(0): Year: 2008  Week: 14
[    55.938] (II) RADEON(0): EDID Version: 1.3
[    55.938] (II) RADEON(0): Analog Display Input,  Input Voltage Level: 0.700/0.300 V
[    55.938] (II) RADEON(0): Sync:  Separate  Composite  SyncOnGreen
[    55.938] (II) RADEON(0): Max Image Size [cm]: horiz.: 52  vert.: 32
[    55.938] (II) RADEON(0): Gamma: 2.60
[    55.938] (II) RADEON(0): DPMS capabilities: Off; RGB/Color Display
[    55.938] (II) RADEON(0): First detailed timing is preferred mode
[    55.938] (II) RADEON(0): redX: 0.653 redY: 0.337   greenX: 0.295 greenY: 0.607
[    55.938] (II) RADEON(0): blueX: 0.144 blueY: 0.075   whiteX: 0.312 whiteY: 0.329
[    55.938] (II) RADEON(0): Supported established timings:Mar 10 10:48:28 localhost systemd[775]: Time has been changed

and from the system log when the fun starts (there is nothing new int the xorg log)

Mar 10 10:48:28 localhost systemd[1]: Time has been changed
Mar 10 10:48:52 localhost kernel: [  231.782359] radeon 0000:01:00.0: GPU fault detected: 146 0x0aa20804
Mar 10 10:48:52 localhost kernel: [  231.782361] radeon 0000:01:00.0:   VM_CONTEXT1_PROTECTION_FAULT_ADDR   0x00002F55
Mar 10 10:48:52 localhost kernel: [  231.782362] radeon 0000:01:00.0:   VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x02008004
Mar 10 10:48:52 localhost kernel: [  231.782363] VM fault (0x04, vmid 1) at page 12117, read from 'TC0' (0x54433000) (8)
Mar 10 10:48:52 localhost kernel: [  231.782366] radeon 0000:01:00.0: GPU fault detected: 146 0x0aa20404
Mar 10 10:48:52 localhost kernel: [  231.782367] radeon 0000:01:00.0:   VM_CONTEXT1_PROTECTION_FAULT_ADDR   0x00000000
Mar 10 10:48:52 localhost kernel: [  231.782367] radeon 0000:01:00.0:   VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x00000000
Mar 10 10:48:52 localhost kernel: [  231.782368] VM fault (0x00, vmid 0) at page 0, read from '' (0x00000000) (0)
Mar 10 10:48:52 localhost kernel: [  231.782374] radeon 0000:01:00.0: GPU fault detected: 146 0x0aa24404
Mar 10 10:48:52 localhost kernel: [  231.782374] radeon 0000:01:00.0:   VM_CONTEXT1_PROTECTION_FAULT_ADDR   0x00002F55
Mar 10 10:48:52 localhost kernel: [  231.782375] radeon 0000:01:00.0:   VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x02044004
Mar 10 10:48:52 localhost kernel: [  231.782376] VM fault (0x04, vmid 1) at page 12117, read from 'TC3' (0x54433300) (68)
Mar 10 10:48:52 localhost kernel: [  231.782378] radeon 0000:01:00.0: GPU fault detected: 146 0x0aa24804
Mar 10 10:48:52 localhost kernel: [  231.782379] radeon 0000:01:00.0:   VM_CONTEXT1_PROTECTION_FAULT_ADDR   0x00000000
Mar 10 10:48:52 localhost kernel: [  231.782379] radeon 0000:01:00.0:   VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x00000000
Mar 10 10:48:52 localhost kernel: [  231.782380] VM fault (0x00, vmid 0) at page 0, read from '' (0x00000000) (0)
Mar 10 10:48:52 localhost kernel: [  231.782382] radeon 0000:01:00.0: GPU fault detected: 146 0x0aa24804
Mar 10 10:48:52 localhost kernel: [  231.782383] radeon 0000:01:00.0:   VM_CONTEXT1_PROTECTION_FAULT_ADDR   0x00000000
Mar 10 10:48:52 localhost kernel: [  231.782383] radeon 0000:01:00.0:   VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x00000000
Mar 10 10:48:52 localhost kernel: [  231.782384] VM fault (0x00, vmid 0) at page 0, read from '' (0x00000000) (0)
Mar 10 10:48:52 localhost kernel: [  231.782386] radeon 0000:01:00.0: GPU fault detected: 146 0x0aa20804
Mar 10 10:48:52 localhost kernel: [  231.782387] radeon 0000:01:00.0:   VM_CONTEXT1_PROTECTION_FAULT_ADDR   0x00000000
Mar 10 10:48:52 localhost kernel: [  231.782388] radeon 0000:01:00.0:   VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x00000000
Mar 10 10:48:52 localhost kernel: [  231.782388] VM fault (0x00, vmid 0) at page 0, read from '' (0x00000000) (0)
Mar 10 10:48:52 localhost kernel: [  231.782391] radeon 0000:01:00.0: GPU fault detected: 146 0x0aa20404
Mar 10 10:48:52 localhost kernel: [  231.782391] radeon 0000:01:00.0:   VM_CONTEXT1_PROTECTION_FAULT_ADDR   0x00000000
Mar 10 10:48:52 localhost kernel: [  231.782392] radeon 0000:01:00.0:   VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x00000000
Mar 10 10:48:52 localhost kernel: [  231.782392] VM fault (0x00, vmid 0) at page 0, read from '' (0x00000000) (0)
Mar 10 10:48:52 localhost kernel: [  231.782398] radeon 0000:01:00.0: GPU fault detected: 146 0x0aa24404
Mar 10 10:48:52 localhost kernel: [  231.782399] radeon 0000:01:00.0:   VM_CONTEXT1_PROTECTION_FAULT_ADDR   0x00000000
Mar 10 10:48:52 localhost kernel: [  231.782399] radeon 0000:01:00.0:   VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x00000000

The above type of errors repeat hundreds of times and eventually the display freezes (this box does not have a serial console and it did not check with ssh)

When X started I did notice some corruption.  There are sets of two rectangles about of a height of 2 or 3 mm, width of 25m or so with a second
about a cm below.  The often occurs in chomium especially when scrolling.  Runing the  unigine-sanctuary or unigine-tropics demo/benchmark
programs also produce the above problems and eventually stall.

The problem is reproducible - I am not that familiar with gpu problems though, what else will help debug this?

TIA
Ed Tomlinson