Radeon RX550 4GB stuck during kernel boot

[bjdooks-sifive]

The 2022.10 release works fine with a R7 4GB but my RX550 4GB is failing during boot.

[   27.788035] snd_hda_intel 0000:07:00.1: enabling device (0000 -> 0002)       
[   27.788085] snd_hda_intel 0000:07:00.1: Force to snoop mode by module option 
[   28.656527] input: HDA ATI HDMI HDMI/DP,pcm=3 as /devices/platform/soc/e00004
[   28.656927] input: HDA ATI HDMI HDMI/DP,pcm=7 as /devices/platform/soc/e00005
[   28.657418] input: HDA ATI HDMI HDMI/DP,pcm=8 as /devices/platform/soc/e00006
[   28.657896] input: HDA ATI HDMI HDMI/DP,pcm=9 as /devices/platform/soc/e00007
[   28.658522] input: HDA ATI HDMI HDMI/DP,pcm=10 as /devices/platform/soc/e0008
[   63.486141] rcu: INFO: rcu_sched self-detected stall on CPU                  
[   63.486165] rcu:     2-....: (5249 ticks this GP) idle=915/1/0x4000000000000 
[   63.486184]  (t=5250 jiffies g=2341 q=4293)                                  

the rest of the probe seems to work but there is no display

[   87.463962] [drm] amdgpu kernel modesetting enabled.                         
[   87.464313] amdgpu 0000:07:00.0: vgaarb: deactivate vga console              
[   87.464943] amdgpu 0000:07:00.0: enabling device (0000 -> 0003)              
[   87.464968] [drm] initializing kernel modesetting (POLARIS12 0x1002:0x699F 0.
[   87.464984] amdgpu 0000:07:00.0: amdgpu: Trusted Memory Zone (TMZ) feature nd
[   87.465005] [drm] register mmio base: 0x60300000                             
[   87.465011] [drm] register mmio size: 262144                                 
[   87.467187] [drm] add ip block number 0 <vi_common>                          
[   87.467206] [drm] add ip block number 1 <gmc_v8_0>                           
[   87.467212] [drm] add ip block number 2 <tonga_ih>                           
[   87.467217] [drm] add ip block number 3 <gfx_v8_0>                           
[   87.467223] [drm] add ip block number 4 <sdma_v3_0>                          
[   87.467228] [drm] add ip block number 5 <powerplay>                          
[   87467235] [drm] add ip block number 6 <dm>                                 
[   87.467246] [drm] add ip block number 8 <vce_v3_0>                           
[   87.678280] amdgpu 0000:07:00.0: amdgpu: Fetched VBIOS from ROM BAR          
[   87.678309] amdgpu: ATOM BIOS: 113-3672S-US1                                 
[   87.678478] [drm] UVD is enabled in VM mode                                  
[   87.678486] [drm] UVD ENC is enabled in VM mode                              
[   87.678495] [drm] VCE enabled in VM mode                                     
[   87.678509] amdgpu 0000:07:00.0: amdgpu: PCIE atomic ops is not supported    
[   87.678549] [drm] GPU posting now...                                         
[   87.802730] [drm] vm size is 64 GB, 2 levels, block size is 10-bit, fragmentt
[   87.945391] amdgpu 0000:07:00.0: BAR 2: releasing [mem 0x2010000000-0x20101f]
[   87.945417] amdgpu 0000:07:00.0: BAR 0: releasing [mem 0x2000000000-0x200fff]
[   87.945479] pcieport 0000:02:08.0: BAR 15: releasing [mem 0x2000000000-0x201]
[   87.945492] pcieport 0000:01:00.0: BAR 15: releasing [mem 0x2000000000-0x201]
[   87.945504] pcieport 0000:00:00.0: BAR 15: releasing [mem 0x2000000000-0x201]
[   87.945573] pcieport 0000:00:00.0: BAR 15: assigned [mem 0x2000000000-0x217f]
[   87.945587] pcieport 0000:01:00.0: BAR 15: assigned [mem 0x2000000000-0x217f]
[   87.945601] pcieport 0000:02:08.0: BAR 15: assigned [mem 0x2000000000-0x217f]
[   87.945845] amdgpu 0000:07:00.0: amdgpu: GART: 256M 0x000000FF00000000 - 0x0F
[   87.945864] [drm] Detected VRAM RAM=4096M, BAR=4096M                         
[   87.945869] [drm] RAM width 128bits GDDR5                                    
[   87.982760] [drm] amdgpu: 4096M of VRAM memory ready                         
[   87.982810] [drm] amdgpu: 4096M of GTT memory ready.                         
[   87.982927] [drm] GART: num cpu pages 65536, num gpu pages 65536             
[   87.983852] [drm] PCIE GART of 256M enabled (table at 0x000000F400000000).   
[   88.117286] [drm] Chained IB support enabled!                                
[   88.582029] amdgpu: hwmgr_sw_init smu backed is polaris10_smu                
[   88.827334] [drm] Found UVD firmware Version: 1.130 Family ID: 16            
[   88.941219] [drm] Found VCE firmware Version: 53.26 Binary ID: 3             
[   89.382375] [drm] Display Core initialized with v3.2.177!                    
[   89.384752] snd_hda_intel 0000:07:00.1: bound 0000:07:00.0 (ops amdgpu_dm_au)
[   89.455446] [drm] UVD and UVD ENC initialized successfully.                  
[   89.556367] [drm] VCE initialized successfully.                              
[   89.558558] amdgpu 0000:07:00.0: amdgpu: SE 2, SH per SE 1, CU per SH 5, act8
[   89.562723] amdgpu 0000:07:00.0: amdgpu: Using BACO for runtime pm           
[   89.564425] [drm] Initialized amdgpu 3.46.0 20150101 for 0000:07:00.0 on min0
[   89.737149] Console: switching to colour frame buffer device 240x67          
[   89.859370] amdgpu 0000:07:00.0: [drm] fb0: amdgpudrmfb frame buffer device  

[tprrt]

Hello @bjdooks-sifive,

Thank you for reporting this issue to us.

Moreover, we reproduced it randomly, by doing several reboots, on RX550 4GB.

In addition, I’m currently working on a kernel update for the next FUSDK release, which contains some fixes for Radeon.

Kind regards, Thomas

[bjdooks-sifive]

I'll try and test this on the latest release to see if it is fixed.