[BUG] Crash when format enabled in topology but not firmware

[cujomalainey]

Describe the bug

[    5.932289] sof-audio-pci 0000:00:1f.3: error : DSP panic!                                                                                                                                  
[    5.932295] sof-audio-pci 0000:00:1f.3: status: fw entered - code 00000005                  
[    5.932393] sof-audio-pci 0000:00:1f.3: error: runtime exception                                                                                                                            
[    5.932395] sof-audio-pci 0000:00:1f.3: error: trace point 00004000                                                                                                                         
[    5.932397] sof-audio-pci 0000:00:1f.3: error: panic at :0                                                                                                                                  
[    5.932399] sof-audio-pci 0000:00:1f.3: error: DSP Firmware Oops                                                                                                                            
[    5.932401] sof-audio-pci 0000:00:1f.3: error: Exception Cause: IllegalInstructionCause, Illegal instruction          
[    5.932403] sof-audio-pci 0000:00:1f.3: EXCCAUSE 0x00000000 EXCVADDR 0x00000000 PS       0x00000000 SAR     0x00000000                                                                      
[    5.932405] sof-audio-pci 0000:00:1f.3: EPC1     0x00000000 EPC2     0x00000000 EPC3     0x00000000 EPC4    0x00000000
[    5.932406] sof-audio-pci 0000:00:1f.3: EPC5     0x00000000 EPC6     0x00000000 EPC7     0x00000000 DEPC    0x00000000                                                                      
[    5.932408] sof-audio-pci 0000:00:1f.3: EPS2     0x00000000 EPS3     0x00000000 EPS4     0x00000000 EPS5    0x00000000
[    5.932410] sof-audio-pci 0000:00:1f.3: EPS6     0x00000000 EPS7     0x00000000 INTENABL 0x00000000 INTERRU 0x00000000  
[    5.932412] sof-audio-pci 0000:00:1f.3: stack dump from 0x00000000                                                                                                                          
[    5.932414] sof-audio-pci 0000:00:1f.3: 0x00000000: 00000000 00000000 00000000 00000000                                                                                                     
[    5.932417] sof-audio-pci 0000:00:1f.3: 0x00000004: 00000000 00000000 00000000 00000000                                                                                                     
[    5.932419] sof-audio-pci 0000:00:1f.3: 0x00000008: 00000000 00000000 00000000 00000000
[    5.932421] sof-audio-pci 0000:00:1f.3: 0x0000000c: 6d588180 ffff9e6c 47f2fdc0 ffffbcce
[    5.932423] sof-audio-pci 0000:00:1f.3: 0x00000010: a341bbe0 ffffffff 6ff29000 ffff9e6c
[    5.932425] sof-audio-pci 0000:00:1f.3: 0x00000014: 30340828 ffff9e6c 000c0800 00000000                                                                                                     
[    5.932427] sof-audio-pci 0000:00:1f.3: 0x00000018: 00000000 00000000 47f2fe50 ffffbcce  
[    5.932429] sof-audio-pci 0000:00:1f.3: 0x0000001c: 47f2fe78 ffffbcce c0587194 ffffffff                                                                                                     
[    5.932431] sof-audio-pci 0000:00:1f.3: error: waking up any trace sleepers                                                                                                                 
[    6.756998] capability: warning: `main' uses 32-bit capabilities (legacy support in use)                                                                                                    
[    7.933258] sof-audio-pci 0000:00:1f.3: error: ipc timed out for 0x60050000 size 12         
[    7.933266] sof-audio-pci 0000:00:1f.3: info: preventing DSP entering D3 state to preserve context                                                                                          
[    7.933278] sof-audio-pci 0000:00:1f.3: status: fw entered - code 00000005                                                                                                                  
[    7.933384] sof-audio-pci 0000:00:1f.3: error: runtime exception                   
[    7.933388] sof-audio-pci 0000:00:1f.3: error: trace point 00004000
[    7.933391] sof-audio-pci 0000:00:1f.3: error: panic at :0                
[    7.933395] sof-audio-pci 0000:00:1f.3: error: DSP Firmware Oops
[    7.933398] sof-audio-pci 0000:00:1f.3: error: Exception Cause: IllegalInstructionCause, Illegal instruction                                                                                
[    7.933402] sof-audio-pci 0000:00:1f.3: EXCCAUSE 0x00000000 EXCVADDR 0x00000000 PS       0x00000000 SAR     0x00000000                                                                      
[    7.933406] sof-audio-pci 0000:00:1f.3: EPC1     0x00000000 EPC2     0x00000000 EPC3     0x00000000 EPC4    0x00000000                                                                      
[    7.933410] sof-audio-pci 0000:00:1f.3: EPC5     0x00000000 EPC6     0x00000000 EPC7     0x00000000 DEPC    0x00000000
[    7.933414] sof-audio-pci 0000:00:1f.3: EPS2     0x00000000 EPS3     0x00000000 EPS4     0x00000000 EPS5    0x00000000
[    7.933417] sof-audio-pci 0000:00:1f.3: EPS6     0x00000000 EPS7     0x00000000 INTENABL 0x00000000 INTERRU 0x00000000
[    7.933420] sof-audio-pci 0000:00:1f.3: stack dump from 0x00000000                                                                                                                          
[    7.933425] sof-audio-pci 0000:00:1f.3: 0x00000000: 00000000 00000000 00000000 00000000                                                                                                     
[    7.933430] sof-audio-pci 0000:00:1f.3: 0x00000004: 00000000 00000000 00000000 00000000                                                                                                     
[    7.933434] sof-audio-pci 0000:00:1f.3: 0x00000008: 00000000 00000000 00000000 00000000
[    7.933438] sof-audio-pci 0000:00:1f.3: 0x0000000c: 6d588180 ffff9e6c 47ef7ab0 ffffbcce
[    7.933442] sof-audio-pci 0000:00:1f.3: 0x00000010: 47ef7a40 ffffbcce ee1a3200 7e4394ef
[    7.933446] sof-audio-pci 0000:00:1f.3: 0x00000014: 30340828 ffff9e6c 2fd3fc28 ffff9e6c
[    7.933450] sof-audio-pci 0000:00:1f.3: 0x00000018: 47ef7b88 ffffbcce c0549237 ffffffff
[    7.933454] sof-audio-pci 0000:00:1f.3: 0x0000001c: a2569819 ffffffff 47ef7b48 ffffbcce
[    7.933465] sof-audio-pci 0000:00:1f.3: error: hda irq intsts 0x00000000 intlctl 0xc0000003 rirb 00                                                                                         
[    7.933468] sof-audio-pci 0000:00:1f.3: error: dsp irq ppsts 0x00000000 adspis 0x00000000                                                                                                   
[    7.933476] sof-audio-pci 0000:00:1f.3: error: host status 0x00000000 dsp status 0x00000000 mask 0x00000003                                                                                 
[    7.933479] sof-audio-pci 0000:00:1f.3: error: waking up any trace sleepers                                                                                                                 
[    7.933732] sof-audio-pci 0000:00:1f.3: ASoC: error at snd_soc_pcm_component_trigger on 0000:00:1f.3: -110                                                                                  
[    7.933740]  DMIC: ASoC: trigger FE cmd: 0 failed: -110  

To Reproduce
E.g. disable FORMAT_S32LE for TGL with main

Reproduction Rate
every time

Expected behavior
hmm, hard to tell

Impact
Blocker to possible memory reduction methods

Environment

1. Branch name and commit hash of the 2 repositories: sof (firmware/topology) and linux (kernel driver).
   • Kernel: chromeos kernel
   • SOF: top of mux: conditionally compile format specific functions #4128
2. Name of the topology file
   • Topology: TGL-*
3. Name of the platform(s) on which the bug is observed.
   • Platform: TGL although i expect this to affect any platform

[juimonen]

I can get this reproduced actually without mux/demux... with mixer pipeline at least. So we have something fishy in the format propagation when restricting to 16 bits. looks like NULL pointer access currently. Debugging more...

[juimonen]

@lgirdwood @cujomalainey I think the problem is in dai.c calling pcm_get_conversion_function() in pcm_converter. With processing reduced to 16 bit, it also compiles out everything else than 16 bit processing. If your dai is configured to 24 or 32 bit, pcm_get_conversion_function returns NULL.... and we have no checks here

I'm not sure what's the proper way to fix this:

1. just return error here if function is NULL -> you're responsible to match the dai format to pipeline format
2. remove the "reduction" from converter -> it should always be able to convert formats from rest of the pipeline to dai

[juimonen]

@cujomalainey can you try out if #4131 at least fixes the crash? we can then see what to do next...

[juimonen]

#4131 is merged. we can still experiment with #4135 and/or close this...

[cujomalainey]

I think we can close this and follow up in the PR for experimentation