Kernel crash on large file operations

phkr

Hi all!
I have some weird bug with Firefly-RK3399 board: kernel suddenly crashes on large file operations (starting from 300Mb). I had this with different kernels (both Rockchip and mainline), so it seems to be hardware problem. In some cases starting infinite loop calling sync command together with main operation can delay or even prevent crash (depends on operation and delay between syncs) but i don't think keeping it running during all session is a good idea. Can it be fixed somehow?
I attach kernel log with modelled operation (1.6Gb was written before crash):

1. dd if=/dev/zero of=test bs=1G count=3
   
2. [  119.555187] SError Interrupt on CPU1, code 0xbf000000 -- SError
   
3. [  119.555189] CPU: 1 PID: 20401 Comm: dd Not tainted 5.4.5-1-ARTIX #1
   
4. [  119.555191] Hardware name: Firefly-RK3399 Board (DT)
   
5. [  119.555192] pstate: 20000005 (nzCv daif -PAN -UAO)
   
6. [  119.555193] pc : __arch_copy_from_user+0x1bc/0x240
   
7. [  119.555194] lr : copyin+0x90/0xa0
   
8. [  119.555196] sp : ffff80001985bb20
   
9. [  119.555196] x29: ffff80001985bb20 x28: 0000000000000000
   
10. [  119.555199] x27: 0000000000001000 x26: ffff800010f5f350
    
11. [  119.555202] x25: 0000000000000000 x24: ffff80001985bd50
    
12. [  119.555204] x23: ffff80001985bd40 x22: 0000000000000000
    
13. [  119.555207] x21: 0000000000001000 x20: ffff000009c01000
    
14. [  119.555209] x19: 0000000000001000 x18: 0000000000000014
    
15. [  119.555212] x17: 00000000f9f64b4b x16: 000000004c3a51cb
    
16. [  119.555214] x15: 000000007aeec261 x14: 0000000000000000
    
17. [  119.555217] x13: 0000000000000000 x12: 0000000000000000
    
18. [  119.555219] x11: 0000000000000000 x10: 0000000000000000
    
19. [  119.555221] x9 : 0000000000000000 x8 : 0000000000000000
    
20. [  119.555224] x7 : 0000000000000000 x6 : ffff000009c00050
    
21. [  119.555226] x5 : ffff000009c01000 x4 : 0000000000000000
    
22. [  119.555229] x3 : 0000ffff8419d000 x2 : 0000000000000f40
    
23. [  119.555231] x1 : 0000ffff8419d090 x0 : ffff000009c00000
    
24. [  119.555234] Kernel panic - not syncing: Asynchronous SError Interrupt
    
25. [  119.555236] CPU: 1 PID: 20401 Comm: dd Not tainted 5.4.5-1-ARTIX #1
    
26. [  119.555237] Hardware name: Firefly-RK3399 Board (DT)
    
27. [  119.555238] Call trace:
    
28. [  119.555239]  dump_backtrace+0x0/0x170
    
29. [  119.555240]  show_stack+0x24/0x30
    
30. [  119.555241]  dump_stack+0xac/0xd0
    
31. [  119.555242]  panic+0x144/0x31c
    
32. [  119.555243]  __stack_chk_fail+0x0/0x28
    
33. [  119.555244]  arm64_serror_panic+0x84/0x90
    
34. [  119.555245]  do_serror+0x11c/0x120
    
35. [  119.555246]  el1_error+0x84/0xf8
    
36. [  119.555248]  __arch_copy_from_user+0x1bc/0x240
    
37. [  119.555249]  iov_iter_copy_from_user_atomic+0xec/0x370
    
38. [  119.555250]  generic_perform_write+0xf0/0x1d0
    
39. [  119.555251]  __generic_file_write_iter+0x134/0x1b8
    
40. [  119.555252]  ext4_file_write_iter+0xd0/0x338
    
41. [  119.555253]  new_sync_write+0xf8/0x190
    
42. [  119.555254]  __vfs_write+0x74/0x90
    
43. [  119.555255]  vfs_write+0xe4/0x1c8
    
44. [  119.555257]  ksys_write+0x78/0x100
    
45. [  119.555257]  __arm64_sys_write+0x24/0x30
    
46. [  119.555259]  el0_svc_handler+0x84/0x190
    
47. [  119.555260]  el0_svc+0x8/0xc
    
48. [  119.555280] SMP: stopping secondary CPUs
    
49. [  119.555280] Kernel Offset: disabled
    
50. [  119.555282] CPU features: 0x0002,20006008
    
51. [  119.555283] Memory Limit: none

Copy the code

phkr

The problem seems to have gone after i switched u-boot from slarm build to mainline. Looks weird but the same command worked correctly after this change