Hi Salvatore, Am 26.07.24 um 23:09 schrieb Salvatore Bonaccorso:
Now that you were able to pinpoint two versions which are very close together, would you be able to bisect the changes between 6.3.5 and 6.3.7 to find the breaking commit upstream in that 6.3.y series?
not without spending more time I can effort ATM and without finding a better test procedure (currently 1-2h per run). I'm not convinced that it is an ext4 bug. The corruption may also occur when the data is written to the nvm. The hardware (chipset-less AM5 mainboard) is quite new, the only vendor (Asrock) probably does not care about Linux and maybe AMD did not tested is carefully enough. Thus, there a quite a lot of changes from 6.3.5 -> 6.3.7 that could cause the bug. I'll try to find out more next week (other ways to reproduce the error, other hardware, ...), especially whether it is a more general issue or hardware specific. Regards Stefan
