On Sat, 3 Apr 2021, Maciej W. Rozycki wrote:
> I'll see if I can bisect the problem and report back. I don't have a
> FlashPoint adapter to verify that part of the driver, but I guess for your
> change alone a MultiMaster one such as mine will suffice.
FYI I have now tracked down the cause of the regression, which is outside
BusLogic code; specifically commit af73623f5f10 ("[SCSI] sd: Reduce buffer
size for vpd request"). It wasn't exactly easy as for an odd reason Linux
versions ~3.15 through to ~4.5 usually crash with this system in early
startup before anything is output to the serial console.
I'm still working on a proper fix, but with said commit reverted the
system boots and with your series applied it has survived system-wide fsck
even, which happened to get scheduled. So for BT-958:
Tested-by: Maciej W. Rozycki <macro@xxxxxxxxxxx>
Maciej
