On Tue, Oct 25, 2022 at 12:48 AM Linus Torvalds
<torvalds@xxxxxxxxxxxxxxxxxxxx> wrote:
>
> Thanks for noticing, and I'll also Cc this reply to the linux-scsi
> mailing list so that I can link to this longer explanation in my fix
> commit message.
Ok, the fix is now there as 1a2dcbdde82e ("scsi: mpt3sas: re-do lost
mpt3sas DMA mask fix").
Re-reading my commit message, I realize it's a bit confusing, because
I kept the authorship of the fix as Sreekanth, but then I wrote the
explanation as myself.
Oh well, the actual important information should be there - and
perhaps even more importantly, the fix itself should be active again.
And thanks again for noticing, this was all on me.
And honestly, it wasn't even a complicated merge conflict.
I think that what happened here was simply that the commit that
conflicted with the one-liner fix was
(a) also from Sreekanth
(b) also had that "DMA_BIT_MASK(32)" in it
So visually it looked like "Ahh, the SCSI tree already has this fix",
and so I just took the SCSI tree version of the code.
But while it had the same author, and while it had the same
"DMA_BIT_MASK(32)" textual pattern right there, it was just a
_different_ case of DMA_BIT_MASK(32).
So the DMA_BIT_MASK(32) fix two lines earlier from the new one got dropped.
Linus "D'oh" Torvalds
