Re: [PATCH net-next v2 0/5] net/mlx4: TX helpers and deferred doorbell fix
From: Joe Damato
Date: Thu Nov 19 2026 - 19:31:31 EST
On Thu, Nov 19, 2026 at 04:05:59PM -0800, Jakub Kicinski wrote:
> On Thu, 19 Nov 2026 15:34:21 -0800 Joe Damato wrote:
> > On Thu, Nov 19, 2026 at 02:38:52PM -0800, Jakub Kicinski wrote:
> > > On Tue, 17 Nov 2026 08:43:25 -0800 Joe Damato wrote:
> > > > The only fix in this series is the deferred doorbell flush (patch 5), but I
> > > > need the other code (patches 1-4) for SW USO support I plan to submit next
> > > > if this series is applied.
> > >
> > > Not sure if this is what you meant but this may read as if you were
> > > sending a fix for net-next for convenience of sending more patches.
> > > To be clear - that's not a valid excuse. That said, here, we're talking
> > > about a 10 year old error path bug which nobody apparently have been
> > > bothered by. We don't have space in net for such things at this point
> > > of the cycle, anyway :S
> >
> > I'm probably just being dense and not following what you mean here.
> >
> > I was looking at a local branch that implements SW USO in mlx4 and it was
> > getting to be quite long with a lot of patches. I thought that breaking it up
> > and moving some of the patches into a standalone series would make it easier
> > to review.
> >
> > I suppose you are saying I should have just kept it as one longer series,
> > instead?
>
> No, not at all.
>
> You said "The only fix in this series is....I plan to submit next if
> this series is applied". Which triggers my suspicion that you packed
> a fix with -next stuff for expedience / not having to wait for a
> net->net-next merge before you send the rest of USO. I don't think
> that's what you meant but since I had to ping Tariq anyway I decided
> to wag a finger at you.
Oh, OK.
To be clear: I was not trying to circumvent the system or the rules and if you
(or Tariq) prefers, I am happy to peel this apart and send it to net in the
next cycle.
I am not trying to rush anything in; I'm happy to wait until the next
cycle if that is what is preferred.
I included it to net-next because my perception (which is clearly incorrect)
is that net is much stricter lately and issues which have never been triggered
IRL or reported by users may not meet the bar required for net.
I'll be more careful in the future; I just legitimately did not think this
change would be accepted to net.
> IOW I'm commenting purely on packing what can be perceived as a net
> fix in a net-next series.
Sorry for the confusion I caused :(