On Sat, 6 Nov 2021 22:13:34 +0100 Vlastimil Babka <vbabka@xxxxxxx> wrote:
> On 11/6/21 22:12, Linus Torvalds wrote:
> > On Sat, Nov 6, 2021 at 1:49 PM Linus Torvalds
> > <torvalds@xxxxxxxxxxxxxxxxxxxx> wrote:
> >>
> >> This workflow can result in more conflicts for me than what Andrew
> >> used to do ("send against current linus tip"), but it means that when
> >> conflicts happen, they get all the merge resolution help that git
> >> gives you, and hopefully what gets tested (over the months that it can
> >> be in -mm) is closer to what gets sent to me.
> >
> > .. and resolving the conflicts (none of which looked bad), I think
> > that part of the resolution ends up doing very similar things to your
> > fixup patch.
>
> If this needed resolution, didn't the resolution exist in -next already?
Yes, but I had it queued after linux-next.patch so it got lost in the
unholy mess that linux-next becomes during the merge window.
I'm still figuring this out. In retrospect I should have moved this
patch "mm/vmscan: throttle reclaim until some writeback completes if
congested" to the post-linux-next section weeks ago, then waited for
the prerequisites to be merged into mainline. That way the unaltered,
tested patch would have smoothly slotted in late in the merge window.
