+Arnd, who probably saw this already...
On Tue, 31 May 2022 14:52:16 PDT (-0700), Linus Torvalds wrote:
> On Tue, May 31, 2022 at 10:13 AM Palmer Dabbelt <[email protected]> wrote:
>>
>> I have a single manual resolution in the fe510 device tree, just the result of
>> some conflicting fixes. It's not showing up in my merge, but a bunch of other
>> automatic resolutions are which is a bit worrisome on my end -- I don't usually
>> touch other trees, but we had a handful of big cross-tree things this time.
>
> Gaah. I'd have normally preferred for things like this to go through
> Arnd, but it looks like he at least ack'ed these things..
Ya, makes sense -- I don't really like touching other peoples' trees,
it's a headache for everyone. There is a bit of overhead involved in
doing one of the multi-tree merges, though, so I'm never quite sure
where to draw the line. We did one for the spinlocks where it was
pretty clear that was the way to go, as it was used by a handful of
trees and didn't take that long to get the RISC-V bits cleaned up.
The compat stuff was a mess for a cycle or two, though, and RISC-V was
the only user of the new bits. I always feel bad trying to dump messes
on other folks, so I figured it was easier to just fix it myself and by
the time that happened it looked like everyone else had stopped paying
attention. I poked it a few times both before and after putting into my
for-next, but with the Acks I just took it.
After seeing those conflicts I kind of wanted to push for it to get
merged a different way. Had it not been both before some other
stuff and a persistent headache I probably would have just sent the PR
before that merge and asked folks again, but I guess I just wanted to
finally have this one done.
Though now that I say that, merging something because it was a headache
is probably the wrong message to send folks... ;)
> I've obviously pulled it, as you can see from the pr-tracker-bot reply
> that already went out.
Thanks, I'll try not to make a mess next time.