Re: [PATCH v2 1/8] lib/printbuf: New data structure for heap-allocated strings

[Kent Overstreet <kent.overstreet@gmail.com>]

On Fri, Apr 22, 2022 at 07:22:08AM +0200, Christoph Hellwig wrote:
> On Fri, Apr 22, 2022 at 01:14:48AM -0400, Kent Overstreet wrote:
> > Christoph, you have no problem making more work for me but I can't even get you
> 
> I think you are misunderstanding this.  You are trying to create more
> work for people maintainaing the kernel by creating duplicate
> infrastructure.  The burden is always on the submitter.
>
> > to look at the bugs you introuduce in your refactorings that I report to you.
> > 
> > Still waiting on you to look at oops you introduced in bio_copy_data_iter...
> 
> I'm not sure why I shoud care about your out of tree code making
> assumptions about block layer helpers.

Wasn't just bcachefs, it affected bcache too, as Coly also reported. And I wrote
that code originally (and the whole fucking modern bvec iter infrastracture,
mind you) so please don't lecture me on making assumptions on block layer
helpers.

Here's the thing, I think you and I have somewhat different approaches to
engineering. Personaly, I find good engineering to be about tradeoffs, not
absolutism, and not letting perfect be the enemy of good.

So I'm honestly not super eager to start modifying tricky arch code that I can't
test, and digging into what looked like non trivial interactions between the way
the traceing code using seq_buf (naturally, given that's where it originates).

I like to push out code that I have high confidence in, and the patch series I
pushed out I do have confidence in, given that it's been in use for awhile and
it's well tested in my tree.

Now yes, I _could_ do a wholesale conversion of seq_buf to printbuf and delete
that code, but doing that job right, to be confident that I'm not introducing
bugs, is going to take more time than I really want to invest right now. I
really don't like to play fast and loose with that stuff.

And the reason getting this from you really irks me is that _practically every
single time_ I trip over a something nasty when I rebase and I git bisect or
blame it's something you did. I don't even bother reporting most of them to you.

I don't want to be calling you out for the work you do because on the whole it's
good and appreciated - I saw the patch series go by getting request_queue out of
filesystem land, I'm happy that's getting done. But I've also seen the stuff you
submit get _really_ churny at times for no good reason, and some really nasty,
data corrupting bugs go by, so...

Please chill out a bit if I'm not super in a rush to do it your way.