On Mon, Aug 1, 2022 at 4:48 AM Logan Gunthorpe <logang@xxxxxxxxxxxx> wrote: > > > > On July 29, 2022 7:48:48 PM ADT, Song Liu <song@xxxxxxxxxx> wrote: > >On Thu, Jul 28, 2022 at 7:13 AM Christoph Hellwig <hch@xxxxxxxxxxxxx> wrote: > >> > >> On Wed, Jul 27, 2022 at 03:05:56PM -0600, Logan Gunthorpe wrote: > >> > Refactor the raid5_get_active_stripe() to read more linearly in > >> > the order it's typically executed. > >> > > >> > The init_stripe() call is called if a free stripe is found and the > >> > function is exited early which removes a lot of if (sh) checks and > >> > unindents the following code. > >> > > >> > Remove the while loop in favour of the 'goto retry' pattern, which > >> > reduces indentation further. And use a 'goto wait_for_stripe' instead > >> > of an additional indent seeing it is the unusual path and this makes > >> > the code easier to read. > >> > > >> > No functional changes intended. Will make subsequent changes > >> > in patches easier to understand. > >> > >> I find the new loop even more confusing than the old one. I'd go > >> with something like the version below (on top of the whol md-next tree > >> that pulled this in way too fast..) > > > >This looks good to me. Christoph, would you mind send official patch > >for this? > > > >Thanks, > >Song > > I'm on vacation this week, but I'd be happy to send patches addressing Christoph's feedback when I'm back next week. We are in the merge window right now. So the timing is a little tricky. I will try to send pull requests with this set as-is. Then we can do follow-ups. Thanks, Song