On Wed, Aug 7, 2024 at 9:52 AM Al Viro <viro@xxxxxxxxxxxxxxxxxx> wrote: > > On Wed, Aug 07, 2024 at 09:22:59AM +0200, Mateusz Guzik wrote: > > > Well it's your call, you wrote the thing and I need the problem out of > > the way, so I'm not going to argue about the patchset. > > > > I verified it boots and provides the expected perf win [I have to > > repeat it is highly variable between re-runs because of ever-changing > > offsets between different inode allocations resulting in different > > false-sharing problems; i'm going to separately mail about that] > > > > I think it will be fine to copy the result from my commit message and > > denote it's from a different variant achieving the same goal. > > > > That said feel free to use my commit message in whatever capacity, > > there is no need to mention me. > > Original analysis had been yours, same for "let's change the calling > conventions for do_dentry_open() wrt path refcounting", same for > the version I'd transformed into that... FWIW, my approach to > that had been along the lines of "how do we get it consistently, > whether we go through vfs_open() or finish_open()", which pretty > much required keeping hold on the path until just before > terminate_walk(). This "transfer from nd->path to whatever borrowed > it" was copied from path_openat() (BTW, might be worth an inlined helper > next to terminate_walk(), just to document that it's not an accidental > property of terminate_walk()) and that was pretty much it. > > Co-developed-by: seems to be the usual notation these days for > such situations - that really had been incremental changes. > > Anyway, I really need to get some sleep before writing something > usable as commit messages... Nobody is getting a Turing award for noticing the extra ref trip and eliding it. Co-developed-by is fine with me if you insist on sharing credit. My only objective here is to expedite the fix so that I can get on with speeding up refcount management. :) -- Mateusz Guzik <mjguzik gmail.com>
