Luis Chamberlain <mcgrof@xxxxxxxxxx> writes: > On Mon, Dec 11, 2023 at 12:25:10PM +0100, Thomas Weißschuh wrote: >> Before sending it I'd like to get feedback on the internal rework of the >> is_empty detection from you and/or Luis. >> >> https://git.sr.ht/~t-8ch/linux/commit/ea27507070f3c47be6febebe451bbb88f6ea707e >> or the attached patch. > > Please send as a new patch as RFC and please ensure on the To field is > first "Eric W. Biederman" <ebiederm@xxxxxxxxxxxx> with a bit more > elaborate commit log as suggested from my review below. If there are > any hidden things me and Joel could probably miss I'm sure Eric will > be easily able to spot it. > >> From ea27507070f3c47be6febebe451bbb88f6ea707e Mon Sep 17 00:00:00 2001 >> From: =?UTF-8?q?Thomas=20Wei=C3=9Fschuh?= <linux@xxxxxxxxxxxxxx> >> Date: Sun, 3 Dec 2023 21:56:46 +0100 >> Subject: [PATCH] sysctl: move permanently empty flag to ctl_dir >> MIME-Version: 1.0 >> Content-Type: text/plain; charset=UTF-8 >> Content-Transfer-Encoding: 8bit >> >> Simplify the logic by always keeping the permanently_empty flag on the >> ctl_dir. >> The previous logic kept the flag in the leaf ctl_table and from there >> transferred it to the ctl_table from the directory. >> >> This also removes the need to have a mutable ctl_table and will allow >> the constification of those structs. > Please elaborate a bit more on this here in your next RFC. > It's a pretty aggressive cleanup, specially with the new hipster guard() > call but I'd love Eric's eyeballs on a proper v2. I will look at a v2 time permitting. My sense is that changing the locking probably make sense as a separate patch. Eric
