Filipe Manana @ 2026-02-17 10:53 GMT: > On Tue, Feb 17, 2026 at 10:18 AM Miquel Sabaté Solà wrote: >> >> Hello, >> >> Burenchev Evgenii @ 2026-02-16 16:16 GMT: >> >> > From ff2df73ba6483b0dc67b3ed89d2a43c49f1c2eb8 Mon Sep 17 00:00:00 2001 >> > From: Evgenii Burenchev >> > Date: Mon, 16 Feb 2026 18:39:30 +0300 >> > Subject: [PATCH] btrfs: remove dead assignment to dirid in >> > btrfs_search_path_in_tree() >> > >> > After the introduction of btrfs_search_backwards(), the directory >> > traversal state in btrfs_search_path_in_tree() is fully maintained via >> > struct btrfs_key. The local variable 'dirid' is no longer used to control >> > the search and the assignment >> > >> > dirid = key.objectid; >> > >> > has no observable effect and is dead code. >> > >> > Remove the unused assignment to avoid confusion and silence static >> > analysis warnings. >> > >> > No functional change. >> > >> > Found by Linux Verification Center (linuxtesting.org) with SVACE. >> > >> > Signed-off-by: Evgenii Burenchev >> > --- >> > fs/btrfs/ioctl.c | 1 - >> > 1 file changed, 1 deletion(-) >> > >> > diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c >> > index a6cc2d3b414c..292043b11207 100644 >> > --- a/fs/btrfs/ioctl.c >> > +++ b/fs/btrfs/ioctl.c >> > @@ -1708,7 +1708,6 @@ static noinline int btrfs_search_path_in_tree(struct btrfs_fs_info *info, >> > btrfs_release_path(path); >> > key.objectid = key.offset; >> > key.offset = (u64)-1; >> > - dirid = key.objectid; >> > } >> > memmove(name, ptr, total_len); >> > name[total_len] = '\0'; >> >> I would add a Fixes tag with the commit that made this assignment >> useless. > > No. We add Fixes tags for patches we want to backport to stable > releases, things that fix a problem affecting users, like functional > bugs or performance regressions for example. > For cleanups, removing unnecessary code, we don't want a Fixes tag, as > that will trigger unnecessary backport overhead. > Got it, thanks for the explanation! >> As far as I can tell this is commit 98d377a0894e ("Btrfs: don't >> miss inode ref items in BTRFS_IOC_INO_LOOKUP"), which made the update >> inside of the loop of 'dirid' no longer needed. This way it's easier to >> track from where this was no longer needed and whether that's really the >> case. >> >> Other than that, and taking a quick look at this function, I can already >> tell that the 'ret' variable could be initialized to 0 in the beginning >> and make the last assignment to it (just before the "out" label) not >> needed. Hence, while we are at it, could you also remove that >> assignment? >> >> Cheers, >> Miquel