It is not possible with the current fuse code, but let's protect ourselves from regressions in the future. Cc: Christian Brauner <brauner@xxxxxxxxxx> Cc: Seth Forshee <sforshee@xxxxxxxxxx> Cc: Miklos Szeredi <miklos@xxxxxxxxxx> Cc: Amir Goldstein <amir73il@xxxxxxxxx> Cc: Bernd Schubert <bschubert@xxxxxxx> Cc: <linux-fsdevel@xxxxxxxxxxxxxxx> Signed-off-by: Alexander Mikhalitsyn <aleksandr.mikhalitsyn@xxxxxxxxxxxxx> --- v4: - this commit added --- fs/fuse/dir.c | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/fs/fuse/dir.c b/fs/fuse/dir.c index d316223bd00b..dd967402bf12 100644 --- a/fs/fuse/dir.c +++ b/fs/fuse/dir.c @@ -1473,6 +1473,14 @@ static int fuse_access(struct inode *inode, int mask) BUG_ON(mask & MAY_NOT_BLOCK); + /* + * We should not send FUSE_ACCESS to the userspace + * when idmapped mounts are enabled as for this case + * we have fc->default_permissions = 1 and access + * permission checks are done on the kernel side. + */ + WARN_ON_ONCE(!(fm->sb->s_iflags & SB_I_NOIDMAP)); + if (fm->fc->no_access) return 0; -- 2.34.1
