On Sun, 2016-07-03 at 18:51 -0400, Jeff Mahoney wrote:
> On 7/3/16 6:29 AM, Tomohiro Kusumi wrote:
> > It's invalid if the given mode is neither dir nor link,
> > so don't let it fallthrough on else case.
>
> Is this case really worth crashing the entire system?
>
> Wouldn't a WARN_ON + return -EINVAL be sufficient? At the very least it
> means that a developer doesn't need to reboot to reload the module if
> they hit this.
That error return should be sufficient to cover this case and the WARN_ON() is
useful too.
And while there is some justification for the BUG_ON() earlier I also think an E
INVAL return would be more sensible there too.
BUG()ing on errors that should be handled is really bad IMHO.
>
> -Jeff
>
> > Signed-off-by: Tomohiro Kusumi <kusumi.tomohiro@xxxxxxxxx>
> > ---
> > fs/autofs4/inode.c | 4 +++-
> > 1 file changed, 3 insertions(+), 1 deletion(-)
> >
> > diff --git a/fs/autofs4/inode.c b/fs/autofs4/inode.c
> > index ad0ee93..7f35c82 100644
> > --- a/fs/autofs4/inode.c
> > +++ b/fs/autofs4/inode.c
> > @@ -305,6 +305,7 @@ int autofs4_fill_super(struct super_block *s, void
> > *data, int silent)
> > if (autofs_type_trigger(sbi->type))
> > __managed_dentry_set_managed(root);
> >
> > + BUG_ON(!root_inode->i_fop || !root_inode->i_op);
> > root_inode->i_fop = &autofs4_root_operations;
> > root_inode->i_op = &autofs4_dir_inode_operations;
> >
> > @@ -368,7 +369,8 @@ struct inode *autofs4_get_inode(struct super_block *sb,
> > umode_t mode)
> > inode->i_fop = &autofs4_dir_operations;
> > } else if (S_ISLNK(mode)) {
> > inode->i_op = &autofs4_symlink_inode_operations;
> > - }
> > + } else
> > + BUG_ON(1);
> >
> > return inode;
> > }
> >
>
>
--
To unsubscribe from this list: send the line "unsubscribe autofs" in
