On Wed, Jan 15, 2020 at 06:38:31PM -0500, Masayoshi Mizuma wrote:
[..]
> > +/* Should be called with inode->plock_mutex held */
> > +static struct lo_inode_plock *lookup_create_plock_ctx(struct lo_data *lo,
> > + struct lo_inode
> > *inode,
> > + uint64_t lock_owner,
> > + pid_t pid, int *err)
> > +{
> > + struct lo_inode_plock *plock;
> > + char procname[64];
> > + int fd;
> > +
> > + plock =
> > + g_hash_table_lookup(inode->posix_locks,
> > GUINT_TO_POINTER(lock_owner));
> > +
> > + if (plock) {
> > + return plock;
> > + }
> > +
> > + plock = malloc(sizeof(struct lo_inode_plock));
> > + if (!plock) {
> > + *err = ENOMEM;
> > + return NULL;
> > + }
> > +
> > + /* Open another instance of file which can be used for ofd locks. */
> > + sprintf(procname, "%i", inode->fd);
> > +
> > + /* TODO: What if file is not writable? */
> > + fd = openat(lo->proc_self_fd, procname, O_RDWR);
> > + if (fd == -1) {
>
> > + *err = -errno;
>
> I think the errno is positive value, so the minus isn't needed?
>
> *err = errno;
That's sounds right. Thanks.
David, will you be able to do this tweak in your tree or you want me to
send a separate fix patch.
Thanks
Vivek
>
> Otherwise looks good to me.
>
> Reviewed-by: Masayoshi Mizuma <m.miz...@jp.fujitsu.com>
>
> Thanks,
> Masa
>
