On Wed, Jan 04, 2017 at 10:36:18AM +0100, Jiri Kosina wrote:
> On Wed, 4 Jan 2017, Xu, Even wrote:
>
> > > [ ... snip ... ]
> > > >
> > > > +static ssize_t ishtp_cl_write(struct file *file, const char __user
> > > > *ubuf,
> > > > + size_t length, loff_t *offset)
> > > > +{
> > > > + struct ishtp_cl_miscdev *ishtp_cl_misc = file-
> > > > >private_data;
> > > > + struct ishtp_cl *cl;
> > > > + void *write_buf;
> > > > + struct ishtp_device *dev;
> > > > + int ret;
> > > > +
> > > > + /* Non-blocking semantics are not supported */
> > > > + if (file->f_flags & O_NONBLOCK) {
> > > > + ret = -EINVAL;
> > > > + goto out_unlock;
> > > When taking the error path here you'd try to unlock
> > > ishtp_cl_misc->cl_mutex before actually acquiring it.
> > >
> >
> > Thanks for your comments, Jiri, I update my patch below:
>
> Greg, how would you like to handle this (after you're done with reviewing
> it of course), given that this driver depends on the preceeding 6 patches
> to intel-ish-hid driver?
Let me go review it now, but if it's ok, I have no problem taking it in
my tree, or if you want to take it through yours, that's fine with me,
which ever is easiest for you.
thanks,
greg k-h
