Re: [PATCH v6 2/3] zram: fix deadlock with sysfs attribute usage and module removal

[Greg KH]

On Sat, Jul 10, 2021 at 12:28:51PM -0700, Andrew Morton wrote:
> On Fri,  2 Jul 2021 17:19:57 -0700 Luis Chamberlain <mcgrof@xxxxxxxxxx> wrote:
> 
> > +#define MODULE_DEVICE_ATTR_FUNC_STORE(_name) \
> > +static ssize_t module_ ## _name ## _store(struct device *dev, \
> > +				   struct device_attribute *attr, \
> > +				   const char *buf, size_t len) \
> > +{ \
> > +	ssize_t __ret; \
> > +	if (!try_module_get(THIS_MODULE)) \
> > +		return -ENODEV; \
> > +	__ret = _name ## _store(dev, attr, buf, len); \
> > +	module_put(THIS_MODULE); \
> > +	return __ret; \
> > +}
> 
> I assume that Greg's comments on try_module_get() are applicable here
> also.

Yes, this is still broken code and does not do what it says it does,
please do not merge it.

Again, almost anything that does try_module_get(THIS_MODULE) is broken,
this code included.  I'll write more in a week or so when I get a chance
to get to this series in my reviews...

thanks,

greg k-h