Hello Sebastian,
Thank you for the patches. By no means am I questioning your work;
I have a question, though. Please see inline.
On Wednesday, October 17, 2012 9:07 PM Sebastian Andrzej Siewior wrote:
<snip>
> +static struct usb_function *try_get_usb_function(const char *name)
> +{
> + struct usb_d_function *f;
> + struct usb_function *uf;
> +
> + mutex_lock(&func_lock);
> + list_for_each_entry(f, &func_list, list) {
> + if (!strcmp(name, f->name)) {
> + bool ok;
> + ok = try_module_get(f->mod);
Suppose this executes fine; module refcount is increased.
> + uf = ERR_PTR(-EBUSY);
> + if (!ok)
> + goto out;
> + uf = f->alloc();
But then suppose this failed
<snip>
> +
> +struct usb_function *usb_get_function(const char *name)
> +{
> + struct usb_function *uf;
> + int ret;
> +
> + uf = try_get_usb_function(name);
> + if (!IS_ERR(uf))
> + return uf;
> + ret = PTR_ERR(uf);
> + if (ret != -ENOENT)
> + return uf;
> + ret = request_module("usbfunc:%s", name);
> + if (ret < 0)
> + return ERR_PTR(ret);
> + return try_get_usb_function(name);
> +}
> +EXPORT_SYMBOL_GPL(usb_get_function);
and, in consequence, usb_get_function returns -ENOMEM.
Which is the hint to the caller that there is no need to call
usb_put_function. And so the module refcount remains increased.
Even if the caller did call usb_put_function, the uf->mod
is not set anyway, so module_put() is not called.
Am I missing something?
Thanks,
AP
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
