On Tue, 2016-02-16 at 10:31 +0300, Dan Carpenter wrote:
> [ checkpatch.pl told someone to introduce a bug and they did... ]
Yeah, it happens.
People should not take a message like "generally" as
an absolute.
"else is not generally useful after a break or return"
checkpatch isn't a flow control analysis tool.
coccinelle might be able to do this better.
There isn't a coccinelle rule for this as far as I know.
> Hello Erik Arfvidson,
>
> The patch 05f1b17ec7aa: "staging: unisys: fix else statement in
> visornic_main.c" from Feb 8, 2016, leads to the following static
> checker warning:
>
> drivers/staging/unisys/visornic/visornic_main.c:381 visornic_serverdown()
> error: double unlock 'spin_lock:&devdata->priv_lock'
>
> drivers/staging/unisys/visornic/visornic_main.c
> 356 static int
> 357 visornic_serverdown(struct visornic_devdata *devdata,
> 358 visorbus_state_complete_func complete_func)
> 359 {
> 360 unsigned long flags;
> 361
> 362 spin_lock_irqsave(&devdata->priv_lock, flags);
> 363 if (!devdata->server_down && !devdata->server_change_state) {
> 364 if (devdata->going_away) {
> 365 spin_unlock_irqrestore(&devdata->priv_lock, flags);
> 366 dev_dbg(&devdata->dev->device,
> 367 "%s aborting because device removal pending\n",
> 368 __func__);
> 369 return -ENODEV;
> 370 }
> 371 devdata->server_change_state = true;
> 372 devdata->server_down_complete_func = complete_func;
> 373 spin_unlock_irqrestore(&devdata->priv_lock, flags);
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> 374 visornic_serverdown_complete(devdata);
> 375 } else if (devdata->server_change_state) {
> 376 dev_dbg(&devdata->dev->device, "%s changing state\n",
> 377 __func__);
> 378 spin_unlock_irqrestore(&devdata->priv_lock, flags);
> 379 return -EINVAL;
> 380 }
> 381 spin_unlock_irqrestore(&devdata->priv_lock, flags);
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> 382 return 0;
> 383 }
>
> regards,
> dan carpenter
_______________________________________________
devel mailing list
devel@xxxxxxxxxxxxxxxxxxxxxx
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel
