On Mon, Nov 21, 2016 at 03:17:16PM +0200, Andy Shevchenko wrote: > On Mon, 2016-11-21 at 13:29 +0100, Greg Kroah-Hartman wrote: > > On Wed, Nov 16, 2016 at 08:06:32PM +0200, Andy Shevchenko wrote: > > > Static analyzer can not see balanced locking if the lock is acquired > > > and > > > released conditionally. > > > > > > drivers/tty/serial/serial_core.c:131:17: warning: context imbalance > > > in 'uart_stop' - unexpected unlock > > > include/linux/spinlock.h:362:9: warning: context imbalance in > > > 'uart_start' - unexpected unlock > > > include/linux/spinlock.h:362:9: warning: context imbalance in > > > 'uart_put_char' - unexpected unlock > > > include/linux/spinlock.h:362:9: warning: context imbalance in > > > 'uart_write' - unexpected unlock > > > include/linux/spinlock.h:362:9: warning: context imbalance in > > > 'uart_write_room' - unexpected unlock > > > include/linux/spinlock.h:362:9: warning: context imbalance in > > > 'uart_chars_in_buffer' - unexpected unlock > > > include/linux/spinlock.h:362:9: warning: context imbalance in > > > 'uart_flush_buffer' - unexpected unlock > > > > > > Get rid of macros and implement their functionality in place. This > > > will also > > > help to enable runtime PM in more clean way later. > > > > Really? Because now the code is harder to follow from a "what is this > > doing" point of view as you have to duplicate logic in a number of > > different places. > > We have number of places in kernel where we take reference count of > something and then using spin lock. I don't see how this makes harder to > follow. You are "open-coding" these macros, for no other reason than to silence a tool that is incorrect. This makes further changes in the code harder to make. > > What happened to making these just static inline functions instead? > > So, the function needs annotation for checker and flags parameter will > go like a) taking address of it, b) dereference it. > > For me it will look not shiny. This looks worse than the code today, a very non-shiny patch as-is, sorry, but I'm not going to take this. greg k-h -- To unsubscribe from this list: send the line "unsubscribe linux-serial" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
