> On 25.03.2019, at 11:06, Stefan Wahren <stefan.wahren@xxxxxxxx> wrote: > > Am 25.03.19 um 10:52 schrieb kernel@xxxxxxxxxxxxxxxx: >>> On 25.03.2019, at 10:28, Stefan Wahren <stefan.wahren@xxxxxxxx> wrote: >>> >>> Hi Martin, >>> >>> Am 24.03.19 um 18:50 schrieb kernel@xxxxxxxxxxxxxxxx: >>>> + >>>> +#if defined(CONFIG_DEBUG_FS) >>>> +#define BCM2835_AUX_INCR(field) ((field)++) >>>> + u64 count_transfer_polling; >>>> + u64 count_transfer_irq; >>>> + u64 count_transfer_irq_after_poll; >>>> + >>>> + struct dentry *debugfs_dir; >>>> +#else >>>> +#define BCM2835_AUX_INCR(field) >>>> +#endif /* CONFIG_DEBUG_FS */ >>>> }; >>>> >>> is there a chance to avoid these ifdefs CONFIG_DEBUG_FS? >>> >> The idea was not to add those stats at all when debugfs is disabled. >> Other groups (e.g net/can) prefer it this way… > > Yes, this is always a balance between size reduction and code readability. > > My biggest concern was about BCM2835_AUX_INCR > just tell me which way you prefer it - anyway: disabling DEBUG is quite hard to do in the first place (too many components select it automatically) so I am always wondering why I need to hop thtu those loops. If you agree I will remove those specific ifdefs and macros. Martin