Re: [PATCH v2 02/11] kcov: apply clang-format to kcov code

[Alexander Potapenko <glider@google.com>]

On Fri, Jun 27, 2025 at 10:02 AM Peter Zijlstra <peterz@infradead.org> wrote:
>
> On Thu, Jun 26, 2025 at 03:41:49PM +0200, Alexander Potapenko wrote:
> > kcov used to obey clang-format style, but somehow diverged over time.
> > This patch applies clang-format to kernel/kcov.c and
> > include/linux/kcov.h, no functional change.
>
> I'm not sure I agree this is in fact a good thing. Very questionable
> style choices made.

Adding Miguel, who maintains clang-format.

> I had to kill clang-format hard in my nvim-lsp-clangd setup, because
> clang-format is such a piece of shit.

Random fact that I didn't know before: 1788 out of 35503 kernel .c
files are already formatted according to the clang-format style.
(I expected the number to be much lower)

>
> > -static inline void kcov_task_init(struct task_struct *t) {}
> > -static inline void kcov_task_exit(struct task_struct *t) {}
> > -static inline void kcov_prepare_switch(struct task_struct *t) {}
> > -static inline void kcov_finish_switch(struct task_struct *t) {}
> > -static inline void kcov_remote_start(u64 handle) {}
> > -static inline void kcov_remote_stop(void) {}
> > +static inline void kcov_task_init(struct task_struct *t)
> > +{
> > +}
> > +static inline void kcov_task_exit(struct task_struct *t)
> > +{
> > +}
> > +static inline void kcov_prepare_switch(struct task_struct *t)
> > +{
> > +}
> > +static inline void kcov_finish_switch(struct task_struct *t)
> > +{
> > +}
> > +static inline void kcov_remote_start(u64 handle)
> > +{
> > +}
> > +static inline void kcov_remote_stop(void)
> > +{
> > +}
>
> This is not an improvement.

Fair enough.
I think we can fix this by setting AllowShortFunctionsOnASingleLine:
Empty, SplitEmptyFunction: false in .clang-format

Miguel, do you think this is a reasonable change?


> >
> >  struct kcov_percpu_data {
> > -     void                    *irq_area;
> > -     local_lock_t            lock;
> > -
> > -     unsigned int            saved_mode;
> > -     unsigned int            saved_size;
> > -     void                    *saved_area;
> > -     struct kcov             *saved_kcov;
> > -     int                     saved_sequence;
> > +     void *irq_area;
> > +     local_lock_t lock;
> > +
> > +     unsigned int saved_mode;
> > +     unsigned int saved_size;
> > +     void *saved_area;
> > +     struct kcov *saved_kcov;
> > +     int saved_sequence;
> >  };
> >
> >  static DEFINE_PER_CPU(struct kcov_percpu_data, kcov_percpu_data) = {
>
> This is just plain wrong. Making something that was readable into a
> trainwreck.

Setting AlignConsecutiveDeclarations: AcrossEmptyLinesAndComments will
replace the above with the following diff:

 struct kcov_percpu_data {
-       void                    *irq_area;
-       local_lock_t            lock;
-
-       unsigned int            saved_mode;
-       unsigned int            saved_size;
-       void                    *saved_area;
-       struct kcov             *saved_kcov;
-       int                     saved_sequence;
+       void        *irq_area;
+       local_lock_t lock;
+
+       unsigned int saved_mode;
+       unsigned int saved_size;
+       void        *saved_area;
+       struct kcov *saved_kcov;
+       int          saved_sequence;
 };

(a bit denser, plus it aligns the variable names, not the pointer signs)
Does this look better?

>
> Please either teach clang-format sensible style choices, or refrain from
> using it.



-- 
Alexander Potapenko
Software Engineer

Google Germany GmbH
Erika-Mann-Straße, 33
80636 München

Geschäftsführer: Paul Manicle, Liana Sebastian
Registergericht und -nummer: Hamburg, HRB 86891
Sitz der Gesellschaft: Hamburg