Re: CVS commit: src/sys/dev/usb

[Valery Ushakov <uwe%stderr.spb.ru@localhost>]

On Sat, Mar 14, 2020 at 10:27:27 -0400, Christos Zoulas wrote:

> > I don't belive that "if".  It's like claiming you got rid of a stain
> > on a wallpaper after you demolish a wall (not load-bearing,
> > fortunately) and have to put it back and put new wallpaper. :) Get rid
> > of the stain, sure; but may be looking closely with a bit of patience
> > might have been less drastic and as effective.
> 
> To fix the kernhist ones I looked with a lot of patience and even then,
> I missed quite a few ones (the ones in the final commit). It is really
> difficult to find them, specially because the DPRINTF macros are
> used sometimes for regular debugging and other times for kernhist.
> In the end I had to add a fake printf function in kernhist.h like below,
> and then filter out the error messages about too many arguments for
> format.
> 
> christos
> 
> --- kernhist.h  2020-03-13 23:03:13.973939910 -0400
> +++ kernhist.h.orig     2020-03-13 22:59:37.237495925 -0400
> @@ -207,6 +207,11 @@
>  #define KERNHIST_PRINTNOW(E) /* nothing */
>  #endif
> 
> +// Just for format checking
> +static __inline __printflike(1, 2) void
> +__kernhist_printf(const char *fmt __unused, ...) {
> +}
> +
>  #define KERNHIST_LOG(NAME,FMT,A,B,C,D) \
>  do { \
>         unsigned int _i_, _j_; \
> @@ -227,6 +232,7 @@
>         _e_->v[1] = (uintmax_t)(B); \
>         _e_->v[2] = (uintmax_t)(C); \
>         _e_->v[3] = (uintmax_t)(D); \
> +       __kernhist_printf(FMT, _e_->v[0], _e_->v[1], _e_->v[2], _e_->v[3]); \
>         KERNHIST_PRINTNOW(_e_); \
>  } while (/*CONSTCOND*/ 0)

How is is affected by the decision to change (or not) 0x%x to %#x?

-uwe