Re: [PATCH] printf: add support for printing symbolic error codes

[Andy Shevchenko <andy.shevchenko@gmail.com>]

On Sat, Aug 31, 2019 at 12:48 AM Rasmus Villemoes
<linux@rasmusvillemoes.dk> wrote:
>
> It has been suggested several times to extend vsnprintf() to be able
> to convert the numeric value of ENOSPC to print "ENOSPC". This is yet
> another attempt. Rather than adding another %p extension, simply teach
> plain %p to convert ERR_PTRs. While the primary use case is
>
>   if (IS_ERR(foo)) {
>     pr_err("Sorry, can't do that: %p\n", foo);
>     return PTR_ERR(foo);
>   }
>
> it is also more helpful to get a symbolic error code (or, worst case,
> a decimal number) in case an ERR_PTR is accidentally passed to some
> %p<something>, rather than the (efault) that check_pointer() would
> result in.
>
> With my embedded hat on, I've made it possible to remove this.
>
> I've tested that the #ifdeffery in errcode.c is sufficient to make
> this compile on arm, arm64, mips, powerpc, s390, x86 - I'm sure the
> 0day bot will tell me which ones I've missed.
>
> The symbols to include have been found by massaging the output of
>
>   find arch include -iname 'errno*.h' | xargs grep -E 'define\s*E'
>
> In the cases where some common aliasing exists
> (e.g. EAGAIN=EWOULDBLOCK on all platforms, EDEADLOCK=EDEADLK on most),
> I've moved the more popular one (in terms of 'git grep -w Efoo | wc)
> to the bottom so that one takes precedence.

> +/*
> + * Ensure these tables to not accidentally become gigantic if some
> + * huge errno makes it in. On most architectures, the first table will
> + * only have about 140 entries, but mips and parisc have more sparsely
> + * allocated errnos (with EHWPOISON = 257 on parisc, and EDQUOT = 1133
> + * on mips), so this wastes a bit of space on those - though we
> + * special case the EDQUOT case.
> + */
> +#define E(err) [err + BUILD_BUG_ON_ZERO(err <= 0 || err > 300)] = #err

Hmm... Perhaps better to define the upper boundary with something like

#define __E_POSIX_UPPER_BOUNDARY 300 // name sucks, I know

> +#define E(err) [err - 512 + BUILD_BUG_ON_ZERO(err < 512 || err > 550)] = #err

Similar to 550?

> +const char *errcode(int err)
> +{
> +       /* Might as well accept both -EIO and EIO. */
> +       if (err < 0)
> +               err = -err;
> +       if (err <= 0) /* INT_MIN or 0 */
> +               return NULL;
> +       if (err < ARRAY_SIZE(codes_0))
> +               return codes_0[err];
> +       if (err >= 512 && err - 512 < ARRAY_SIZE(codes_512))
> +               return codes_512[err - 512];
> +       /* But why? */
> +       if (IS_ENABLED(CONFIG_MIPS) && err == EDQUOT) /* 1133 */
> +               return "EDQUOT";

Another possibility is to initialize the errors at run time with radix tree.

> +       return NULL;
> +}

> @@ -2111,6 +2112,31 @@ static noinline_for_stack
>  char *pointer(const char *fmt, char *buf, char *end, void *ptr,
>               struct printf_spec spec)
>  {
> +       /* %px means the user explicitly wanted the pointer formatted as a hex value. */
> +       if (*fmt == 'x')
> +               return pointer_string(buf, end, ptr, spec);

But instead of breaking switch case apart can we use...

> +
> +       /* If it's an ERR_PTR, try to print its symbolic representation. */
> +       if (IS_ERR(ptr)) {

...  if (IS_ERR() && *fmt != 'x') {
here?

> +               long err = PTR_ERR(ptr);
> +               const char *sym = errcode(-err);
> +               if (sym)
> +                       return string_nocheck(buf, end, sym, spec);
> +               /*
> +                * Funky, somebody passed ERR_PTR(-1234) or some other
> +                * non-existing Efoo - or more likely
> +                * CONFIG_SYMBOLIC_ERRCODE=n. None of the
> +                * %p<something> extensions can make any sense of an
> +                * ERR_PTR(), and if this was just a plain %p, the
> +                * user is still better off getting the decimal
> +                * representation rather than the hash value that
> +                * ptr_to_id() would generate.
> +                */
> +               spec.flags |= SIGN;
> +               spec.base = 10;
> +               return number(buf, end, err, spec);
> +       }


-- 
With Best Regards,
Andy Shevchenko