On Mon, Jul 26, 2021 at 03:44:28PM -0400, Andrew MacLeod via Gcc-patches wrote:
> I'm testing this... Think this'll work better? we don't support any
> character sets where a space is 32 but it isnt ASCI.. and 64 isnt EBCDIC?
>
> This should handle those 2 sets.. and if is something else, it bails.
>
> This seems OK? Im testing it now on x86-64... but dont have an ebcdic
> target...
> --- a/gcc/gimple-range-fold.cc
> +++ b/gcc/gimple-range-fold.cc
> @@ -37,6 +37,7 @@ along with GCC; see the file COPYING3. If not see
> #include "cfgloop.h"
> #include "tree-ssa-loop.h"
> #include "tree-scalar-evolution.h"
> +#include "langhooks.h"
> #include "vr-values.h"
> #include "range.h"
> #include "value-query.h"
> @@ -835,6 +836,43 @@ fold_using_range::range_of_builtin_ubsan_call (irange
> &r, gcall *call,
> r.set_varying (type);
> }
>
> +// Return TRUE if we recognize the target character set and return the
> +// range for lower case and upper case letters.
> +
> +static bool
> +get_letter_range (tree type, irange &lowers, irange &uppers)
> +{
> + // ASCII
> + if (lang_hooks.to_target_charset (' ') == 0x20)
> + {
> + lowers = int_range<2> (build_int_cst (type, 'a'),
> + build_int_cst (type, 'z'));
> + uppers = int_range<2> (build_int_cst (type, 'A'),
> + build_int_cst (type, 'Z'));
Wouldn't it be safer to lang_hooks.to_target_charset ('a')
(and 'z') and just verify that their difference is that of 'z' - 'a'
and use that range and IMHO just don't bother with EBCDIC or other charsets?
Jakub
