* src/conf_post.h (ATTRIBUTE_UNUSED): Remove.  All uses replaced
by MAYBE_UNUSED, and moved to start as needed for C2x.

* src/conf_post.h [HYBRID_MALLOC || DARWIN_OS && HAVE_UNEXEC]:
Include <stdlib.h> here, before redirecting 'malloc' and friends
to their hybrid_* and unexec_* equivalents.  #undef malloc and
friends before redefining.  Provide prototypes for the
replacements.  Suggested by Paul Eggert <eggert@cs.ucla.edu>.
* src/gmalloc.c [HYBRID_MALLOC]: Remove declarations of 'malloc'
and friends, as they are now redundant: we include <stdlib.h> in
conf_post.h before redefining 'malloc' etc., and that provides
prototypes from system headers.

* configure.ac (HYBRID_MALLOC): Remove kludge to avoid replacement
of 'free' by Gnulib.  (Bug#36649)

* src/conf_post.h (HAS_ATTRIBUTE):
* src/emacs-module.h.in (EMACS_ATTRIBUTE_NONNULL):
Port to clang 3.4 and earlier.

Run "TZ=UTC0 admin/update-copyright".

Problem reported by Andreas Schwab in:
https://lists.gnu.org/r/emacs-devel/2020-05/msg00650.html
* src/conf_post.h (HAS_ATTR_no_sanitize): Define to false in case
cpp is picky, fixing a longstanding glitch here.
(ATTRIBUTE_NO_SANITIZE_ADDRESS, ATTRIBUTE_NO_SANITIZE_UNDEFINED):
Use HAS_ATTRIBUTE, not __has_attribute.

attribute.h is partly designed for C2X forward compatibility,
since C2X will add some standard attributes.  Using its macros
should help insulate Emacs from C2X teething problems.
* src/conf_post.h: Include attribute.h.
(HAS_ATTRIBUTE, HAS_FEATURE): Rename from __has_attribute and
__has_feature, to avoid polluting the builtin namespace.
All uses changed.
(ATTRIBUTE_COLD, ATTRIBUTE_FORMAT, FALLTHROUGH, ATTRIBUTE_CONST)
(ATTRIBUTE_PURE, ATTRIBUTE_UNUSED, ATTRIBUTE_MAY_ALIAS)
(ATTRIBUTE_MALLOC, ATTRIBUTE_ALLOC_SIZE)
(ATTRIBUTE_RETURNS_NONNULL): Remove, as attribute.h does this now.
(NO_INLINE, EXTERNALLY_VISIBLE, ARG_NONNULL, ATTRIBUTE_UNUSED):
Simplify by defining in terms of attribute.h macros.
* src/systhread.h (ATTRIBUTE_WARN_UNUSED_RESULT): Remove.
All uses replaced by attribute.h’s NODISCARD.

In character.h, replace macros with inline functions or enums
when this is easy.  This improves maintainability and
on my platform (Fedora 31 x86-64, gcc -O2) improved CPU
performance very slightly (0.3%) on ‘make compile-always’.
* src/buffer.h (SANE_TAB_WIDTH, CHARACTER_WIDTH):
Move here from character.h, and make them inline functions.
Tune CHARACTER_WIDTH so that ASCII_CHAR_WIDTH is no longer needed.
(sanitize_tab_width, sanitize_char_width):
Move here from character.h.
* src/character.h (MAX_CHAR, MAX_UNICODE_CHAR, MAX_1_BYTE_CHAR)
(MAX_2_BYTE_CHAR, MAX_3_BYTE_CHAR, MAX_4_BYTE_CHAR)
(MAX_5_BYTE_CHAR, MIN_MULTIBYTE_LEADING_CODE)
(MAX_MULTIBYTE_LEADING_CODE, MAX_MULTIBYTE_LENGTH):
Now enum constants instead of macros.
* src/character.h (CHAR_BYTES): Redo to avoid conditional branches.
(CHAR_BYTE8_P, BYTE8_TO_CHAR, UNIBYTE_TO_CHAR, CHAR_TO_BYTE8)
(CHAR_TO_BYTE_SAFE, CHAR_BYTE8_HEAD_P, CHARACTERP)
(CHECK_CHARACTER, CHECK_CHARACTER_CAR, CHECK_CHARACTER_CDR)
(CHAR_PRINTABLE_P, CHAR_BYTES, CHAR_LEADING_CODE, BYTE8_STRING)
(LEADING_CODE_P, TRAILING_CODE_P, CHAR_HEAD_P)
(BYTES_BY_CHAR_HEAD):
Now inline functions instead of macros.
(ASCII_CHAR_WIDTH): Remove; no longer used.
* src/conf_post.h (ATTRIBUTE_PURE): New macro.
* src/lisp.h (char_table_ref): Use it, for better inlining.
* src/fns.c (base64_decode_1): Add now-necessary casts.

* src/conf_post.h (EXTERN_INLINE) [!EMACS_EXTERN_INLINE]:
Make it static inline, not merely static.  This is a worthwhile
performance improvement on my two platforms A and B (see below).
On my platform A this change improves user+system CPU performance of
‘make compile-always’ by 52% on an -Og build, and by 1.4% on the
default -O2 build.  On my platform B this improves the same benchmark
by 41% on an -Og build, and by -0.8% on the default -O2 build.
That "-0.8%" is a small negative for this change, and I recall that
it is why I didn't make this change earlier.  However, Platform B uses
an older GCC so we needn't worry overmuch about this small negative.

With this change the performance advantage of -O2 over -Og has dropped
on platform A; formerly -O2 was 87% faster than -Og, and now it is
only 25% faster.  On platform B the performance advantage of -O2 over
-Og has dropped from being 62% faster to being 14% faster.

Platform A is GCC 9.3.1 20200317 (Red Hat 9.3.1-1) on Fedora 31
x86-64 (AMD Phenom II X4 910e, circa 2010).  Platform B is GCC (Ubuntu
7.5.0-3ubuntu1~18.04) 7.5.0 on Ubuntu 18.04.4 (Intel Xeon E3-1225 V2,
circa 2012).

This patch was inspired by a suggestion by Andrea Corallo in:
https://lists.gnu.org/r/emacs-devel/2020-04/msg00263.html

Original problem report by N. Jackson in:
https://lists.gnu.org/r/emacs-devel/2020-03/msg00047.html
I found some other warnings when I used gcc, and fixed them
with this patch.
* lib-src/etags.c: Include verify.h.
(xnmalloc, xnrealloc): Tell the compiler that NITEMS is
nononnegative and ITEM_SIZE is positive.
* src/conf_post.h (__has_attribute_returns_nonnull)
(ATTRIBUTE_RETURNS_NONNULL): New macros.
* src/editfns.c (Fuser_full_name): Don’t assume Fuser_login_name
returns non-nil.
* src/intervals.c (rotate_right, rotate_left, update_interval):
* src/intervals.h (LENGTH, LEFT_TOTAL_LENGTH, RIGHT_TOTAL_LENGTH):
Use TOTAL_LENGTH0 or equivalent on intervals that might be null.
* src/intervals.h (TOTAL_LENGTH): Assume arg is nonnull.
(TOTAL_LENGTH0): New macro, with the old TOTAL_LENGTH meaning.
(make_interval, split_interval_right): Add ATTRIBUTE_RETURNS_NONNULL.
* src/pdumper.c (dump_check_dump_off): Now returns void, since
no caller uses the return value.  Redo assert to pacify GCC.
(decode_emacs_reloc): Add a seemingly-random eassume to pacify GCC.
Ugly, and I suspect due to a bug in GCC.

Run "TZ=UTC0 admin/update-copyright $(git ls-files)".