[src/trunk]: src/usr.bin/make make: document platform dependency in string pa...

[rillig <rillig%NetBSD.org@localhost>]

details:   https://anonhg.NetBSD.org/src/rev/8e16f6c1acff
branches:  trunk
changeset: 366081:8e16f6c1acff
user:      rillig <rillig%NetBSD.org@localhost>
date:      Fri May 13 21:42:30 2022 +0000

description:
make: document platform dependency in string pattern matching

No unit test for this edge case since all other unit tests are platform-
independent.

To reproduce:
$ make clean
$ make -s PROG=s-make NOMAN=yes USER_CFLAGS=-fsigned-char
$ make clean
$ make -s PROG=u-make NOMAN=yes USER_CFLAGS=-funsigned-char
$ make clean
$ range=$(lua -e 'print(("[%c-%c]"):format(0xe4, 0x61))')
$ ./s-make -V "\${:UM:M$range}\${:UN:N$range}"
M
$ ./u-make -V "\${:UM:M$range}\${:UN:N$range}"
N

diffstat:

 usr.bin/make/str.c |  28 +++++++++++++++++++++++-----
 1 files changed, 23 insertions(+), 5 deletions(-)

diffs (56 lines):

diff -r 3ffe1c68d021 -r 8e16f6c1acff usr.bin/make/str.c
--- a/usr.bin/make/str.c        Fri May 13 20:37:01 2022 +0000
+++ b/usr.bin/make/str.c        Fri May 13 21:42:30 2022 +0000
@@ -1,4 +1,4 @@
-/*     $NetBSD: str.c,v 1.90 2022/05/13 20:37:01 rillig Exp $  */
+/*     $NetBSD: str.c,v 1.91 2022/05/13 21:42:30 rillig Exp $  */
 
 /*
  * Copyright (c) 1988, 1989, 1990, 1993
@@ -71,7 +71,7 @@
 #include "make.h"
 
 /*     "@(#)str.c      5.8 (Berkeley) 6/1/90"  */
-MAKE_RCSID("$NetBSD: str.c,v 1.90 2022/05/13 20:37:01 rillig Exp $");
+MAKE_RCSID("$NetBSD: str.c,v 1.91 2022/05/13 21:42:30 rillig Exp $");
 
 
 static HashTable interned_strings;
@@ -293,6 +293,26 @@
 }
 
 /*
+ * XXX: In the extreme edge case that one of the characters is from the basic
+ * execution character set and the other isn't, the result of the comparison
+ * differs depending on whether plain char is signed or unsigned.
+ *
+ * An example is the character range from \xE4 to 'a', where \xE4 may come
+ * from U+00E4 'Latin small letter A with diaeresis'.
+ *
+ * If char is signed, \xE4 evaluates to -28, the first half of the condition
+ * becomes -28 <= '0' && '0' <= 'a', which evaluates to true.
+ *
+ * If char is unsigned, \xE4 evaluates to 228, the second half of the
+ * condition becomes 'a' <= '0' && '0' <= 228, which evaluates to false.
+ */
+static bool
+in_range(char e1, char c, char e2)
+{
+       return (e1 <= c && c <= e2) || (e2 <= c && c <= e1);
+}
+
+/*
  * Str_Match -- Test if a string matches a pattern like "*.[ch]".
  * The following special characters are known *?\[] (as in fnmatch(3)).
  *
@@ -355,9 +375,7 @@
                                if (pat[1] == '-') {
                                        if (pat[2] == '\0')
                                                return neg;
-                                       if (pat[0] <= *str && *str <= pat[2])
-                                               break;
-                                       if (pat[2] <= *str && *str <= pat[0])
+                                       if (in_range(pat[0], *str, pat[2]))
                                                break;
                                        pat += 2;
                                }