Bugfix in regex char ranges affecting ranges whose upper end

corresponds to the high bit of a bitmap cell: for instance
the character \x7f when the cell size is 32 bits.

* regex.c (BITCELL_ALL1): Unused macro removed.
(BITCELL_BIT): New macro to replace occurrences of a repeated
expression.
(CHAR_SET_INDEX, CHAR_SET_BIT): Updated to use BITCELL_BIT.
(L0_fill_range): Bugfix: the mask1 calculation was producing all-zero
in the condition bt1 == BITCELL_BIT; it should produce an all-ones
mask.

2 files changed, 21 insertions, 4 deletions

diff --git a/ChangeLog b/ChangeLog
index a56288cf..d3191013 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,5 +1,19 @@
 2014-01-13  Kaz Kylheku  <kaz@kylheku.com>
 
+	Bugfix in regex char ranges affecting ranges whose upper end
+	corresponds to the high bit of a bitmap cell: for instance
+	the character \x7f when the cell size is 32 bits.
+
+	* regex.c (BITCELL_ALL1): Unused macro removed.
+	(BITCELL_BIT): New macro to replace occurrences of a repeated
+	expression.
+	(CHAR_SET_INDEX, CHAR_SET_BIT): Updated to use BITCELL_BIT.
+	(L0_fill_range): Bugfix: the mask1 calculation was producing all-zero
+	in the condition bt1 == BITCELL_BIT; it should produce an all-ones
+	mask.
+
+2014-01-13  Kaz Kylheku  <kaz@kylheku.com>
+
 	Version 74
 
 	* txr.c (version): Bumped.
diff --git a/regex.c b/regex.c
index dfe872a8..765112ce 100644
--- a/regex.c
+++ b/regex.c
@@ -62,11 +62,12 @@ typedef union regex_machine regex_machine_t;
 
 typedef unsigned int bitcell_t;
 
-#define BITCELL_ALL1 UINT_MAX
 #define CHAR_SET_SIZE (256 / (sizeof (bitcell_t) * CHAR_BIT))
 
-#define CHAR_SET_INDEX(CH) ((CH) / (sizeof (bitcell_t) * CHAR_BIT))
-#define CHAR_SET_BIT(CH) ((CH) % (sizeof (bitcell_t) * CHAR_BIT))
+#define BITCELL_BIT (sizeof (bitcell_t) * CHAR_BIT)
+
+#define CHAR_SET_INDEX(CH) ((CH) / BITCELL_BIT)
+#define CHAR_SET_BIT(CH) ((CH) % BITCELL_BIT)
 
 #define CHAR_SET_L0(CH) ((CH) & 0xFF)
 #define CHAR_SET_L1(CH) (((CH) >> 8) & 0xF)
@@ -236,7 +237,9 @@ static void L0_fill_range(cset_L0_t *L0, wchar_t ch0, wchar_t ch1)
   bitcell_t mask0 = ~(((bitcell_t) 1 << bt0) - 1);
   int bt1 = CHAR_SET_BIT(ch1);
   int bc1 = CHAR_SET_INDEX(ch1);
-  bitcell_t mask1 = (((bitcell_t) 1 << (bt1 + 1) % 32) - 1);
+  bitcell_t mask1 = (bt1 == (BITCELL_BIT - 1))
+                     ? (bitcell_t) -1
+                     : (((bitcell_t) 1 << (bt1 + 1)) - 1);
 
   if (bc1 == bc0) {
     (*L0)[bc0] |= (mask0 & mask1);