Re: [PATCH 3/9] dfa: rewrite handling of multibyte case_fold lexing

[Jim Meyering]

Paolo Bonzini wrote:
> Let dfacomp do the folding to lowercase of multibyte regexes, and remove
> it from grep.c.  Input strings to kwset.c are still folded outside
> kwset.c.
>
> * NEWS: Document bugfixes.
> * .x-sc_cast_of_argument_to_free: Remove.
> * src/dfa.c (wctok, addtok_wc): New.
...

This looks fine.
Two suggestions:

The literal array dimension in buf[16] should probably be
written as buf[MB_LEN_MAX], for which we'd need <limits.h>.
Also, I've added to the description of the new mbtolower function.

diff --git a/src/dfa.c b/src/dfa.c
index 5a549b3..2859570 100644
--- a/src/dfa.c
+++ b/src/dfa.c
@@ -27,6 +27,7 @@
 #include <sys/types.h>
 #include <stddef.h>
 #include <stdlib.h>
+#include <limits.h>
 #include <string.h>
 #include <locale.h>

@@ -1165,7 +1166,7 @@ addtok (token t)
 static void
 addtok_wc (wint_t wc)
 {
-  unsigned char buf[16];
+  unsigned char buf[MB_LEN_MAX];
   mbstate_t s;
   int i;
   memset (&s, 0, sizeof(s));
diff --git a/src/search.c b/src/search.c
index 6533c16..1554c08 100644
--- a/src/search.c
+++ b/src/search.c
@@ -73,10 +73,14 @@ kwsinit (void)
 }

 #ifdef MBS_SUPPORT
-/* Convert the string from BEG to N to lowercase.  Overwrite N
-   with the length of the new string, and return a pointer to
-   the lowercase string.  Successive calls to mbtolower will
-   rewrite the output buffer.  */
+/* Convert the *N-byte string, BEG, to lowercase, and write the
+   NUL-terminated result into malloc'd storage.  Upon success, set *N
+   to the length (in bytes) of the resulting string (not including the
+   trailing NUL byte), and return a pointer to the lowercase string.
+   Note that while this function returns a pointer to malloc'd storage,
+   the caller must not free it, since this function retains a pointer
+   to the buffer and reuses it on any subsequent call.
+   Upon memory allocation failure, this function exits.  */
 static char *
 mbtolower (const char *beg, size_t *n)
 {