[musl] Warnings in fnmatch implementation

[musl] Warnings in fnmatch implementation

[Vincent Torri]

Hello


when compiling fnmatch.c, I have there following warnings :


fnmatch.c: In function 'str_next':
fnmatch.c:34:13: warning: comparison of integer expressions of
different signedness: 'char' and 'unsigned int' [-Wsign-compare]
   34 |  if (str[0] >= 128U) {
      |             ^~
fnmatch.c: In function 'pat_next':
fnmatch.c:87:13: warning: comparison of integer expressions of
different signedness: 'char' and 'unsigned int' [-Wsign-compare]
   87 |  if (pat[0] >= 128U) {
      |             ^~
fnmatch.c: In function 'match_bracket':
fnmatch.c:129:24: warning: comparison of integer expressions of
different signedness: 'unsigned int' and 'int' [-Wsign-compare]
  129 |     if ((unsigned)k-wc <= wc2-wc ||
      |                        ^~
fnmatch.c:130:28: warning: comparison of integer expressions of
different signedness: 'unsigned int' and 'int' [-Wsign-compare]
  130 |         (unsigned)kfold-wc <= wc2-wc)
      |                            ^~
fnmatch.c:150:10: warning: comparison of integer expressions of
different signedness: 'char' and 'unsigned int' [-Wsign-compare]
  150 |   if (*p < 128U) {
      |          ^
fnmatch.c: In function 'fnmatch_internal':
fnmatch.c:232:13: warning: comparison of integer expressions of
different signedness: 'char' and 'unsigned int' [-Wsign-compare]
  232 |   if (s[-1] < 128U || MB_CUR_MAX==1) s--;
      |             ^

All the 128U can be replaced by just 128, and (unsigned) by just (int)

Maybe it is worth fixing them

best regards

Vincent Torri

Re: [musl] Warnings in fnmatch implementation

[Rich Felker]

On Sun, May 02, 2021 at 02:11:03PM +0200, Vincent Torri wrote:
> Hello
> 
> 
> when compiling fnmatch.c, I have there following warnings :
> 
> 
> fnmatch.c: In function 'str_next':
> fnmatch.c:34:13: warning: comparison of integer expressions of
> different signedness: 'char' and 'unsigned int' [-Wsign-compare]
>    34 |  if (str[0] >= 128U) {
>       |             ^~
> fnmatch.c: In function 'pat_next':
> fnmatch.c:87:13: warning: comparison of integer expressions of
> different signedness: 'char' and 'unsigned int' [-Wsign-compare]
>    87 |  if (pat[0] >= 128U) {
>       |             ^~
> fnmatch.c: In function 'match_bracket':
> fnmatch.c:129:24: warning: comparison of integer expressions of
> different signedness: 'unsigned int' and 'int' [-Wsign-compare]
>   129 |     if ((unsigned)k-wc <= wc2-wc ||
>       |                        ^~
> fnmatch.c:130:28: warning: comparison of integer expressions of
> different signedness: 'unsigned int' and 'int' [-Wsign-compare]
>   130 |         (unsigned)kfold-wc <= wc2-wc)
>       |                            ^~
> fnmatch.c:150:10: warning: comparison of integer expressions of
> different signedness: 'char' and 'unsigned int' [-Wsign-compare]
>   150 |   if (*p < 128U) {
>       |          ^
> fnmatch.c: In function 'fnmatch_internal':
> fnmatch.c:232:13: warning: comparison of integer expressions of
> different signedness: 'char' and 'unsigned int' [-Wsign-compare]
>   232 |   if (s[-1] < 128U || MB_CUR_MAX==1) s--;
>       |             ^
> 
> All the 128U can be replaced by just 128, and (unsigned) by just (int)

They definitely can't be replaced with just 128. Changing them to 128
changes the meaning of the expression and introduces a serious bug
(prevents handling of any non-ASCII characters on archs where plain
char is a signed type). For example -64 < 128 is true but -64 < 128U
is false.

Rich