Comparing paths with special characters on Mac UTF-8 - c

Question

We have a kext that checks if a path is a subdir of another path and does some magic if it is.
This all works fine as long we don't have special characters in our path (characters like ë)
We feed some working paths into the system by a helper application that can communicate with the kext.
I've isolated the problem to this code:
#include <stdio.h>
#include <string.h>
int main ()
{
char* path = "/Users/user/test/tëst/test"; //Sent by the system
char* wp = "/Users/user/test/tëst"; //Some path we claim to be ours
size_t wp_len = strlen(wp);
if (strncmp (wp,path,wp_len) == 0) //Check is path is a subpath
{
printf ("matched %s\n", path);
}else {
printf ("could not match\n");
}
return 0;
}
I've created a Gist, so the encoding does not go lost with the browser: https://gist.github.com/fvandepitte/ec28f4321a48061808d0095853af7bd7
Someone knows how i can check if path is a subpath of wp without losing too much performance (this code runs in the kernel)?

Answer 1

I've copy/pasted the source straight from the browser into a file (test.c). It prints could not match for me.
If I dump the file using od this is what I see:
bash-3.2$ od -c test.c
0000000 # i n c l u d e < s t d i o .
0000020 h > \n # i n c l u d e < s t r
0000040 i n g . h > \n \n i n t m a i n
0000060 ( ) \n { \n c h a r * p a
0000100 t h = " / U s e r s / u s e
0000120 r / t e s t / t ë ** s t / t e s
0000140 t " ; / / S e n t b y t h
0000160 e s y s t e m \n c h a r *
0000200 w p = " / U s e r s /
0000220 u s e r / t e s t / t e ̈ ** s t
0000240 " ; / / S o m e p a t h w
Notice that the tëst of path comes out as t ë ** s t,
but the tëst of wp comes out as t e ̈ ** s t, which is different: so strncmp will fail when comparing ë and e.
If I copy the tëst from path paste that into wp's assignment then I get matched /Users/user/test/tëst/test, so strncmp seems to work fine.
I don't know these two strings differ like this, I can only assume that the two strings are using different encodings somehow. The strncmp function compares strings per byte, so ë and e ̈ are considered different. If you want to use strncmp, then unfortunately there's no easy solution to this other than insuring that both strings use the same encoding.
FWIW - I'm running on macOS 10.12.1, with clang version Apple LLVM version 8.0.0 (clang-800.0.42.1)
EDIT: I've downloaded pathtest.cpp from your github link just to double-check things. I've run od -c pathtest.cpp and I see the same problem.