On Tue, Apr 4, 2023 at 10:07 AM Darrick J. Wong <djwong@xxxxxxxxxx> wrote:
>
\> Now that we've made kernel and userspace use the same tolower code for
> computing directory index hashes, add that to the selftest code.
Please just delete this test. It's really really fundamentally wrong.
The fact that you even *think* that you use the same tolower() as user
space does shows just that you don't even understand how user space
works.
Really. The only thing this series shows is that you do not understand
the complexities.
Lookie here: compile and run this program:
#include <stdio.h>
#include <ctype.h>
#include <locale.h>
int main(int argc, char **argv)
{
printf("tolower(0xc4)=%#x\n", tolower(0xc4));
setlocale(LC_ALL, "C");
printf("tolower(0xc4)=%#x\n", tolower(0xc4));
setlocale(LC_ALL, "sv_SE.iso88591");
printf("tolower(0xc4)=%#x\n", tolower(0xc4));
}
and on my machine, I get this:
tolower(0xc4)=0xc4
tolower(0xc4)=0xc4
tolower(0xc4)=0xe4
and the important thing to note is that "on my machine". The first
line could be *different* on some other machine (and the last line
could be too: there's no guarantee that the sv_SE locale even exists).
So this whole "kernel and userspace use the same tolower code"
sentence is simply completely and utterly wrong. It's not even "wrong"
in the sense fo "that's not true". It's "wrong" in the sense "that
shows that you didn't understand the problem at all".
Put another way: saying "5+8=10" is wrong. But saying "5+8=tiger" is
nonsensical.
Your patches are nonsensical.
Linus
