Re: [PATCH v6 1/5] unicode: Add standard casefolded d_ops

From: Gabriel Krisman Bertazi
Date: Sun Feb 02 2020 - 20:46:11 EST


Daniel Rosenberg <drosen@xxxxxxxxxx> writes:

> diff --git a/include/linux/unicode.h b/include/linux/unicode.h
> index 990aa97d80496..5de313abeaf98 100644
> --- a/include/linux/unicode.h
> +++ b/include/linux/unicode.h
> @@ -4,6 +4,8 @@
>
> #include <linux/init.h>
> #include <linux/dcache.h>
> +#include <linux/fscrypt.h>
> +#include <linux/fs.h>
>
> struct unicode_map {
> const char *charset;
> @@ -30,4 +32,19 @@ int utf8_casefold(const struct unicode_map *um, const struct qstr *str,
> struct unicode_map *utf8_load(const char *version);
> void utf8_unload(struct unicode_map *um);
>
> +int utf8_ci_d_hash(const struct dentry *dentry, struct qstr *str);
> +int utf8_ci_d_compare(const struct dentry *dentry, unsigned int len,
> + const char *str, const struct qstr *name);


I don't think fs/unicode is the right place for these very specific
filesystem functions, just because they happen to use unicode. It is an
encoding library, it doesn't care about dentries, nor should know how to
handle them. It exposes a simple api to manipulate and convert utf8 strings.

I saw change was after the desire to not have these functions polluting
the VFS hot path, but that has nothing to do with placing them here.

Would libfs be better? or a casefolding library in fs/casefold.c?


--
Gabriel Krisman Bertazi