"Theodore Y. Ts'o" [off-list ref] writes:

On Mon, Oct 15, 2018 at 05:12:14PM -0400, Gabriel Krisman Bertazi wrote:

quoted

+#define EXT4_ENC_STRICT_MODE_FL		(1 << 0) /* Reject invalid sequences? */

Why the question mark?

Hi Ted,

The question mark is very redundant for a flag, I admit :).  It meant to
say "Whether to reject invalid sequences" or something like that.  Will
fix in the v2.

quoted

+#define UTF8_NORMALIZATION_TYPE_NFKD	(1 << 1)
+#define UTF8_CASEFOLD_TYPE_NFKDCF	(1 << 4)
+
+static const struct ext4_sb_encoding_map {
+	char *name;
+	__u16 default_flags;
+} ext4_encoding_map[] = {
+	/* 0x0 */ { "ascii", 0x0},
+	/* 0x1 */ {"utf8-10.0.0", UTF8_NORMALIZATION_TYPE_NFKD|UTF8_CASEFOLD_TYPE_NFKDCF},
+	{0x0, 0x0},
+};
+
 #endif	/* _LINUX_EXT2_FS_H */

What uses this?  I can't find any other references in either the kernel or
e2fsprogs patches.

Only the instance ext4_encoding_map, itself, is used in this patch and
in the next one, which modifies mkefs.  It stores the string for
comparison with what the user passed in the command line.

I guess naming the structure is unnecessary, since we have only this
single const static instance.  I will change that in the v2, as well.

The current series doesn't include the huge utf8 stuff, which makes use
of the rest of the flags, but I will add that in v2 as well.


Thanks!

-- 
Gabriel Krisman Bertazi