From: Pali Rohár <pali.rohar@gmail.com>
Date: Tue, 16 May 2017 21:52:05 +0000 (+0200)
Subject: libblkid: udf: Correctly handle UDF strings encoded in 8bit OSTA Compressed Unicode
X-Git-Tag: v2.30-rc2~33
X-Git-Url: http://git.ipfire.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=22acc2681de1ae2e10d7f1e7e62538af37bc2982;p=thirdparty%2Futil-linux.git

libblkid: udf: Correctly handle UDF strings encoded in 8bit OSTA Compressed Unicode

String encoded in 8bit OSTA Compressed Unicode contains one Unicode
codepoint per 8bits. Maximal Unicode codepoint is U+FF. Which effectively
means that it is equivalent to Latin1 encoding.

Before this patch libblkid copied raw 8bit OSTA Compressed Unicode from
disk and treated it as UTF-8. It worked fine just for UTF-8 invariants,
other characters were incorrectly encoded. This patch fixes this problem.

Note that processing UUID is not fixed in this patch.
---

diff --git a/libblkid/src/superblocks/udf.c b/libblkid/src/superblocks/udf.c
index 36769779f5..96254d8813 100644
--- a/libblkid/src/superblocks/udf.c
+++ b/libblkid/src/superblocks/udf.c
@@ -218,8 +218,9 @@ real_blksz:
 			if (!have_volid) {
 				uint8_t clen = vd->type.primary.ident.clen;
 				if (clen == 8)
-					have_volid = !blkid_probe_set_id_label(pr, "VOLUME_ID",
-							vd->type.primary.ident.c, 31);
+					have_volid = !blkid_probe_set_utf8_id_label(pr, "VOLUME_ID",
+							vd->type.primary.ident.c, 31,
+							BLKID_ENC_LATIN1);
 				else if (clen == 16)
 					have_volid = !blkid_probe_set_utf8_id_label(pr, "VOLUME_ID",
 							vd->type.primary.ident.c, 31,
@@ -261,8 +262,9 @@ real_blksz:
 			if (!have_volsetid) {
 				uint8_t clen = vd->type.primary.volset_id.clen;
 				if (clen == 8)
-					have_volsetid = !blkid_probe_set_id_label(pr, "VOLUME_SET_ID",
-							vd->type.primary.volset_id.c, 127);
+					have_volsetid = !blkid_probe_set_utf8_id_label(pr, "VOLUME_SET_ID",
+							vd->type.primary.volset_id.c, 127,
+							BLKID_ENC_LATIN1);
 				else if (clen == 16)
 					have_volsetid = !blkid_probe_set_utf8_id_label(pr, "VOLUME_SET_ID",
 							vd->type.primary.volset_id.c, 127,
@@ -299,11 +301,13 @@ real_blksz:
 				uint8_t clen = vd->type.logical.logvol_id.clen;
 				if (clen == 8) {
 					if (!have_label)
-						have_label = !blkid_probe_set_label(pr,
-								vd->type.logical.logvol_id.c, 127);
+						have_label = !blkid_probe_set_utf8label(pr,
+								vd->type.logical.logvol_id.c, 127,
+								BLKID_ENC_LATIN1);
 					if (!have_logvolid)
-						have_logvolid = !blkid_probe_set_id_label(pr, "LOGICAL_VOLUME_ID",
-								vd->type.logical.logvol_id.c, 127);
+						have_logvolid = !blkid_probe_set_utf8_id_label(pr, "LOGICAL_VOLUME_ID",
+								vd->type.logical.logvol_id.c, 127,
+								BLKID_ENC_LATIN1);
 				} else if (clen == 16) {
 					if (!have_label)
 						have_label = !blkid_probe_set_utf8label(pr,