Skip to contentSkip to conversation

gobex: Replace g_convert by utf16_to_utf8

PATCH Superseded Frédéric Danis 3 participants 4 messages Sep 10, 2024 Bluetooth drivers

v3 of 4 Superseded Latest is v4 →

lore
Share & tools
lore
#0 lore raw

PATCH v3 1/1 gobex: Replace g_convert by utf16_to_utf8

The glibc's iconv implementation is based around plug in modules for specific translations which may not been built on the platform and prevent to use g_convert().
This commit replaces it by a function similar to the existing utf8_to_utf16() function.
---
v1 -> v2: Fix missing g_free
v2 -> v3: Replace g_malloc0 by alloca
Fix UTF-16 buffer size allocation
Ensure that UTF-16 buffer is terminated by '\0'

 gobex/gobex-header.c | 41 +++++++++++++++++++++++++++++++++++------
 1 file changed, 35 insertions(+), 6 deletions(-)
gobex/gobex-header.c +35 -6
index 002ba8861..889c794df 100644
@@ -62,6 +62,34 @@ static glong utf8_to_utf16(gunichar2 **utf16, const char *utf8) {
 	return utf16_len;
 }
 
+static glong utf16_to_utf8(char **utf8, const gunichar2 *utf16, guint16 len,
+				GError **err)
+{
+	glong utf8_len;
+	guint16 utf16_len, i;
+	gunichar2 *buf;
+
+	if (*utf16 == '\0') {
+		*utf8 = NULL;
+		return 0;
+	}
+
+	/* OBEX requires network byteorder (big endian) UTF-16
+	 * but g_utf16_to_utf8 expects host-byteorder UTF-8
+	 */
+	utf16_len = len / sizeof(gunichar2);
+	buf = alloca(sizeof(gunichar2) * utf16_len);
+	for (i = 0; i < utf16_len; i++)
+		(buf)[i] = g_ntohs(utf16[i]);
+	buf[utf16_len] = '\0';
+
+	*utf8 = g_utf16_to_utf8(buf, -1, NULL, &utf8_len, err);
+	if (*utf8 == NULL)
+		utf8_len = -1;
+
+	return utf8_len;
+}
+
 static guint8 *put_bytes(guint8 *to, const void *from, gsize count)
 {
 	memcpy(to, from, count);
@@ -130,7 +158,7 @@ GObexHeader *g_obex_header_decode(const void *data, gsize len,
 	GObexHeader *header;
 	const guint8 *ptr = data;
 	guint16 hdr_len;
-	gsize str_len;
+	glong str_len;
 	GError *conv_err = NULL;
 
 	if (len < 2) {
@@ -177,13 +205,14 @@ GObexHeader *g_obex_header_decode(const void *data, gsize len,
 			goto failed;
 		}
 
-		header->v.string = g_convert((const char *) ptr, hdr_len - 5,
-						"UTF-8", "UTF-16BE",
-						NULL, &str_len, &conv_err);
-		if (header->v.string == NULL) {
+		str_len = utf16_to_utf8(&header->v.string,
+					(const gunichar2 *) ptr,
+					hdr_len - 5,
+					&conv_err);
+		if (str_len < 0) {
 			g_set_error(err, G_OBEX_ERROR,
 					G_OBEX_ERROR_PARSE_ERROR,
-					"Unicode conversion failed: %s",
+					"UTF16 to UTF8 conversion failed: %s",
 					conv_err->message);
 			g_error_free(conv_err);
 			goto failed;
Signature · 1 line
2.34.1
#1 lore raw

This is automated email and please do not reply to this email!

Dear submitter,

Thank you for submitting the patches to the linux bluetooth mailing list. This is a CI test results with your patch series:
PW Link:https://patchwork.kernel.org/project/bluetooth/list/?series=888782

---Test result---

Test Summary:

CheckPatch                    PASS      0.46 seconds
GitLint                       PASS      0.29 seconds
BuildEll                      PASS      25.24 seconds
BluezMake                     PASS      1670.15 seconds
MakeCheck                     FAIL      10.99 seconds
MakeDistcheck                 PASS      177.86 seconds
CheckValgrind                 PASS      253.85 seconds
CheckSmatch                   PASS      359.61 seconds
bluezmakeextell               PASS      120.52 seconds
IncrementalBuild              PASS      1623.29 seconds
ScanBuild                     WARNING   1051.26 seconds

Details
##############################
Test: MakeCheck - FAIL
Desc: Run Bluez Make Check
Output:

make[3]: *** [Makefile:11773: test-suite.log] Error 1
make[2]: *** [Makefile:11881: check-TESTS] Error 2
make[1]: *** [Makefile:12310: check-am] Error 2
make: *** [Makefile:12312: check] Error 2
##############################
Test: ScanBuild - WARNING
Desc: Run Scan Build
Output:
gobex/gobex-header.c:95:2: warning: Null pointer passed to 2nd parameter expecting 'nonnull'

        memcpy(to, from, count);
        ^~~~~~~~~~~~~~~~~~~~~~~

1 warning generated.

---
Regards,
Linux Bluetooth

#2 lore raw
In reply to Frédéric Danis

Hi Frédéric,

Quoted 33 lines · Frédéric Danis

The glibc's iconv implementation is based around plug in modules for specific translations which may not been built on the platform and prevent to use g_convert().
This commit replaces it by a function similar to the existing utf8_to_utf16() function.
---
v1 -> v2: Fix missing g_free
v2 -> v3: Replace g_malloc0 by alloca
Fix UTF-16 buffer size allocation
Ensure that UTF-16 buffer is terminated by '\0'

 gobex/gobex-header.c | 41 +++++++++++++++++++++++++++++++++++------
 1 file changed, 35 insertions(+), 6 deletions(-)
gobex/gobex-header.c +10 -0
index 002ba8861..889c794df 100644
@@ -62,6 +62,34 @@ static glong utf8_to_utf16(gunichar2 **utf16, const char *utf8) {
        return utf16_len;
 }
 
+static glong utf16_to_utf8(char **utf8, const gunichar2 *utf16, guint16 len,
+                               GError **err)
+{
+       glong utf8_len;
+       guint16 utf16_len, i;
+       gunichar2 *buf;
+
+       if (*utf16 == '\0') {
+               *utf8 = NULL;
+               return 0;

gobex/gobex-header.c static glong utf8_to_utf16(gunichar2 **utf16, const char *utf8) { · line 84

Quoted 10 lines
gobex/gobex-header.c +10 -0
+       }
+
+       /* OBEX requires network byteorder (big endian) UTF-16
+        * but g_utf16_to_utf8 expects host-byteorder UTF-8
+        */
+       utf16_len = len / sizeof(gunichar2);
+       buf = alloca(sizeof(gunichar2) * utf16_len);
+       for (i = 0; i < utf16_len; i++)
+               (buf)[i] = g_ntohs(utf16[i]);
+       buf[utf16_len] = '\0';

You will need to allocate one extra index if you want it to be NULL terminated otherwise you will be accessing past allocated area, that said can't you just pass the utf16_len to g_utf16_to_utf8 instead of -1?

Quoted 44 lines
gobex/gobex-header.c +15 -7
+
+       *utf8 = g_utf16_to_utf8(buf, -1, NULL, &utf8_len, err);
+       if (*utf8 == NULL)
+               utf8_len = -1;
+
+       return utf8_len;
+}
+
 static guint8 *put_bytes(guint8 *to, const void *from, gsize count)
 {
        memcpy(to, from, count);
@@ -130,7 +158,7 @@ GObexHeader *g_obex_header_decode(const void *data, gsize len,
        GObexHeader *header;
        const guint8 *ptr = data;
        guint16 hdr_len;
-       gsize str_len;
+       glong str_len;
        GError *conv_err = NULL;
 
        if (len < 2) {
@@ -177,13 +205,14 @@ GObexHeader *g_obex_header_decode(const void *data, gsize len,
                        goto failed;
                }
 
-               header->v.string = g_convert((const char *) ptr, hdr_len - 5,
-                                               "UTF-8", "UTF-16BE",
-                                               NULL, &str_len, &conv_err);
-               if (header->v.string == NULL) {
+               str_len = utf16_to_utf8(&header->v.string,
+                                       (const gunichar2 *) ptr,
+                                       hdr_len - 5,
+                                       &conv_err);
+               if (str_len < 0) {
                        g_set_error(err, G_OBEX_ERROR,
                                        G_OBEX_ERROR_PARSE_ERROR,
-                                       "Unicode conversion failed: %s",
+                                       "UTF16 to UTF8 conversion failed: %s",
                                        conv_err->message);
                        g_error_free(conv_err);
                        goto failed;
--

2.34.1

Signature · 1 line
Luiz Augusto von Dentz
#3 lore raw

Hi Luiz,

Quoted 42 lines · Luiz Augusto von Dentz

Hi Frédéric,

Quoted 38 lines · Frédéric Danis

The glibc's iconv implementation is based around plug in modules for specific translations which may not been built on the platform and prevent to use g_convert().
This commit replaces it by a function similar to the existing utf8_to_utf16() function.
---
v1 -> v2: Fix missing g_free
v2 -> v3: Replace g_malloc0 by alloca
Fix UTF-16 buffer size allocation
Ensure that UTF-16 buffer is terminated by '\0'

  gobex/gobex-header.c | 41 +++++++++++++++++++++++++++++++++++------
  1 file changed, 35 insertions(+), 6 deletions(-)
gobex/gobex-header.c +16 -0
index 002ba8861..889c794df 100644
@@ -62,6 +62,34 @@ static glong utf8_to_utf16(gunichar2 **utf16, const char *utf8) {
         return utf16_len;
  }
 
+static glong utf16_to_utf8(char **utf8, const gunichar2 *utf16, guint16 len,
+                               GError **err)
+{
+       glong utf8_len;
+       guint16 utf16_len, i;
+       gunichar2 *buf;
+
+       if (*utf16 == '\0') {
+               *utf8 = NULL;
+               return 0;
+       }
+
+       /* OBEX requires network byteorder (big endian) UTF-16
+        * but g_utf16_to_utf8 expects host-byteorder UTF-8
+        */
+       utf16_len = len / sizeof(gunichar2);
Quoted 8 lines

gobex/gobex-header.c static glong utf8_to_utf16(gunichar2 **utf16, const char *utf8) { · line 84

Quoted 4 lines
gobex/gobex-header.c +4 -0
+       buf = alloca(sizeof(gunichar2) * utf16_len);
+       for (i = 0; i < utf16_len; i++)
+               (buf)[i] = g_ntohs(utf16[i]);
+       buf[utf16_len] = '\0';

You will need to allocate one extra index if you want it to be NULL terminated otherwise you will be accessing past allocated area, that said can't you just pass the utf16_len to g_utf16_to_utf8 instead of -1?

iiuc g_utf16_to_utf8() doc, if the source len is passed the dest string may include embedded NUL characters or none, while when passing -1 it will stop at the first NUL character.I will update allocation length.

Quoted 45 lines
Quoted 44 lines
gobex/gobex-header.c +15 -7
+
+       *utf8 = g_utf16_to_utf8(buf, -1, NULL, &utf8_len, err);
+       if (*utf8 == NULL)
+               utf8_len = -1;
+
+       return utf8_len;
+}
+
  static guint8 *put_bytes(guint8 *to, const void *from, gsize count)
  {
         memcpy(to, from, count);
@@ -130,7 +158,7 @@ GObexHeader *g_obex_header_decode(const void *data, gsize len,
         GObexHeader *header;
         const guint8 *ptr = data;
         guint16 hdr_len;
-       gsize str_len;
+       glong str_len;
         GError *conv_err = NULL;
 
         if (len < 2) {
@@ -177,13 +205,14 @@ GObexHeader *g_obex_header_decode(const void *data, gsize len,
                         goto failed;
                 }
 
-               header->v.string = g_convert((const char *) ptr, hdr_len - 5,
-                                               "UTF-8", "UTF-16BE",
-                                               NULL, &str_len, &conv_err);
-               if (header->v.string == NULL) {
+               str_len = utf16_to_utf8(&header->v.string,
+                                       (const gunichar2 *) ptr,
+                                       hdr_len - 5,
+                                       &conv_err);
+               if (str_len < 0) {
                         g_set_error(err, G_OBEX_ERROR,
                                         G_OBEX_ERROR_PARSE_ERROR,
-                                       "Unicode conversion failed: %s",
+                                       "UTF16 to UTF8 conversion failed: %s",
                                         conv_err->message);
                         g_error_free(conv_err);
                         goto failed;
--

2.34.1

Signature · 6 lines
Frédéric Danis
Senior Software Engineer

Collabora Ltd.
Platinum Building, St John's Innovation Park, Cambridge CB4 0DS, United Kingdom
Registered in England & Wales, no. 5513718