Skip to contentSkip to conversation

sockets: avoid formatting buffer that may not be NULL terminated

PATCH Superseded Daniel P. Berrange 4 participants 6 messages Jun 26, 2017 – Jul 5, 2017

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 sockets: avoid formatting buffer that may not be NULL terminated

The 'sun_path' field in the sockaddr_un struct is not required to be NULL termianted, so when reporting an error, we must use the separate 'path' variable which is guaranteed terminated.

Fixes a bug spotted by coverity that was introduced in

commit ad9579aaa16d5b385922d49edac2c96c79bcfb62
Author: Daniel P. Berrange <[email protected]>
Date: Thu May 25 16:53:00 2017 +0100

sockets: improve error reporting if UNIX socket path is too long

---
 util/qemu-sockets.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
util/qemu-sockets.c +1 -1
index 51bf279..37d386f 100644
@@ -930,7 +930,7 @@ static int unix_listen_saddr(UnixSocketAddress *saddr,
     strncpy(un.sun_path, path, sizeof(un.sun_path));
 
     if (bind(sock, (struct sockaddr*) &un, sizeof(un)) < 0) {
-        error_setg_errno(errp, errno, "Failed to bind socket to %s", un.sun_path);
+        error_setg_errno(errp, errno, "Failed to bind socket to %s", path);
         goto err;
     }
     if (listen(sock, 1) < 0) {
Signature · 1 line
2.9.3
Reviewed-by on v1 1/1
#1 lore raw
Quoted 2 lines · Daniel P. Berrange

The 'sun_path' field in the sockaddr_un struct is not required to be NULL termianted, so when reporting an error, we must use

s/NULL/NUL/

NULL is a pointer, NUL is the '\0' character.

Quoted 20 lines

the separate 'path' variable which is guaranteed terminated.

Fixes a bug spotted by coverity that was introduced in

commit ad9579aaa16d5b385922d49edac2c96c79bcfb62
Author: Daniel P. Berrange <[email protected]>
Date: Thu May 25 16:53:00 2017 +0100

sockets: improve error reporting if UNIX socket path is too long

---
 util/qemu-sockets.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
util/qemu-sockets.c +0 -0
index 51bf279..37d386f 100644

util/qemu-sockets.c static int unix_listen_saddr(UnixSocketAddress *saddr, · line 936

Quoted 8 lines
util/qemu-sockets.c +1 -1
@@ -930,7 +930,7 @@ static int unix_listen_saddr(UnixSocketAddress *saddr,
     strncpy(un.sun_path, path, sizeof(un.sun_path));
 
     if (bind(sock, (struct sockaddr*) &un, sizeof(un)) < 0) {
-        error_setg_errno(errp, errno, "Failed to bind socket to %s", un.sun_path);
+        error_setg_errno(errp, errno, "Failed to bind socket to %s", path);
         goto err;
     }
     if (listen(sock, 1) < 0) {

Apart from the nit in the comment, the patch looks right to me:

#2 lore raw
Quoted 7 lines · Thomas Huth
Quoted 2 lines · Daniel P. Berrange

The 'sun_path' field in the sockaddr_un struct is not required to be NULL termianted, so when reporting an error, we must use

s/NULL/NUL/

NULL is a pointer, NUL is the '\0' character.

I wanted to point out the same thing to someone recently, so I chased up a reference to the NUL character in RFC 20 "ASCII format for Network Interchange". After all, no one can argue with an RFC.

What I found shocked me! There must be a typo in the ASCII RFC:
https://tools.ietf.org/html/rfc20#section-5.2

I closed my browser tab quickly and headed to Wikipedia instead. If the primary source didn't support my argument, I could always count on good old Wikipedia...

But do you know what I found? Someone had conflated nul and null on the Wikipedia entry:
https://en.wikipedia.org/wiki/Null_character

Amateurs! The Wikipedia editors probably didn't have the intellectual calibre to question the correctness of the RFC text the way I did.

But to cut a long story short, as my search continued the evidence became overwhelming. It is acceptable to refer to the nul character as the null character.

Coming back to the patch in question, although we can't complain about the "NULL" it's with considerable joy that I'd like to highlight:

s/termianted/terminated/

:)
Stefan

#3 lore raw
Quoted 24 lines · Stefan Hajnoczi
Quoted 7 lines · Thomas Huth
Quoted 2 lines · Daniel P. Berrange

The 'sun_path' field in the sockaddr_un struct is not required to be NULL termianted, so when reporting an error, we must use

s/NULL/NUL/

NULL is a pointer, NUL is the '\0' character.

I wanted to point out the same thing to someone recently, so I chased up a reference to the NUL character in RFC 20 "ASCII format for Network Interchange". After all, no one can argue with an RFC.

What I found shocked me! There must be a typo in the ASCII RFC:
https://tools.ietf.org/html/rfc20#section-5.2

I closed my browser tab quickly and headed to Wikipedia instead. If the primary source didn't support my argument, I could always count on good old Wikipedia...

But do you know what I found? Someone had conflated nul and null on the Wikipedia entry:
https://en.wikipedia.org/wiki/Null_character

Quoted 6 lines

Amateurs! The Wikipedia editors probably didn't have the intellectual calibre to question the correctness of the RFC text the way I did.

But to cut a long story short, as my search continued the evidence became overwhelming. It is acceptable to refer to the nul character as the null character.

Well, I don't see a real problem here - as long as you write "null" with lowercase letters. "NULL" with uppercase letters is the pointer. "NUL" with uppercase letters is the character. And "null" with lowercase letters is just a context-sensitive word :-)

Thomas

#4 lore raw
In reply to Stefan Hajnoczi
Quoted 31 lines · Stefan Hajnoczi
Quoted 7 lines · Thomas Huth
Quoted 2 lines · Daniel P. Berrange

The 'sun_path' field in the sockaddr_un struct is not required to be NULL termianted, so when reporting an error, we must use

s/NULL/NUL/

NULL is a pointer, NUL is the '\0' character.

I wanted to point out the same thing to someone recently, so I chased up a reference to the NUL character in RFC 20 "ASCII format for Network Interchange". After all, no one can argue with an RFC.

What I found shocked me! There must be a typo in the ASCII RFC:
https://tools.ietf.org/html/rfc20#section-5.2

I closed my browser tab quickly and headed to Wikipedia instead. If the primary source didn't support my argument, I could always count on good old Wikipedia...

But do you know what I found? Someone had conflated nul and null on the Wikipedia entry:
https://en.wikipedia.org/wiki/Null_character

Amateurs! The Wikipedia editors probably didn't have the intellectual calibre to question the correctness of the RFC text the way I did.

But to cut a long story short, as my search continued the evidence became overwhelming. It is acceptable to refer to the nul character as the null character.

Quoted 7 lines

Coming back to the patch in question, although we can't complain about the "NULL" it's with considerable joy that I'd like to highlight:

s/termianted/terminated/

:)
Stefan

Despite this shocking, shocking precedence, I formally object to the use of NULL in the context of C for anything other than "one particular null pointer". C's confusing enough without overloading identifiers.

Reviewed-by on v1 1/1
#5 lore raw
In reply to Daniel P. Berrange
Quoted 10 lines · Daniel P. Berrange

The 'sun_path' field in the sockaddr_un struct is not required to be NULL termianted, so when reporting an error, we must use the separate 'path' variable which is guaranteed terminated.

Fixes a bug spotted by coverity that was introduced in

commit ad9579aaa16d5b385922d49edac2c96c79bcfb62
Author: Daniel P. Berrange <[email protected]>
Date: Thu May 25 16:53:00 2017 +0100

Quoted 6 lines

sockets: improve error reporting if UNIX socket path is too long

---
 util/qemu-sockets.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)