Skip to contentSkip to conversation

ocfs2: tighten up strlen() checking

PATCH In review Dan Carpenter 4 participants 7 messages Jul 10, 2010 – Jul 12, 2010

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 ocfs2: tighten up strlen() checking

This function is only called from one place and it's like this:

	dlm_register_domain(conn->cc_name, dlm_key, &fs_version);

The "conn->cc_name" is 64 characters long. If strlen(conn->cc_name) were equal to O2NM_MAX_NAME_LEN (64) that would be a bug because strlen() doesn't count the NULL character.

In fact, if you look how O2NM_MAX_NAME_LEN is used, it mostly describes 64 character buffers. The only exception is nd_name from struct o2nm_node.

Anyway I looked into it and in this case the domain string comes from osb->uuid_str in ocfs2_setup_osb_uuid(). That's 32 characters and NULL which easily fits into O2NM_MAX_NAME_LEN. This patch doesn't change how the code works, but I think it makes the code a little cleaner.

---
Or we could get rid of check entirely.

fs/ocfs2/dlm/dlmdomain.c +1 -1
index 6b5a492..084b051 100644
@@ -1671,7 +1671,7 @@ struct dlm_ctxt * dlm_register_domain(const char *domain,
 	struct dlm_ctxt *dlm = NULL;
 	struct dlm_ctxt *new_ctxt = NULL;
 
-	if (strlen(domain) > O2NM_MAX_NAME_LEN) {
+	if (strlen(domain) >= O2NM_MAX_NAME_LEN) {
 		ret = -ENAMETOOLONG;
 		mlog(ML_ERROR, "domain name length too long\n");
 		goto leave;
#1 lore raw

Hi Dan,

I think O2NM_MAX_NAME_LEN is the max valid length of the domain name. Regarding your patch, it changed to be that a domain name with length O2NM_MAX_NAME_LEN (NULL character not included) is not permitted.

Though that check seems useless for current calls, we'd better keep it.

Checking the structure,

99 struct ocfs2_cluster_connection {
100 char cc_name[GROUP_NAME_MAX];
101 int cc_namelen;

cc_name is not a NULL tailed string. the cc_namelen specifies the length of it. There does is misuse of cc_name, such as

7 832 fs/ocfs2/stack_user.c <<user_cluster_connect>>

             rc = dlm_new_lockspace(conn->cc_name, strlen(conn->cc_name),
   5    308  fs/ocfs2/stack_o2cb.c <<o2cb_cluster_connect>>
             dlm = dlm_register_domain(conn->cc_name, dlm_key, &fs_version);

Also, the uuid shouldn't be treated as NULL tailed string. 142 struct ocfs2_control_message_down {

143         char    tag[OCFS2_CONTROL_MESSAGE_OP_LEN];
144         char    space1;
145         char    uuid[OCFS2_TEXT_UUID_LEN];
146         char    space2;

thus,
the calling of
fs/ocfs2/stack_user.c:474: ocfs2_control_send_down(msg->uuid, nodenum);
->ocfs2_connection_find(uuid)

	->size_t len = strlen(name);

is suspectable.

Could you please make patch for that instead?

regards,
wengang.

Quoted 33 lines · Dan Carpenter

This function is only called from one place and it's like this:

	dlm_register_domain(conn->cc_name, dlm_key, &fs_version);

The "conn->cc_name" is 64 characters long. If strlen(conn->cc_name) were equal to O2NM_MAX_NAME_LEN (64) that would be a bug because strlen() doesn't count the NULL character.

In fact, if you look how O2NM_MAX_NAME_LEN is used, it mostly describes 64 character buffers. The only exception is nd_name from struct o2nm_node.

Anyway I looked into it and in this case the domain string comes from osb->uuid_str in ocfs2_setup_osb_uuid(). That's 32 characters and NULL which easily fits into O2NM_MAX_NAME_LEN. This patch doesn't change how the code works, but I think it makes the code a little cleaner.

---
Or we could get rid of check entirely.

fs/ocfs2/dlm/dlmdomain.c +1 -1
index 6b5a492..084b051 100644
@@ -1671,7 +1671,7 @@ struct dlm_ctxt * dlm_register_domain(const char *domain,
 	struct dlm_ctxt *dlm = NULL;
 	struct dlm_ctxt *new_ctxt = NULL;
 
-	if (strlen(domain) > O2NM_MAX_NAME_LEN) {
+	if (strlen(domain) >= O2NM_MAX_NAME_LEN) {
 		ret = -ENAMETOOLONG;
 		mlog(ML_ERROR, "domain name length too long\n");
 		goto leave;
#2 lore raw
Quoted 6 lines · Wengang Wang

Also, the uuid shouldn't be treated as NULL tailed string. 142 struct ocfs2_control_message_down {

143         char    tag[OCFS2_CONTROL_MESSAGE_OP_LEN];
144         char    space1;
145         char    uuid[OCFS2_TEXT_UUID_LEN];
146         char    space2;

The space1 and space2 characters are NULL terminators:

From ocfs2_control_do_down_msg():

        msg->space1 = msg->space2 = msg->newline = '\0';

I would have thought it had to be a "packed" struct, but it works because there are only chars in that struct. So that code is fine.

regards,
dan carpenter

#3 lore raw

Hi Dan,

Quoted 8 lines · Dan Carpenter
Quoted 6 lines · Wengang Wang

Also, the uuid shouldn't be treated as NULL tailed string. 142 struct ocfs2_control_message_down {

143         char    tag[OCFS2_CONTROL_MESSAGE_OP_LEN];
144         char    space1;
145         char    uuid[OCFS2_TEXT_UUID_LEN];
146         char    space2;
Quoted 7 lines

The space1 and space2 characters are NULL terminators:

From ocfs2_control_do_down_msg():

        msg->space1 = msg->space2 = msg->newline = '\0';

I would have thought it had to be a "packed" struct, but it works because there are only chars in that struct. So that code is fine.

Yes, it is working fine.
I would rather think that structure is misleading than smart :-D. I prefer this:

	struct ocfs2_control_message_down {
		char    tag[OCFS2_CONTROL_MESSAGE_OP_LEN + 1];
	#define space1 tag[OCFS2_CONTROL_MESSAGE_OP_LEN]
		char    uuid[OCFS2_TEXT_UUID_LEN + 1];
	#define space2 uuid[OCFS2_TEXT_UUID_LEN]
		....

regards,
wengang.

#4 lore raw
In reply to Wengang Wang

So o2dlm expects a null terminated domain name. The original patch is good as it adds the check in o2dlm only.

For userspace, we allow non-null terminating group (domain) name. That remains unchanged.

Quoted 80 lines · Wengang Wang

Hi Dan,

I think O2NM_MAX_NAME_LEN is the max valid length of the domain name. Regarding your patch, it changed to be that a domain name with length O2NM_MAX_NAME_LEN (NULL character not included) is not permitted.

Though that check seems useless for current calls, we'd better keep it.

Checking the structure,

99 struct ocfs2_cluster_connection {
100 char cc_name[GROUP_NAME_MAX];
101 int cc_namelen;

cc_name is not a NULL tailed string. the cc_namelen specifies the length of it. There does is misuse of cc_name, such as

    7    832  fs/ocfs2/stack_user.c<<user_cluster_connect>>
              rc = dlm_new_lockspace(conn->cc_name, strlen(conn->cc_name),

    5    308  fs/ocfs2/stack_o2cb.c<<o2cb_cluster_connect>>
              dlm = dlm_register_domain(conn->cc_name, dlm_key,&fs_version);

Also, the uuid shouldn't be treated as NULL tailed string. 142 struct ocfs2_control_message_down {

143         char    tag[OCFS2_CONTROL_MESSAGE_OP_LEN];
144         char    space1;
145         char    uuid[OCFS2_TEXT_UUID_LEN];
146         char    space2;

thus,
the calling of

fs/ocfs2/stack_user.c:474:      ocfs2_control_send_down(msg->uuid, nodenum);
    ->ocfs2_connection_find(uuid)
	->size_t len = strlen(name);

is suspectable.

Could you please make patch for that instead?

regards,
wengang.

Quoted 34 lines · Dan Carpenter

This function is only called from one place and it's like this:

	dlm_register_domain(conn->cc_name, dlm_key,&fs_version);

The "conn->cc_name" is 64 characters long. If strlen(conn->cc_name) were equal to O2NM_MAX_NAME_LEN (64) that would be a bug because strlen() doesn't count the NULL character.

In fact, if you look how O2NM_MAX_NAME_LEN is used, it mostly describes 64 character buffers. The only exception is nd_name from struct o2nm_node.

Anyway I looked into it and in this case the domain string comes from osb->uuid_str in ocfs2_setup_osb_uuid(). That's 32 characters and NULL which easily fits into O2NM_MAX_NAME_LEN. This patch doesn't change how the code works, but I think it makes the code a little cleaner.

---
Or we could get rid of check entirely.

fs/ocfs2/dlm/dlmdomain.c +1 -1
index 6b5a492..084b051 100644
@@ -1671,7 +1671,7 @@ struct dlm_ctxt * dlm_register_domain(const char *domain,
  	struct dlm_ctxt *dlm = NULL;
  	struct dlm_ctxt *new_ctxt = NULL;
 
-	if (strlen(domain)>  O2NM_MAX_NAME_LEN) {
+	if (strlen(domain)>= O2NM_MAX_NAME_LEN) {
  		ret = -ENAMETOOLONG;
  		mlog(ML_ERROR, "domain name length too long\n");
  		goto leave;
Acked-by on v1 1/1
#5 lore raw
In reply to Dan Carpenter
Quoted 34 lines · Dan Carpenter

This function is only called from one place and it's like this:

	dlm_register_domain(conn->cc_name, dlm_key,&fs_version);

The "conn->cc_name" is 64 characters long. If strlen(conn->cc_name) were equal to O2NM_MAX_NAME_LEN (64) that would be a bug because strlen() doesn't count the NULL character.

In fact, if you look how O2NM_MAX_NAME_LEN is used, it mostly describes 64 character buffers. The only exception is nd_name from struct o2nm_node.

Anyway I looked into it and in this case the domain string comes from osb->uuid_str in ocfs2_setup_osb_uuid(). That's 32 characters and NULL which easily fits into O2NM_MAX_NAME_LEN. This patch doesn't change how the code works, but I think it makes the code a little cleaner.

---
Or we could get rid of check entirely.

fs/ocfs2/dlm/dlmdomain.c +1 -1
index 6b5a492..084b051 100644
@@ -1671,7 +1671,7 @@ struct dlm_ctxt * dlm_register_domain(const char *domain,
  	struct dlm_ctxt *dlm = NULL;
  	struct dlm_ctxt *new_ctxt = NULL;
 
-	if (strlen(domain)>  O2NM_MAX_NAME_LEN) {
+	if (strlen(domain)>= O2NM_MAX_NAME_LEN) {
  		ret = -ENAMETOOLONG;
  		mlog(ML_ERROR, "domain name length too long\n");
  		goto leave;
#6 lore raw
In reply to Dan Carpenter
Quoted 11 lines · Dan Carpenter

This function is only called from one place and it's like this:

	dlm_register_domain(conn->cc_name, dlm_key, &fs_version);

The "conn->cc_name" is 64 characters long. If strlen(conn->cc_name) were equal to O2NM_MAX_NAME_LEN (64) that would be a bug because strlen() doesn't count the NULL character.

In fact, if you look how O2NM_MAX_NAME_LEN is used, it mostly describes 64 character buffers. The only exception is nd_name from struct o2nm_node.

Quoted 6 lines

Anyway I looked into it and in this case the domain string comes from osb->uuid_str in ocfs2_setup_osb_uuid(). That's 32 characters and NULL which easily fits into O2NM_MAX_NAME_LEN. This patch doesn't change how the code works, but I think it makes the code a little cleaner.

This patch is now in the fixes branch of ocfs2.git.

Joel

Signature · 9 lines
Viro's Razor:
	Any race condition, no matter how unlikely, will occur just
	often enough to bite you.

Joel Becker
Consulting Software Developer
Oracle
E-mail: [email protected]
Phone: (650) 506-8127