Skip to contentSkip to conversation

genetlink custom attribute type

RFC In review Johannes Berg 3 participants 6 messages Sep 14, 2006 – Sep 25, 2006

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

RFC 1/1 genetlink custom attribute type

This patch adds an NLA_CUSTOM_CHECK type for netlink attributes in order to be able to centrally define new attribute structures instead of having to check these special types in each function that uses such an attribute.

nl80211 will benefit from this because it needs an information element attribute with a fixed type|length|value structure defined in IEEE 802.11 which ought to be checked when passing an IE from userspace.

include/net/netlink.h +5 -0
@@ -159,6 +159,7 @@ enum {
 	NLA_MSECS,
 	NLA_NESTED,
 	NLA_NUL_STRING,
+	NLA_CUSTOM_CHECK,
 	__NLA_TYPE_MAX,
 };
 
@@ -176,6 +177,8 @@ enum {
  *    NLA_STRING           Maximum length of string
  *    NLA_NUL_STRING       Maximum length of string (excluding NUL)
  *    NLA_FLAG             Unused
+ *    NLA_CUSTOM_CHECK     Unused, check function must be assigned,
+ *                         it must return 0 if the attribute is valid
  *    All other            Exact length of attribute payload
  *
  * Example:
@@ -183,11 +186,13 @@ enum {
  * 	[ATTR_FOO] = { .type = NLA_U16 },
  *	[ATTR_BAR] = { .type = NLA_STRING, len = BARSIZ },
  *	[ATTR_BAZ] = { .len = sizeof(struct mystruct) },
+ *	[ATTR_CST] = { .type = NLA_CUSTOM_CHECK, .check = my_check },
  * };
  */
 struct nla_policy {
 	u16		type;
 	u16		len;
+	int		(*check)(struct nlattr *nla);
 };
 
 extern void		netlink_run_queue(struct sock *sk, unsigned int *qlen,
net/netlink/attr.c +3 -0
@@ -67,6 +67,9 @@ static int validate_nla(struct nlattr *n
 		}
 		break;
 
+	case NLA_CUSTOM_CHECK:
+		BUG_ON(!pt->check);
+		return pt->check(nla);
 	default:
 		if (pt->len)
 			minlen = pt->len;
#1 lore raw

* Johannes Berg <[email protected]> 2006-09-14 09:44

Quoted 9 lines

This patch adds an NLA_CUSTOM_CHECK type for netlink attributes in order to be able to centrally define new attribute structures instead of having to check these special types in each function that uses such an attribute.

nl80211 will benefit from this because it needs an information element attribute with a fixed type|length|value structure defined in IEEE 802.11 which ought to be checked when passing an IE from userspace.

Looks good, we have to watch the size of struct nla_policy though. This bumps the size from 4 bytes to 16 bytes on 64bit architectures which might become a problem since we always use ATTR_MAX sized arrays.

#2 lore raw
Quoted 4 lines · Thomas Graf

Looks good, we have to watch the size of struct nla_policy though. This bumps the size from 4 bytes to 16 bytes on 64bit architectures which might become a problem since we always use ATTR_MAX sized arrays.

Yes, I'm aware of that, but I couldn't think of a way to handle it well.

I thought about using a second array containing just the check functions, and then (ab)using `len' to index into it but that didn't seem clean enough.

Can you think of something better?

johannes

#3 lore raw

* Johannes Berg <[email protected]> 2006-09-14 11:21

Quoted 7 lines · Thomas Graf
Quoted 5 lines

Looks good, we have to watch the size of struct nla_policy though. This bumps the size from 4 bytes to 16 bytes on 64bit architectures which might become a problem since we always use ATTR_MAX sized arrays.

Quoted 5 lines

Yes, I'm aware of that, but I couldn't think of a way to handle it well.

I thought about using a second array containing just the check functions, and then (ab)using `len' to index into it but that didn't seem clean enough.

I agree, I talked about this with various people and some ideas came up.

Always use function pointers to define the validation policy, i.e. there would be nla_validate_u32() etc. The problem with this approach is that for every string attribute with different length a separate validation function is required which simply adds to code what it saved from text. It also makes exporting the policy to userspace impossible.

Following up on your idea, we could save a bit by using ERR_PTR() to store both the callback pointer and type/len tuple in the very same pointer but that gets very ugly.

I think its best to use your patch for now and see where this leads to.

#4 lore raw
Quoted 1 line · Thomas Graf

I think its best to use your patch for now and see where this leads to.

Alright, should I repost with a proper [PATCH] subject or is it good to take as-is? :)

johannes

maintainer
#5 lore raw

From: Johannes Berg <[email protected]>
Date: Mon, 25 Sep 2006 16:58:28 +0200

Quoted 6 lines · Thomas Graf
Quoted 2 lines

I think its best to use your patch for now and see where this leads to.

Alright, should I repost with a proper [PATCH] subject or is it good to take as-is? :)

Please resend it as I've deleted the copy that was in my inbox.