[PATCH net-next] genetlink: do not parse attributes for families with zero maxattr

Commit c10e6cf85e7d ("net: genetlink: push attrbuf allocation and parsing
to a separate function") moved attribute buffer allocation and attribute
parsing from genl_family_rcv_msg_doit() into a separate function
genl_family_rcv_msg_attrs_parse() which, unlike the previous code, calls
__nlmsg_parse() even if family->maxattr is 0 (i.e. the family does its own
parsing). The parser error is ignored and does not propagate out of
genl_family_rcv_msg_attrs_parse() but an error message ("Unknown attribute
type") is set in extack and if further processing generates no error or
warning, it stays there and is interpreted as a warning by userspace.

Dumpit requests are not affected as genl_family_rcv_msg_dumpit() bypasses
the call of genl_family_rcv_msg_doit() if family->maxattr is zero. Do the
same also in genl_family_rcv_msg_doit().

Fixes: c10e6cf85e7d ("net: genetlink: push attrbuf allocation and parsing to a separate function")
Signed-off-by: Michal Kubecek <redacted>
---
 net/netlink/genetlink.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/net/netlink/genetlink.c b/net/netlink/genetlink.c
index ecc2bd3e73e4..c4bf8830eedf 100644
--- a/net/netlink/genetlink.c
+++ b/net/netlink/genetlink.c

@@ -639,21 +639,23 @@ static int genl_family_rcv_msg_doit(const struct genl_family *family,
 				    const struct genl_ops *ops,
 				    int hdrlen, struct net *net)
 {
-	struct nlattr **attrbuf;
+	struct nlattr **attrbuf = NULL;
 	struct genl_info info;
 	int err;
 
 	if (!ops->doit)
 		return -EOPNOTSUPP;
 
+	if (!family->maxattr)
+		goto no_attrs;
 	attrbuf = genl_family_rcv_msg_attrs_parse(family, nlh, extack,
 						  ops, hdrlen,
 						  GENL_DONT_VALIDATE_STRICT,
-						  family->maxattr &&
 						  family->parallel_ops);
 	if (IS_ERR(attrbuf))
 		return PTR_ERR(attrbuf);
 
+no_attrs:
 	info.snd_seq = nlh->nlmsg_seq;
 	info.snd_portid = NETLINK_CB(skb).portid;
 	info.nlhdr = nlh;

-- 
2.23.0

Wed, Oct 09, 2019 at 06:44:32PM CEST, mkubecek@suse.cz wrote:

Commit c10e6cf85e7d ("net: genetlink: push attrbuf allocation and parsing
to a separate function") moved attribute buffer allocation and attribute
parsing from genl_family_rcv_msg_doit() into a separate function
genl_family_rcv_msg_attrs_parse() which, unlike the previous code, calls
__nlmsg_parse() even if family->maxattr is 0 (i.e. the family does its own
parsing). The parser error is ignored and does not propagate out of
genl_family_rcv_msg_attrs_parse() but an error message ("Unknown attribute
type") is set in extack and if further processing generates no error or
warning, it stays there and is interpreted as a warning by userspace.

Dumpit requests are not affected as genl_family_rcv_msg_dumpit() bypasses
the call of genl_family_rcv_msg_doit() if family->maxattr is zero. Do the
same also in genl_family_rcv_msg_doit().

This is the original code before the changes:

        if (ops->doit == NULL)
                return -EOPNOTSUPP;

        if (family->maxattr && family->parallel_ops) {
                attrbuf = kmalloc_array(family->maxattr + 1,
                                        sizeof(struct nlattr *),
                                        GFP_KERNEL);
                if (attrbuf == NULL)
                        return -ENOMEM;
        } else
                attrbuf = family->attrbuf;

        if (attrbuf) {
                enum netlink_validation validate = NL_VALIDATE_STRICT;

                if (ops->validate & GENL_DONT_VALIDATE_STRICT)
                        validate = NL_VALIDATE_LIBERAL;

                err = __nlmsg_parse(nlh, hdrlen, attrbuf, family->maxattr,
                                    family->policy, validate, extack);
                if (err < 0)
                        goto out;
        }

Looks like the __nlmsg_parse() is called no matter if maxattr if 0 or
not. It is only considered for allocation of attrbuf. This is in-sync
with the current code.

For dumpit, the check was there:

                        if (family->maxattr) {
                                unsigned int validate = NL_VALIDATE_STRICT;

                                if (ops->validate &
                                    GENL_DONT_VALIDATE_DUMP_STRICT)
                                        validate = NL_VALIDATE_LIBERAL;
                                rc = __nla_validate(nlmsg_attrdata(nlh, hdrlen),
                                                    nlmsg_attrlen(nlh, hdrlen),
                                                    family->maxattr,
                                                    family->policy,
                                                    validate, extack);
                                if (rc)
                                        return rc;
                        }

On Thu, Oct 10, 2019 at 10:39:58AM +0200, Jiri Pirko wrote:

Wed, Oct 09, 2019 at 06:44:32PM CEST, mkubecek@suse.cz wrote:

quoted

Commit c10e6cf85e7d ("net: genetlink: push attrbuf allocation and parsing
to a separate function") moved attribute buffer allocation and attribute
parsing from genl_family_rcv_msg_doit() into a separate function
genl_family_rcv_msg_attrs_parse() which, unlike the previous code, calls
__nlmsg_parse() even if family->maxattr is 0 (i.e. the family does its own
parsing). The parser error is ignored and does not propagate out of
genl_family_rcv_msg_attrs_parse() but an error message ("Unknown attribute
type") is set in extack and if further processing generates no error or
warning, it stays there and is interpreted as a warning by userspace.

Dumpit requests are not affected as genl_family_rcv_msg_dumpit() bypasses
the call of genl_family_rcv_msg_doit() if family->maxattr is zero. Do the
same also in genl_family_rcv_msg_doit().

This is the original code before the changes:

        if (ops->doit == NULL)
                return -EOPNOTSUPP;

        if (family->maxattr && family->parallel_ops) {
                attrbuf = kmalloc_array(family->maxattr + 1,
                                        sizeof(struct nlattr *),
                                        GFP_KERNEL);
                if (attrbuf == NULL)
                        return -ENOMEM;
        } else
                attrbuf = family->attrbuf;

        if (attrbuf) {
                enum netlink_validation validate = NL_VALIDATE_STRICT;

                if (ops->validate & GENL_DONT_VALIDATE_STRICT)
                        validate = NL_VALIDATE_LIBERAL;

                err = __nlmsg_parse(nlh, hdrlen, attrbuf, family->maxattr,
                                    family->policy, validate, extack);
                if (err < 0)
                        goto out;
        }

Looks like the __nlmsg_parse() is called no matter if maxattr if 0 or
not. It is only considered for allocation of attrbuf. This is in-sync
with the current code.

If family->maxattr is 0, genl_register_family() sets family->attrbuf to
NULL so that attrbuf is also NULL and the whole "if (attrbuf) { ... }"
block is not executed.

Michal

Thu, Oct 10, 2019 at 11:13:47AM CEST, mkubecek@suse.cz wrote:

On Thu, Oct 10, 2019 at 10:39:58AM +0200, Jiri Pirko wrote:

quoted

Wed, Oct 09, 2019 at 06:44:32PM CEST, mkubecek@suse.cz wrote:

quoted

Commit c10e6cf85e7d ("net: genetlink: push attrbuf allocation and parsing
to a separate function") moved attribute buffer allocation and attribute
parsing from genl_family_rcv_msg_doit() into a separate function
genl_family_rcv_msg_attrs_parse() which, unlike the previous code, calls
__nlmsg_parse() even if family->maxattr is 0 (i.e. the family does its own
parsing). The parser error is ignored and does not propagate out of
genl_family_rcv_msg_attrs_parse() but an error message ("Unknown attribute
type") is set in extack and if further processing generates no error or
warning, it stays there and is interpreted as a warning by userspace.

Dumpit requests are not affected as genl_family_rcv_msg_dumpit() bypasses
the call of genl_family_rcv_msg_doit() if family->maxattr is zero. Do the
same also in genl_family_rcv_msg_doit().

This is the original code before the changes:

        if (ops->doit == NULL)
                return -EOPNOTSUPP;

        if (family->maxattr && family->parallel_ops) {
                attrbuf = kmalloc_array(family->maxattr + 1,
                                        sizeof(struct nlattr *),
                                        GFP_KERNEL);
                if (attrbuf == NULL)
                        return -ENOMEM;
        } else
                attrbuf = family->attrbuf;

        if (attrbuf) {
                enum netlink_validation validate = NL_VALIDATE_STRICT;

                if (ops->validate & GENL_DONT_VALIDATE_STRICT)
                        validate = NL_VALIDATE_LIBERAL;

                err = __nlmsg_parse(nlh, hdrlen, attrbuf, family->maxattr,
                                    family->policy, validate, extack);
                if (err < 0)
                        goto out;
        }

Looks like the __nlmsg_parse() is called no matter if maxattr if 0 or
not. It is only considered for allocation of attrbuf. This is in-sync
with the current code.

If family->maxattr is 0, genl_register_family() sets family->attrbuf to
NULL so that attrbuf is also NULL and the whole "if (attrbuf) { ... }"
block is not executed.

Ah, I missed that. Thanks!

Michal

Wed, Oct 09, 2019 at 06:44:32PM CEST, mkubecek@suse.cz wrote:

Please also adjust genl_family_rcv_msg_attrs_free() call arg
below in this function in the similar way.

	if (IS_ERR(attrbuf))
		return PTR_ERR(attrbuf);

+no_attrs:
	info.snd_seq = nlh->nlmsg_seq;
	info.snd_portid = NETLINK_CB(skb).portid;
	info.nlhdr = nlh;
-- 
2.23.0

On Thu, Oct 10, 2019 at 11:31:53AM +0200, Jiri Pirko wrote:

Wed, Oct 09, 2019 at 06:44:32PM CEST, mkubecek@suse.cz wrote:

quoted

Commit c10e6cf85e7d ("net: genetlink: push attrbuf allocation and parsing
to a separate function") moved attribute buffer allocation and attribute
parsing from genl_family_rcv_msg_doit() into a separate function
genl_family_rcv_msg_attrs_parse() which, unlike the previous code, calls
__nlmsg_parse() even if family->maxattr is 0 (i.e. the family does its own
parsing). The parser error is ignored and does not propagate out of
genl_family_rcv_msg_attrs_parse() but an error message ("Unknown attribute
type") is set in extack and if further processing generates no error or
warning, it stays there and is interpreted as a warning by userspace.

Dumpit requests are not affected as genl_family_rcv_msg_dumpit() bypasses
the call of genl_family_rcv_msg_doit() if family->maxattr is zero. Do the
same also in genl_family_rcv_msg_doit().

Fixes: c10e6cf85e7d ("net: genetlink: push attrbuf allocation and parsing to a separate function")
Signed-off-by: Michal Kubecek <redacted>
---
net/netlink/genetlink.c | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/net/netlink/genetlink.c b/net/netlink/genetlink.c
index ecc2bd3e73e4..c4bf8830eedf 100644
--- a/net/netlink/genetlink.c
+++ b/net/netlink/genetlink.c

@@ -639,21 +639,23 @@ static int genl_family_rcv_msg_doit(const struct genl_family *family,

				    const struct genl_ops *ops,
				    int hdrlen, struct net *net)
{
-	struct nlattr **attrbuf;
+	struct nlattr **attrbuf = NULL;
	struct genl_info info;
	int err;

	if (!ops->doit)
		return -EOPNOTSUPP;

+	if (!family->maxattr)
+		goto no_attrs;
	attrbuf = genl_family_rcv_msg_attrs_parse(family, nlh, extack,
						  ops, hdrlen,
						  GENL_DONT_VALIDATE_STRICT,
-						  family->maxattr &&
						  family->parallel_ops);

Please also adjust genl_family_rcv_msg_attrs_free() call arg
below in this function in the similar way.

Sent v2.

Michal