| Message ID | 20241115003150.733141-1-kuba@kernel.org (mailing list archive) |
|---|---|
| State | Superseded |
| Delegated to: | Netdev Maintainers |
| Headers | show |
| Series | [net,1/2] netlink: fix false positive warning in extack during dumps | expand |
On Thu, Nov 14, 2024 at 04:31:49PM -0800, Jakub Kicinski wrote: > Commit under fixes extended extack reporting to dumps. > It works under normal conditions, because extack errors are > usually reported during ->start() or the first ->dump(), > it's quite rare that the dump starts okay but fails later. > If the dump does fail later, however, the input skb will > already have the initiating message pulled, so checking > if bad attr falls within skb->data will fail. > > Switch the check to using nlh, which is always valid. > > syzbot found a way to hit that scenario by filling up > the receive queue. In this case we initiate a dump > but don't call ->dump() until there is read space for > an skb. > > RIP: 0010:netlink_ack_tlv_fill+0x1a8/0x560 net/netlink/af_netlink.c:2209 > Call Trace: > <TASK> > netlink_dump_done+0x513/0x970 net/netlink/af_netlink.c:2250 > netlink_dump+0x91f/0xe10 net/netlink/af_netlink.c:2351 > netlink_recvmsg+0x6bb/0x11d0 net/netlink/af_netlink.c:1983 > sock_recvmsg_nosec net/socket.c:1051 [inline] > sock_recvmsg+0x22f/0x280 net/socket.c:1073 > __sys_recvfrom+0x246/0x3d0 net/socket.c:2267 > __do_sys_recvfrom net/socket.c:2285 [inline] > __se_sys_recvfrom net/socket.c:2281 [inline] > __x64_sys_recvfrom+0xde/0x100 net/socket.c:2281 > do_syscall_x64 arch/x86/entry/common.c:52 [inline] > do_syscall_64+0xf3/0x230 arch/x86/entry/common.c:83 > entry_SYSCALL_64_after_hwframe+0x77/0x7f > RIP: 0033:0x7ff37dd17a79 > > Reported-by: syzbot+d4373fa8042c06cefa84@syzkaller.appspotmail.com > Fixes: 8af4f60472fc ("netlink: support all extack types in dumps") > Signed-off-by: Jakub Kicinski <kuba@kernel.org> > --- > CC: dsahern@kernel.org > --- > include/net/netlink.h | 13 +++++++++++++ > net/netlink/af_netlink.c | 13 +++++-------- > 2 files changed, 18 insertions(+), 8 deletions(-) > > diff --git a/include/net/netlink.h b/include/net/netlink.h > index db6af207287c..efd906aff22f 100644 > --- a/include/net/netlink.h > +++ b/include/net/netlink.h > @@ -659,6 +659,19 @@ nlmsg_next(const struct nlmsghdr *nlh, int *remaining) > return (struct nlmsghdr *) ((unsigned char *) nlh + totlen); > } > > +/** > + * nlmsg_addr_in_payload - address points to a byte within the message payload > + * @nlh: netlink message header nit: something about @addr should go here. > + * > + * Returns: true if address is within the payload of the message > + */ > +static inline bool nlmsg_addr_in_payload(const struct nlmsghdr *nlh, > + const void *addr) > +{ > + return addr >= nlmsg_data(nlh) && > + addr - (const void *) nlh < nlh->nlmsg_len; > +} > + > /** > * nla_parse - Parse a stream of attributes into a tb buffer > * @tb: destination array with maxtype+1 elements Otherwise, LGTM.
On Fri, 15 Nov 2024 14:00:07 +0000 Simon Horman wrote: > > + * nlmsg_addr_in_payload - address points to a byte within the message payload > > + * @nlh: netlink message header > > nit: something about @addr should go here. Ah, thanks, I just sent a mass conversion, the missing Return statements make the warnings hard to spot. I'll also move the warning into the helper in v2, I hate the double negation in the caller.
diff --git a/include/net/netlink.h b/include/net/netlink.h index db6af207287c..efd906aff22f 100644 --- a/include/net/netlink.h +++ b/include/net/netlink.h @@ -659,6 +659,19 @@ nlmsg_next(const struct nlmsghdr *nlh, int *remaining) return (struct nlmsghdr *) ((unsigned char *) nlh + totlen); } +/** + * nlmsg_addr_in_payload - address points to a byte within the message payload + * @nlh: netlink message header + * + * Returns: true if address is within the payload of the message + */ +static inline bool nlmsg_addr_in_payload(const struct nlmsghdr *nlh, + const void *addr) +{ + return addr >= nlmsg_data(nlh) && + addr - (const void *) nlh < nlh->nlmsg_len; +} + /** * nla_parse - Parse a stream of attributes into a tb buffer * @tb: destination array with maxtype+1 elements diff --git a/net/netlink/af_netlink.c b/net/netlink/af_netlink.c index f84aad420d44..b49c9d745239 100644 --- a/net/netlink/af_netlink.c +++ b/net/netlink/af_netlink.c @@ -2177,8 +2177,7 @@ netlink_ack_tlv_len(struct netlink_sock *nlk, int err, } static void -netlink_ack_tlv_fill(struct sk_buff *in_skb, struct sk_buff *skb, - const struct nlmsghdr *nlh, int err, +netlink_ack_tlv_fill(struct sk_buff *skb, const struct nlmsghdr *nlh, int err, const struct netlink_ext_ack *extack) { if (extack->_msg) @@ -2191,8 +2190,7 @@ netlink_ack_tlv_fill(struct sk_buff *in_skb, struct sk_buff *skb, return; if (extack->bad_attr && - !WARN_ON((u8 *)extack->bad_attr < in_skb->data || - (u8 *)extack->bad_attr >= in_skb->data + in_skb->len)) + !WARN_ON(!nlmsg_addr_in_payload(nlh, extack->bad_attr))) WARN_ON(nla_put_u32(skb, NLMSGERR_ATTR_OFFS, (u8 *)extack->bad_attr - (const u8 *)nlh)); if (extack->policy) @@ -2202,8 +2200,7 @@ netlink_ack_tlv_fill(struct sk_buff *in_skb, struct sk_buff *skb, WARN_ON(nla_put_u32(skb, NLMSGERR_ATTR_MISS_TYPE, extack->miss_type)); if (extack->miss_nest && - !WARN_ON((u8 *)extack->miss_nest < in_skb->data || - (u8 *)extack->miss_nest > in_skb->data + in_skb->len)) + !WARN_ON(!nlmsg_addr_in_payload(nlh, extack->miss_nest))) WARN_ON(nla_put_u32(skb, NLMSGERR_ATTR_MISS_NEST, (u8 *)extack->miss_nest - (const u8 *)nlh)); } @@ -2232,7 +2229,7 @@ static int netlink_dump_done(struct netlink_sock *nlk, struct sk_buff *skb, if (extack_len) { nlh->nlmsg_flags |= NLM_F_ACK_TLVS; if (skb_tailroom(skb) >= extack_len) { - netlink_ack_tlv_fill(cb->skb, skb, cb->nlh, + netlink_ack_tlv_fill(skb, cb->nlh, nlk->dump_done_errno, extack); nlmsg_end(skb, nlh); } @@ -2491,7 +2488,7 @@ void netlink_ack(struct sk_buff *in_skb, struct nlmsghdr *nlh, int err, } if (tlvlen) - netlink_ack_tlv_fill(in_skb, skb, nlh, err, extack); + netlink_ack_tlv_fill(skb, nlh, err, extack); nlmsg_end(skb, rep);
Commit under fixes extended extack reporting to dumps. It works under normal conditions, because extack errors are usually reported during ->start() or the first ->dump(), it's quite rare that the dump starts okay but fails later. If the dump does fail later, however, the input skb will already have the initiating message pulled, so checking if bad attr falls within skb->data will fail. Switch the check to using nlh, which is always valid. syzbot found a way to hit that scenario by filling up the receive queue. In this case we initiate a dump but don't call ->dump() until there is read space for an skb. RIP: 0010:netlink_ack_tlv_fill+0x1a8/0x560 net/netlink/af_netlink.c:2209 Call Trace: <TASK> netlink_dump_done+0x513/0x970 net/netlink/af_netlink.c:2250 netlink_dump+0x91f/0xe10 net/netlink/af_netlink.c:2351 netlink_recvmsg+0x6bb/0x11d0 net/netlink/af_netlink.c:1983 sock_recvmsg_nosec net/socket.c:1051 [inline] sock_recvmsg+0x22f/0x280 net/socket.c:1073 __sys_recvfrom+0x246/0x3d0 net/socket.c:2267 __do_sys_recvfrom net/socket.c:2285 [inline] __se_sys_recvfrom net/socket.c:2281 [inline] __x64_sys_recvfrom+0xde/0x100 net/socket.c:2281 do_syscall_x64 arch/x86/entry/common.c:52 [inline] do_syscall_64+0xf3/0x230 arch/x86/entry/common.c:83 entry_SYSCALL_64_after_hwframe+0x77/0x7f RIP: 0033:0x7ff37dd17a79 Reported-by: syzbot+d4373fa8042c06cefa84@syzkaller.appspotmail.com Fixes: 8af4f60472fc ("netlink: support all extack types in dumps") Signed-off-by: Jakub Kicinski <kuba@kernel.org> --- CC: dsahern@kernel.org --- include/net/netlink.h | 13 +++++++++++++ net/netlink/af_netlink.c | 13 +++++-------- 2 files changed, 18 insertions(+), 8 deletions(-)