commit 4356e9f841 upstream.
We've had issues with gcc and 'asm goto' before, and we created a
'asm_volatile_goto()' macro for that in the past: see commits
3f0116c323 ("compiler/gcc4: Add quirk for 'asm goto' miscompilation
bug") and a9f180345f ("compiler/gcc4: Make quirk for
asm_volatile_goto() unconditional").
Then, much later, we ended up removing the workaround in commit
43c249ea0b ("compiler-gcc.h: remove ancient workaround for gcc PR
58670") because we no longer supported building the kernel with the
affected gcc versions, but we left the macro uses around.
Now, Sean Christopherson reports a new version of a very similar
problem, which is fixed by re-applying that ancient workaround. But the
problem in question is limited to only the 'asm goto with outputs'
cases, so instead of re-introducing the old workaround as-is, let's
rename and limit the workaround to just that much less common case.
It looks like there are at least two separate issues that all hit in
this area:
(a) some versions of gcc don't mark the asm goto as 'volatile' when it
has outputs:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98619https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110420
which is easy to work around by just adding the 'volatile' by hand.
(b) Internal compiler errors:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110422
which are worked around by adding the extra empty 'asm' as a
barrier, as in the original workaround.
but the problem Sean sees may be a third thing since it involves bad
code generation (not an ICE) even with the manually added 'volatile'.
but the same old workaround works for this case, even if this feels a
bit like voodoo programming and may only be hiding the issue.
Reported-and-tested-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/all/20240208220604.140859-1-seanjc@google.com/
Cc: Nick Desaulniers <ndesaulniers@google.com>
Cc: Uros Bizjak <ubizjak@gmail.com>
Cc: Jakub Jelinek <jakub@redhat.com>
Cc: Andrew Pinski <quic_apinski@quicinc.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 60c0c230c6 upstream.
rbtree lazy gc on insert might collect an end interval element that has
been just added in this transactions, skip end interval elements that
are not yet active.
Fixes: f718863aca ("netfilter: nft_set_rbtree: fix overlap expiration walk")
Cc: stable@vger.kernel.org
Reported-by: lonial con <kongln9170@gmail.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit 5a8cdf6fd8 ]
use ->scratch for both avx2 and the generic implementation.
After previous change the scratch->map member is always aligned properly
for AVX2, so we can just use scratch->map in AVX2 too.
The alignoff delta is stored in the scratchpad so we can reconstruct
the correct address to free the area again.
Fixes: 7400b06396 ("nft_set_pipapo: Introduce AVX2-based lookup implementation")
Reviewed-by: Stefano Brivio <sbrivio@redhat.com>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 47b1c03c3c ]
After next patch simple kfree() is not enough anymore, so add
a helper for it.
Reviewed-by: Stefano Brivio <sbrivio@redhat.com>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Stable-dep-of: 5a8cdf6fd8 ("netfilter: nft_set_pipapo: remove scratch_aligned pointer")
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 76313d1a4a ]
Pipapo needs a scratchpad area to keep state during matching.
This state can be large and thus cannot reside on stack.
Each set preallocates percpu areas for this.
On each match stage, one scratchpad half starts with all-zero and the other
is inited to all-ones.
At the end of each stage, the half that starts with all-ones is
always zero. Before next field is tested, pointers to the two halves
are swapped, i.e. resmap pointer turns into fill pointer and vice versa.
After the last field has been processed, pipapo stashes the
index toggle in a percpu variable, with assumption that next packet
will start with the all-zero half and sets all bits in the other to 1.
This isn't reliable.
There can be multiple sets and we can't be sure that the upper
and lower half of all set scratch map is always in sync (lookups
can be conditional), so one set might have swapped, but other might
not have been queried.
Thus we need to keep the index per-set-and-cpu, just like the
scratchpad.
Note that this bug fix is incomplete, there is a related issue.
avx2 and normal implementation might use slightly different areas of the
map array space due to the avx2 alignment requirements, so
m->scratch (generic/fallback implementation) and ->scratch_aligned
(avx) may partially overlap. scratch and scratch_aligned are not distinct
objects, the latter is just the aligned address of the former.
After this change, write to scratch_align->map_index may write to
scratch->map, so this issue becomes more prominent, we can set to 1
a bit in the supposedly-all-zero area of scratch->map[].
A followup patch will remove the scratch_aligned and makes generic and
avx code use the same (aligned) area.
Its done in a separate change to ease review.
Fixes: 3c4287f620 ("nf_tables: Add set type for arbitrary concatenation of ranges")
Reviewed-by: Stefano Brivio <sbrivio@redhat.com>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit f82777e8ce ]
Only override userspace verdict if the ct hook returns something
other than ACCEPT.
Else, this replaces NF_REPEAT (run all hooks again) with NF_ACCEPT
(move to next hook).
Fixes: 6291b3a67a ("netfilter: conntrack: convert nf_conntrack_update to netfilter verdicts")
Reported-by: l.6diay@passmail.com
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 7395dfacff ]
Add a timestamp field at the beginning of the transaction, store it
in the nftables per-netns area.
Update set backend .insert, .deactivate and sync gc path to use the
timestamp, this avoids that an element expires while control plane
transaction is still unfinished.
.lookup and .update, which are used from packet path, still use the
current time to check if the element has expired. And .get path and dump
also since this runs lockless under rcu read size lock. Then, there is
async gc which also needs to check the current time since it runs
asynchronously from a workqueue.
Fixes: c3e1b005ed ("netfilter: nf_tables: add set element timeout support")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 38ed1c7062 ]
Direction attribute is ignored, reject it in case this ever needs to be
supported
Fixes: 3087c3f7c2 ("netfilter: nft_ct: Add ct id support")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit ab0beafd52 ]
This has slipped through when reducing memory footprint for set
elements, remove it.
Fixes: 9dad402b89 ("netfilter: nf_tables: expose opaque set element as struct nft_elem_priv")
Reported-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit d694b75489 ]
xt_check_{match,target} expects u16, but NFTA_RULE_COMPAT_PROTO is u32.
NLA_POLICY_MAX(NLA_BE32, 65535) cannot be used because .max in
nla_policy is s16, see 3e48be05f3 ("netlink: add attribute range
validation to policy").
Fixes: 0ca743a559 ("netfilter: nf_tables: add compatibility layer for x_tables")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 292781c3c5 ]
Flag (1 << 0) is ignored is set, never used, reject it it with EINVAL
instead.
Fixes: 0ca743a559 ("netfilter: nf_tables: add compatibility layer for x_tables")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 8059918a13 ]
- Disallow families other than NFPROTO_{IPV4,IPV6,INET}.
- Disallow layer 4 protocol with no ports, since destination port is a
mandatory attribute for this object.
Fixes: 857b46027d ("netfilter: nft_ct: add ct expectations support")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 259eb32971 ]
Module reference is bumped for each user, this should not ever happen.
But BUG_ON check should use rcu_access_pointer() instead.
If this ever happens, do WARN_ON_ONCE() instead of BUG_ON() and
consolidate pointer check under the rcu read side lock section.
Fixes: fab4085f4e ("netfilter: log: nf_log_packet() as real unified interface")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 776d451648 ]
Bail out on using the tunnel dst template from other than netdev family.
Add the infrastructure to check for the family in objects.
Fixes: af308b94a2 ("netfilter: nf_tables: add tunnel support")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit fb366fc754 ]
commit c7aab4f170 ("netfilter: nf_conntrack_tcp: re-init for syn packets
only") introduces a bug where SYNs in ORIGINAL direction on reused 5-tuple
result in incorrect window scale negotiation. This commit merged the SYN
re-initialization and simultaneous open or SYN retransmits cases. Merging
this block added the logic in tcp_init_sender() that performed window scale
negotiation to the retransmitted syn case. Previously. this would only
result in updating the sender's scale and flags. After the merge the
additional logic results in improperly clearing the scale in ORIGINAL
direction before any packets in the REPLY direction are received. This
results in packets incorrectly being marked invalid for being
out-of-window.
This can be reproduced with the following trace:
Packet Sequence:
> Flags [S], seq 1687765604, win 62727, options [.. wscale 7], length 0
> Flags [S], seq 1944817196, win 62727, options [.. wscale 7], length 0
In order to fix the issue, only evaluate window negotiation for packets
in the REPLY direction. This was tested with simultaneous open, fast
open, and the above reproduction.
Fixes: c7aab4f170 ("netfilter: nf_conntrack_tcp: re-init for syn packets only")
Signed-off-by: Ryan Schaefer <ryanschf@amazon.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
commit f342de4e2f upstream.
This reverts commit e0abdadcc6.
core.c:nf_hook_slow assumes that the upper 16 bits of NF_DROP
verdicts contain a valid errno, i.e. -EPERM, -EHOSTUNREACH or similar,
or 0.
Due to the reverted commit, its possible to provide a positive
value, e.g. NF_ACCEPT (1), which results in use-after-free.
Its not clear to me why this commit was made.
NF_QUEUE is not used by nftables; "queue" rules in nftables
will result in use of "nft_queue" expression.
If we later need to allow specifiying errno values from userspace
(do not know why), this has to call NF_DROP_GETERR and check that
"err <= 0" holds true.
Fixes: e0abdadcc6 ("netfilter: nf_tables: accept QUEUE/DROP verdict parameters")
Cc: stable@vger.kernel.org
Reported-by: Notselwyn <notselwyn@pwning.tech>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 01acb2e866 upstream.
Remove netdevice from inet/ingress basechain in case NETDEV_UNREGISTER
event is reported, otherwise a stale reference to netdevice remains in
the hook list.
Fixes: 60a3815da7 ("netfilter: add inet ingress support")
Cc: stable@vger.kernel.org
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit b462579b2b ]
nftables has two types of sets/maps, one where userspace defines the
name, and anonymous sets/maps, where userspace defines a template name.
For the latter, kernel requires presence of exactly one "%d".
nftables uses "__set%d" and "__map%d" for this. The kernel will
expand the format specifier and replaces it with the smallest unused
number.
As-is, userspace could define a template name that allows to move
the set name past the 256 bytes upperlimit (post-expansion).
I don't see how this could be a problem, but I would prefer if userspace
cannot do this, so add a limit of 16 bytes for the '%d' template name.
16 bytes is the old total upper limit for set names that existed when
nf_tables was merged initially.
Fixes: 387454901b ("netfilter: nf_tables: Allow set names of up to 255 chars")
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit c9d9eb9c53 ]
Reject bogus configs where internal token counter wraps around.
This only occurs with very very large requests, such as 17gbyte/s.
Its better to reject this rather than having incorrect ratelimit.
Fixes: d2168e849e ("netfilter: nft_limit: add per-byte limiting")
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit d6938c1c76 ]
Inside decrement_ttl() upon discovering that the packet ttl has exceeded,
__IP_INC_STATS and __IP6_INC_STATS macros can be called from preemptible
context having the following backtrace:
check_preemption_disabled: 48 callbacks suppressed
BUG: using __this_cpu_add() in preemptible [00000000] code: curl/1177
caller is decrement_ttl+0x217/0x830
CPU: 5 PID: 1177 Comm: curl Not tainted 6.7.0+ #34
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 04/01/2014
Call Trace:
<TASK>
dump_stack_lvl+0xbd/0xe0
check_preemption_disabled+0xd1/0xe0
decrement_ttl+0x217/0x830
__ip_vs_get_out_rt+0x4e0/0x1ef0
ip_vs_nat_xmit+0x205/0xcd0
ip_vs_in_hook+0x9b1/0x26a0
nf_hook_slow+0xc2/0x210
nf_hook+0x1fb/0x770
__ip_local_out+0x33b/0x640
ip_local_out+0x2a/0x490
__ip_queue_xmit+0x990/0x1d10
__tcp_transmit_skb+0x288b/0x3d10
tcp_connect+0x3466/0x5180
tcp_v4_connect+0x1535/0x1bb0
__inet_stream_connect+0x40d/0x1040
inet_stream_connect+0x57/0xa0
__sys_connect_file+0x162/0x1a0
__sys_connect+0x137/0x160
__x64_sys_connect+0x72/0xb0
do_syscall_64+0x6f/0x140
entry_SYSCALL_64_after_hwframe+0x6e/0x76
RIP: 0033:0x7fe6dbbc34e0
Use the corresponding preemption-aware variants: IP_INC_STATS and
IP6_INC_STATS.
Found by Linux Verification Center (linuxtesting.org).
Fixes: 8d8e20e2d7 ("ipvs: Decrement ttl")
Signed-off-by: Fedor Pchelkin <pchelkin@ispras.ru>
Acked-by: Julian Anastasov <ja@ssi.bg>
Acked-by: Simon Horman <horms@kernel.org>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 113661e074 ]
It is still possible to set on the NFT_SET_CONCAT flag by specifying a
set size and no field description, report EINVAL in such case.
Fixes: 1b6345d416 ("netfilter: nf_tables: check NFT_SET_CONCAT flag if field_count is specified")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 6b1ca88e4b ]
Delete from packet path relies on the garbage collector to purge
elements with NFT_SET_ELEM_DEAD_BIT on.
Skip these dead elements from nf_tables_dump_setelem() path, I very
rarely see tests/shell/testcases/maps/typeof_maps_add_delete reports
[DUMP FAILED] showing a mismatch in the expected output with an element
that should not be there.
If the netlink dump happens before GC worker run, it might show dead
elements in the ruleset listing.
nft_rhash_get() already skips dead elements in nft_rhash_cmp(),
therefore, it already does not show the element when getting a single
element via netlink control plane.
Fixes: 5f68718b34 ("netfilter: nf_tables: GC transaction API to avoid race with control plane")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 3ce67e3793 ]
The set description provides the size of each field in the set whose sum
should not mismatch the set key length, bail out otherwise.
I did not manage to crash nft_set_pipapo with mismatch fields and set key
length so far, but this is UB which must be disallowed.
Fixes: f3a2181e16 ("netfilter: nf_tables: Support for sets with multiple ranged fields")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit a54e721970 ]
This is a preparation patch for replacing physindev with physinif on
nf_bridge_info structure. We will use dev_get_by_index_rcu to resolve
device, when needed, and it requires net to be available.
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Stable-dep-of: 9874808878 ("netfilter: bridge: replace physindev with physinif in nf_bridge_info")
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit aeaa44075f ]
We don't really need nf_bridge variable here. And nf_bridge_info_exists
is better replacement for nf_bridge_info_get in case we are only
checking for existence.
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Stable-dep-of: 9874808878 ("netfilter: bridge: replace physindev with physinif in nf_bridge_info")
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit c3f9fd54cd ]
We don't use physindev in __build_packet_message except for getting
physinif from it. So let's switch to nf_bridge_get_physinif to get what
we want directly.
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Stable-dep-of: 9874808878 ("netfilter: bridge: replace physindev with physinif in nf_bridge_info")
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 91a139cee1 ]
Bail out if userspace provides unsupported flags, otherwise future
extensions to the limit expression will be silently ignored by the
kernel.
Fixes: c7862a5f0d ("netfilter: nft_limit: allow to invert matching criteria")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
commit b1db244ffd upstream.
When deactivating the catch-all set element, check the state in the next
generation that represents this transaction.
This bug uncovered after the recent removal of the element busy mark
a2dd0233cb ("netfilter: nf_tables: remove busy mark and gc batch API").
Fixes: aaa31047a6 ("netfilter: nftables: add catch-all set element support")
Cc: stable@vger.kernel.org
Reported-by: lonial con <kongln9170@gmail.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit aaba7ddc85 ]
Parse netlink attribute containing the chain type in this update, to
bail out if this is different from the existing type.
Otherwise, it is possible to define a chain with the same name, hook and
priority but different type, which is silently ignored.
Fixes: 96518518cc ("netfilter: add nftables")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 08e4c8c591 ]
If a transaction is aborted, we should mark the to-be-released NEWSET dead,
just like commit path does for DEL and DESTROYSET commands.
In both cases all remaining elements will be released via
set->ops->destroy().
The existing abort code does NOT post the actual release to the work queue.
Also the entire __nf_tables_abort() function is wrapped in gc_seq
begin/end pair.
Therefore, async gc worker will never try to release the pending set
elements, as gc sequence is always stale.
It might be possible to speed up transaction aborts via work queue too,
this would result in a race and a possible use-after-free.
So fix this before it becomes an issue.
Fixes: 5f68718b34 ("netfilter: nf_tables: GC transaction API to avoid race with control plane")
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
In the init path, nft_data_init() bumps the chain reference counter,
decrement it on error by following the error path which calls
nft_data_release() to restore it.
Fixes: 4bedf9eee0 ("netfilter: nf_tables: fix chain binding transaction logic")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
This fixes openvswitch's handling of nat packets in the related state.
In nf_ct_nat_execute(), which is called from nf_ct_nat(), ICMP/ICMPv6
packets in the IP_CT_RELATED or IP_CT_RELATED_REPLY state, which have
not been dropped, will follow the goto, however the placement of the
goto label means that updating the action bit field will be bypassed.
This causes ovs_nat_update_key() to not be called from ovs_ct_nat()
which means the openvswitch match key for the ICMP/ICMPv6 packet is not
updated and the pre-nat value will be retained for the key, which will
result in the wrong openflow rule being matched for that packet.
Move the goto label above where the action bit field is being set so
that it is updated in all cases where the packet is accepted.
Fixes: ebddb14049 ("net: move the nat function to nf_nat_ovs for ovs and tc")
Signed-off-by: Brad Cowie <brad@faucet.nz>
Reviewed-by: Simon Horman <horms@kernel.org>
Acked-by: Xin Long <lucien.xin@gmail.com>
Acked-by: Aaron Conole <aconole@redhat.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
NFT_MSG_DELSET deactivates all elements in the set, skip
set->ops->commit() to avoid the unnecessary clone (for the pipapo case)
as well as the sync GC cycle, which could deactivate again expired
elements in such set.
Fixes: 5f68718b34 ("netfilter: nf_tables: GC transaction API to avoid race with control plane")
Reported-by: Kevin Rich <kevinrich1337@gmail.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Before this patch, transport offset (pkt->thoff) provides an offset
relative to the network header. This is fine for the inet families
because skb->data points to the network header in such case. However,
from netdev/egress, skb->data points to the mac header (if available),
thus, pkt->thoff is missing the mac header length.
Add skb_network_offset() to the transport offset (pkt->thoff) for
netdev, so transport header mangling works as expected. Adjust payload
fast eval function to use skb->data now that pkt->thoff provides an
absolute offset. This explains why users report that matching on
egress/netdev works but payload mangling does not.
This patch implicitly fixes payload mangling for IPv4 packets in
netdev/egress given skb_store_bits() requires an offset from skb->data
to reach the transport header.
I suspect that nft_exthdr and the trace infra were also broken from
netdev/egress because they also take skb->data as start, and pkt->thoff
was not correct.
Note that IPv6 is fine because ipv6_find_hdr() already provides a
transport offset starting from skb->data, which includes
skb_network_offset().
The bridge family also uses nft_set_pktinfo_ipv4_validate(), but there
skb_network_offset() is zero, so the update in this patch does not alter
the existing behaviour.
Fixes: 42df6e1d22 ("netfilter: Introduce egress hook")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
A concurrently running sock_orphan() may NULL the sk_socket pointer in
between check and deref. Follow other users (like nft_meta.c for
instance) and acquire sk_callback_lock before dereferencing sk_socket.
Fixes: 0265ab44ba ("[NETFILTER]: merge ipt_owner/ip6t_owner in xt_owner")
Reported-by: Jann Horn <jannh@google.com>
Signed-off-by: Phil Sutter <phil@nwl.cc>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Validate table family when looking up for it via NFTA_TABLE_HANDLE.
Fixes: 3ecbfd65f5 ("netfilter: nf_tables: allocate handle and delete objects via handle")
Reported-by: Xingyuan Mo <hdthky0@gmail.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
If dynset expressions provided by userspace is larger than the declared
set expressions, then bail out.
Fixes: 48b0ae046e ("netfilter: nftables: netlink support for several set element expressions")
Reported-by: Xingyuan Mo <hdthky0@gmail.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Maze reports "tcp option fastopen exists" fails to match on
OpenWrt 22.03.5, r20134-5f15225c1e (5.10.176) router.
"tcp option fastopen exists" translates to:
inet
[ exthdr load tcpopt 1b @ 34 + 0 present => reg 1 ]
[ cmp eq reg 1 0x00000001 ]
.. but existing nft userspace generates a 1-byte compare.
On LSB (x86), "*reg32 = 1" is identical to nft_reg_store8(reg32, 1), but
not on MSB, which will place the 1 last. IOW, on bigendian aches the cmp8
is awalys false.
Make sure we store this in a consistent fashion, so existing userspace
will also work on MSB (bigendian).
Regardless of this patch we can also change nft userspace to generate
'reg32 == 0' and 'reg32 != 0' instead of u8 == 0 // u8 == 1 when
adding 'option x missing/exists' expressions as well.
Fixes: 3c1fece881 ("netfilter: nft_exthdr: Allow checking TCP option presence, too")
Fixes: b9f9a485fb ("netfilter: nft_exthdr: add boolean DCCP option matching")
Fixes: 055c4b34b9 ("netfilter: nft_fib: Support existence check")
Reported-by: Maciej Żenczykowski <zenczykowski@gmail.com>
Closes: https://lore.kernel.org/netfilter-devel/CAHo-OozyEqHUjL2-ntATzeZOiuftLWZ_HU6TOM_js4qLfDEAJg@mail.gmail.com/
Signed-off-by: Florian Westphal <fw@strlen.de>
Acked-by: Phil Sutter <phil@nwl.cc>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Otherwise set elements can be deactivated twice which will cause a crash.
Reported-by: Xingyuan Mo <hdthky0@gmail.com>
Fixes: 3c4287f620 ("nf_tables: Add set type for arbitrary concatenation of ranges")
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
list_for_each_entry_safe() does not work for the async case which runs
under RCU, therefore, split GC logic for catchall in two functions
instead, one for each of the sync and async GC variants.
The catchall sync GC variant never sees a _DEAD bit set on ever, thus,
this handling is removed in such case, moreover, allocate GC sync batch
via GFP_KERNEL.
Fixes: 93995bf4af ("netfilter: nf_tables: remove catchall element in GC sync path")
Reported-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Linkui Xiao reported that there's a race condition when ipset swap and destroy is
called, which can lead to crash in add/del/test element operations. Swap then
destroy are usual operations to replace a set with another one in a production
system. The issue can in some cases be reproduced with the script:
ipset create hash_ip1 hash:net family inet hashsize 1024 maxelem 1048576
ipset add hash_ip1 172.20.0.0/16
ipset add hash_ip1 192.168.0.0/16
iptables -A INPUT -m set --match-set hash_ip1 src -j ACCEPT
while [ 1 ]
do
# ... Ongoing traffic...
ipset create hash_ip2 hash:net family inet hashsize 1024 maxelem 1048576
ipset add hash_ip2 172.20.0.0/16
ipset swap hash_ip1 hash_ip2
ipset destroy hash_ip2
sleep 0.05
done
In the race case the possible order of the operations are
CPU0 CPU1
ip_set_test
ipset swap hash_ip1 hash_ip2
ipset destroy hash_ip2
hash_net_kadt
Swap replaces hash_ip1 with hash_ip2 and then destroy removes hash_ip2 which
is the original hash_ip1. ip_set_test was called on hash_ip1 and because destroy
removed it, hash_net_kadt crashes.
The fix is to force ip_set_swap() to wait for all readers to finish accessing the
old set pointers by calling synchronize_rcu().
The first version of the patch was written by Linkui Xiao <xiaolinkui@kylinos.cn>.
v2: synchronize_rcu() is moved into ip_set_swap() in order not to burden
ip_set_destroy() unnecessarily when all sets are destroyed.
v3: Florian Westphal pointed out that all netfilter hooks run with rcu_read_lock() held
and em_ipset.c wraps the entire ip_set_test() in rcu read lock/unlock pair.
So there's no need to extend the rcu read locked area in ipset itself.
Closes: https://lore.kernel.org/all/69e7963b-e7f8-3ad0-210-7b86eebf7f78@netfilter.org/
Reported by: Linkui Xiao <xiaolinkui@kylinos.cn>
Signed-off-by: Jozsef Kadlecsik <kadlec@netfilter.org>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
destroy element command bogusly reports ENOENT in case a set element
does not exist. ENOENT errors are skipped, however, err is still set
and propagated to userspace.
# nft destroy element ip raw BLACKLIST { 1.2.3.4 }
Error: Could not process rule: No such file or directory
destroy element ip raw BLACKLIST { 1.2.3.4 }
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Fixes: f80a612dd7 ("netfilter: nf_tables: add support to destroy operation")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
The problem is in nft_byteorder_eval() where we are iterating through a
loop and writing to dst[0], dst[1], dst[2] and so on... On each
iteration we are writing 8 bytes. But dst[] is an array of u32 so each
element only has space for 4 bytes. That means that every iteration
overwrites part of the previous element.
I spotted this bug while reviewing commit caf3ef7468 ("netfilter:
nf_tables: prevent OOB access in nft_byteorder_eval") which is a related
issue. I think that the reason we have not detected this bug in testing
is that most of time we only write one element.
Fixes: ce1e7989d9 ("netfilter: nft_byteorder: provide 64bit le/be conversion")
Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
The code that uses nft_net has been removed, and the nft_pernet function
is merely obtaining a reference to shared data through the net pointer.
The content of the net pointer is not modified or changed, so both of
them should be removed.
silence the warning:
net/netfilter/nft_set_rbtree.c:627:26: warning: variable ‘nft_net’ set but not used
Reported-by: Abaci Robot <abaci@linux.alibaba.com>
Closes: https://bugzilla.openanolis.cn/show_bug.cgi?id=7103
Signed-off-by: Yang Li <yang.lee@linux.alibaba.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
-----BEGIN PGP SIGNATURE-----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=8Kr3
-----END PGP SIGNATURE-----
Merge tag 'nf-23-11-08' of git://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf
Pablo Neira Ayuso says:
====================
Netfilter fixes for net
The following patchset contains Netfilter fixes for net:
1) Add missing netfilter modules description to fix W=1, from Florian Westphal.
2) Fix catch-all element GC with timeout when use with the pipapo set
backend, this remained broken since I tried to fix it this summer,
then another attempt to fix it recently.
3) Add missing IPVS modules descriptions to fix W=1, also from Florian.
4) xt_recent allocated a too small buffer to store an IPv4-mapped IPv6
address which can be parsed by in6_pton(), from Maciej Zenczykowski.
Broken for many releases.
5) Skip IPv4-mapped IPv6, IPv4-compat IPv6, site/link local scoped IPv6
addressses to set up IPv6 NAT redirect, also from Florian. This is
broken since 2012.
* tag 'nf-23-11-08' of git://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf:
netfilter: nat: fix ipv6 nat redirect with mapped and scoped addresses
netfilter: xt_recent: fix (increase) ipv6 literal buffer length
ipvs: add missing module descriptions
netfilter: nf_tables: remove catchall element in GC sync path
netfilter: add missing module descriptions
====================
Link: https://lore.kernel.org/r/20231108155802.84617-1-pablo@netfilter.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The ipv6 redirect target was derived from the ipv4 one, i.e. its
identical to a 'dnat' with the first (primary) address assigned to the
network interface. The code has been moved around to make it usable
from nf_tables too, but its still the same as it was back when this
was added in 2012.
IPv6, however, has different types of addresses, if the 'wrong' address
comes first the redirection does not work.
In Daniels case, the addresses are:
inet6 ::ffff:192 ...
inet6 2a01: ...
... so the function attempts to redirect to the mapped address.
Add more checks before the address is deemed correct:
1. If the packets' daddr is scoped, search for a scoped address too
2. skip tentative addresses
3. skip mapped addresses
Use the first address that appears to match our needs.
Reported-by: Daniel Huhardeaux <tech@tootai.net>
Closes: https://lore.kernel.org/netfilter/71be06b8-6aa0-4cf9-9e0b-e2839b01b22f@tootai.net/
Fixes: 115e23ac78 ("netfilter: ip6tables: add REDIRECT target")
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
