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>
Referenced commit doesn't always set iifidx when offloading the flow to
hardware. Fix the following cases:
- nf_conn_act_ct_ext_fill() is called before extension is created with
nf_conn_act_ct_ext_add() in tcf_ct_act(). This can cause rule offload with
unspecified iifidx when connection is offloaded after only single
original-direction packet has been processed by tc data path. Always fill
the new nf_conn_act_ct_ext instance after creating it in
nf_conn_act_ct_ext_add().
- Offloading of unidirectional UDP NEW connections is now supported, but ct
flow iifidx field is not updated when connection is promoted to
bidirectional which can result reply-direction iifidx to be zero when
refreshing the connection. Fill in the extension and update flow iifidx
before calling flow_offload_refresh().
Fixes: 9795ded7f9 ("net/sched: act_ct: Fill offloading tuple iifidx")
Reviewed-by: Paul Blakey <paulb@nvidia.com>
Signed-off-by: Vlad Buslov <vladbu@nvidia.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Fixes: 6a9bad0069 ("net/sched: act_ct: offload UDP NEW connections")
Link: https://lore.kernel.org/r/20231103151410.764271-1-vladbu@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Since 41f2c7c342 ("net/sched: act_ct: Fix promotion of offloaded
unreplied tuple"), flowtable GC pushes back flows with IPS_SEEN_REPLY
back to classic path in every run, ie. every second. This is because of
a new check for NF_FLOW_HW_ESTABLISHED which is specific of sched/act_ct.
In Netfilter's flowtable case, NF_FLOW_HW_ESTABLISHED never gets set on
and IPS_SEEN_REPLY is unreliable since users decide when to offload the
flow before, such bit might be set on at a later stage.
Fix it by adding a custom .gc handler that sched/act_ct can use to
deal with its NF_FLOW_HW_ESTABLISHED bit.
Fixes: 41f2c7c342 ("net/sched: act_ct: Fix promotion of offloaded unreplied tuple")
Reported-by: Vladimir Smelhaus <vl.sm@email.cz>
Reviewed-by: Paul Blakey <paulb@nvidia.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Return struct nft_elem_priv instead of struct nft_set_ext for
consistency with ("netfilter: nf_tables: expose opaque set element as
struct nft_elem_priv") and to prepare the introduction of element
timeout updates from control path.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Instead of copying struct nft_set_elem into struct nft_trans_elem, store
the pointer to the opaque set element object in the transaction. Adapt
set backend API (and set backend implementations) to take the pointer to
opaque set element representation whenever required.
This patch deconstifies .remove() and .activate() set backend API since
these modify the set element opaque object. And it also constify
nft_set_elem_ext() this provides access to the nft_set_ext struct
without updating the object.
According to pahole on x86_64, this patch shrinks struct nft_trans_elem
size from 216 to 24 bytes.
This patch also reduces stack memory consumption by removing the
template struct nft_set_elem object, using the opaque set element object
instead such as from the set iterator API, catchall elements and the get
element command.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Add placeholder structure and place it at the beginning of each struct
nft_*_elem for each existing set backend, instead of exposing elements
as void type to the frontend which defeats compiler type checks. Use
this pointer to this new type to replace void *.
This patch updates the following set backend API to use this new struct
nft_elem_priv placeholder structure:
- update
- deactivate
- flush
- get
as well as the following helper functions:
- nft_set_elem_ext()
- nft_set_elem_init()
- nft_set_elem_destroy()
- nf_tables_set_elem_destroy()
This patch adds nft_elem_priv_cast() to cast struct nft_elem_priv to
native element representation from the corresponding set backend.
BUILD_BUG_ON() makes sure this .priv placeholder is always at the top
of the opaque set element representation.
Suggested-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
.flush is always successful since this results from iterating over the
set elements to toggle mark the element as inactive in the next
generation.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
The spinlock is back from the day when connabels did not have
a fixed size and reallocation had to be supported.
Remove it. This change also allows to call the helpers from
softirq or timers without deadlocks.
Also add WARN()s to catch refcounting imbalances.
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Add comments for nlpid, family, udlen and udata in struct nft_table, and
afinfo is no longer a member of struct nft_table, so remove the comment
for it.
Signed-off-by: George Guo <guodongtai@kylinos.cn>
Signed-off-by: Florian Westphal <fw@strlen.de>
nf_conntrack_alter_reply doesn't do helper reassignment anymore.
Remove the comments that make this claim.
Furthermore, remove dead code from the function and place ot
in nf_conntrack.h.
Signed-off-by: Florian Westphal <fw@strlen.de>
When more than 255 elements expired we're supposed to switch to a new gc
container structure.
This never happens: u8 type will wrap before reaching the boundary
and nft_trans_gc_space() always returns true.
This means we recycle the initial gc container structure and
lose track of the elements that came before.
While at it, don't deref 'gc' after we've passed it to call_rcu.
Fixes: 5f68718b34 ("netfilter: nf_tables: GC transaction API to avoid race with control plane")
Reported-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Florian Westphal <fw@strlen.de>
pipapo needs to enqueue GC transactions for catchall elements through
nft_trans_gc_queue_sync(). Add nft_trans_gc_catchall_sync() and
nft_trans_gc_catchall_async() to handle GC transaction queueing
accordingly.
Fixes: 5f68718b34 ("netfilter: nf_tables: GC transaction API to avoid race with control plane")
Fixes: f6c383b8c3 ("netfilter: nf_tables: adapt set backend to use GC transaction API")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Don't queue more gc work, else we may queue the same elements multiple
times.
If an element is flagged as dead, this can mean that either the previous
gc request was invalidated/discarded by a transaction or that the previous
request is still pending in the system work queue.
The latter will happen if the gc interval is set to a very low value,
e.g. 1ms, and system work queue is backlogged.
The sets refcount is 1 if no previous gc requeusts are queued, so add
a helper for this and skip gc run if old requests are pending.
Add a helper for this and skip the gc run in this case.
Fixes: f6c383b8c3 ("netfilter: nf_tables: adapt set backend to use GC transaction API")
Signed-off-by: Florian Westphal <fw@strlen.de>
Reviewed-by: Pablo Neira Ayuso <pablo@netfilter.org>
We have to validate all tables in the transaction that are in
VALIDATE_DO state, the blamed commit below did not move the break
statement to its right location so we only validate one table.
Moreover, we can't init table->validate to _SKIP when a table object
is allocated.
If we do, then if a transcaction creates a new table and then
fails the transaction, nfnetlink will loop and nft will hang until
user cancels the command.
Add back the pernet state as a place to stash the last state encountered.
This is either _DO (we hit an error during commit validation) or _SKIP
(transaction passed all checks).
Fixes: 00c320f9b7 ("netfilter: nf_tables: make validation state per table")
Reported-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Florian Westphal <fw@strlen.de>
Jakub Kicinski says:
We've got some new kdoc warnings here:
net/netfilter/nft_set_pipapo.c:1557: warning: Function parameter or member '_set' not described in 'pipapo_gc'
net/netfilter/nft_set_pipapo.c:1557: warning: Excess function parameter 'set' description in 'pipapo_gc'
include/net/netfilter/nf_tables.h:577: warning: Function parameter or member 'dead' not described in 'nft_set'
Fixes: 5f68718b34 ("netfilter: nf_tables: GC transaction API to avoid race with control plane")
Fixes: f6c383b8c3 ("netfilter: nf_tables: adapt set backend to use GC transaction API")
Reported-by: Jakub Kicinski <kuba@kernel.org>
Closes: https://lore.kernel.org/netdev/20230810104638.746e46f1@kernel.org/
Signed-off-by: Florian Westphal <fw@strlen.de>
The set types rhashtable and rbtree use a GC worker to reclaim memory.
From system work queue, in periodic intervals, a scan of the table is
done.
The major caveat here is that the nft transaction mutex is not held.
This causes a race between control plane and GC when they attempt to
delete the same element.
We cannot grab the netlink mutex from the work queue, because the
control plane has to wait for the GC work queue in case the set is to be
removed, so we get following deadlock:
cpu 1 cpu2
GC work transaction comes in , lock nft mutex
`acquire nft mutex // BLOCKS
transaction asks to remove the set
set destruction calls cancel_work_sync()
cancel_work_sync will now block forever, because it is waiting for the
mutex the caller already owns.
This patch adds a new API that deals with garbage collection in two
steps:
1) Lockless GC of expired elements sets on the NFT_SET_ELEM_DEAD_BIT
so they are not visible via lookup. Annotate current GC sequence in
the GC transaction. Enqueue GC transaction work as soon as it is
full. If ruleset is updated, then GC transaction is aborted and
retried later.
2) GC work grabs the mutex. If GC sequence has changed then this GC
transaction lost race with control plane, abort it as it contains
stale references to objects and let GC try again later. If the
ruleset is intact, then this GC transaction deactivates and removes
the elements and it uses call_rcu() to destroy elements.
Note that no elements are removed from GC lockless path, the _DEAD bit
is set and pointers are collected. GC catchall does not remove the
elements anymore too. There is a new set->dead flag that is set on to
abort the GC transaction to deal with set->ops->destroy() path which
removes the remaining elements in the set from commit_release, where no
mutex is held.
To deal with GC when mutex is held, which allows safe deactivate and
removal, add sync GC API which releases the set element object via
call_rcu(). This is used by rbtree and pipapo backends which also
perform garbage collection from control plane path.
Since element removal from sets can happen from control plane and
element garbage collection/timeout, it is necessary to keep the set
structure alive until all elements have been deactivated and destroyed.
We cannot do a cancel_work_sync or flush_work in nft_set_destroy because
its called with the transaction mutex held, but the aforementioned async
work queue might be blocked on the very mutex that nft_set_destroy()
callchain is sitting on.
This gives us the choice of ABBA deadlock or UaF.
To avoid both, add set->refs refcount_t member. The GC API can then
increment the set refcount and release it once the elements have been
free'd.
Set backends are adapted to use the GC transaction API in a follow up
patch entitled:
("netfilter: nf_tables: use gc transaction API in set backends")
This is joint work with Florian Westphal.
Fixes: cfed7e1b1f ("netfilter: nf_tables: add set garbage collection helpers")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Currently nf_conntrack_in() calling nf_ct_find_expectation() will
remove the exp from the hash table. However, in some scenario, we
expect the exp not to be removed when the created ct will not be
confirmed, like in OVS and TC conntrack in the following patches.
This patch allows exp not to be removed by setting IPS_CONFIRMED
in the status of the tmpl.
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Acked-by: Aaron Conole <aconole@redhat.com>
Acked-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
