commit f6c383b8c3 upstream.
Use the GC transaction API to replace the old and buggy gc API and the
busy mark approach.
No set elements are removed from async garbage collection anymore,
instead the _DEAD bit is set on so the set element is not visible from
lookup path anymore. Async GC enqueues transaction work that might be
aborted and retried later.
rbtree and pipapo set backends does not set on the _DEAD bit from the
sync GC path since this runs in control plane path where mutex is held.
In this case, set elements are deactivated, removed and then released
via RCU callback, sync GC never fails.
Fixes: 3c4287f620 ("nf_tables: Add set type for arbitrary concatenation of ranges")
Fixes: 8d8540c4f5 ("netfilter: nft_set_rbtree: add timeout support")
Fixes: 9d0982927e ("netfilter: nft_hash: add support for timeouts")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 24138933b9 upstream.
There is an asymmetry between commit/abort and preparation phase if the
following conditions are met:
1. set is a verdict map ("1.2.3.4 : jump foo")
2. timeouts are enabled
In this case, following sequence is problematic:
1. element E in set S refers to chain C
2. userspace requests removal of set S
3. kernel does a set walk to decrement chain->use count for all elements
from preparation phase
4. kernel does another set walk to remove elements from the commit phase
(or another walk to do a chain->use increment for all elements from
abort phase)
If E has already expired in 1), it will be ignored during list walk, so its use count
won't have been changed.
Then, when set is culled, ->destroy callback will zap the element via
nf_tables_set_elem_destroy(), but this function is only safe for
elements that have been deactivated earlier from the preparation phase:
lack of earlier deactivate removes the element but leaks the chain use
count, which results in a WARN splat when the chain gets removed later,
plus a leak of the nft_chain structure.
Update pipapo_get() not to skip expired elements, otherwise flush
command reports bogus ENOENT errors.
Fixes: 3c4287f620 ("nf_tables: Add set type for arbitrary concatenation of ranges")
Fixes: 8d8540c4f5 ("netfilter: nft_set_rbtree: add timeout support")
Fixes: 9d0982927e ("netfilter: nft_hash: add support for timeouts")
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>
[ Upstream commit f718863aca ]
The lazy gc on insert that should remove timed-out entries fails to release
the other half of the interval, if any.
Can be reproduced with tests/shell/testcases/sets/0044interval_overlap_0
in nftables.git and kmemleak enabled kernel.
Second bug is the use of rbe_prev vs. prev pointer.
If rbe_prev() returns NULL after at least one iteration, rbe_prev points
to element that is not an end interval, hence it should not be removed.
Lastly, check the genmask of the end interval if this is active in the
current generation.
Fixes: c9e6978e27 ("netfilter: nft_set_rbtree: Switch to node list walk for overlap detection")
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
set .destroy callback releases the references to other objects in maps.
This is very late and it results in spurious EBUSY errors. Drop refcount
from the preparation phase instead, update set backend not to drop
reference counter from set .destroy path.
Exceptions: NFT_TRANS_PREPARE_ERROR does not require to drop the
reference counter because the transaction abort path releases the map
references for each element since the set is unbound. The abort path
also deals with releasing reference counter for new elements added to
unbound sets.
Fixes: 591054469b ("netfilter: nf_tables: revisit chain/object refcounting from elements")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
There is no guarantee that rb_prev() will not return NULL in nft_rbtree_gc_elem():
general protection fault, probably for non-canonical address 0xdffffc0000000003: 0000 [#1] PREEMPT SMP KASAN
KASAN: null-ptr-deref in range [0x0000000000000018-0x000000000000001f]
nft_add_set_elem+0x14b0/0x2990
nf_tables_newsetelem+0x528/0xb30
Furthermore, there is a possible use-after-free while iterating,
'node' can be free'd so we need to cache the next value to use.
Fixes: c9e6978e27 ("netfilter: nft_set_rbtree: Switch to node list walk for overlap detection")
Signed-off-by: Florian Westphal <fw@strlen.de>
Skip interference with an ongoing transaction, do not perform garbage
collection on inactive elements. Reset annotated previous end interval
if the expired element is marked as busy (control plane removed the
element right before expiration).
Fixes: 8d8540c4f5 ("netfilter: nft_set_rbtree: add timeout support")
Reviewed-by: Stefano Brivio <sbrivio@redhat.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
...instead of a tree descent, which became overly complicated in an
attempt to cover cases where expired or inactive elements would affect
comparisons with the new element being inserted.
Further, it turned out that it's probably impossible to cover all those
cases, as inactive nodes might entirely hide subtrees consisting of a
complete interval plus a node that makes the current insertion not
overlap.
To speed up the overlap check, descent the tree to find a greater
element that is closer to the key value to insert. Then walk down the
node list for overlap detection. Starting the overlap check from
rb_first() unconditionally is slow, it takes 10 times longer due to the
full linear traversal of the list.
Moreover, perform garbage collection of expired elements when walking
down the node list to avoid bogus overlap reports.
For the insertion operation itself, this essentially reverts back to the
implementation before commit 7c84d41416 ("netfilter: nft_set_rbtree:
Detect partial overlaps on insertion"), except that cases of complete
overlap are already handled in the overlap detection phase itself, which
slightly simplifies the loop to find the insertion point.
Based on initial patch from Stefano Brivio, including text from the
original patch description too.
Fixes: 7c84d41416 ("netfilter: nft_set_rbtree: Detect partial overlaps on insertion")
Reviewed-by: Stefano Brivio <sbrivio@redhat.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
This patch fixes spurious EEXIST errors.
Extend d2df92e98a ("netfilter: nft_set_rbtree: handle element
re-addition after deletion") to deal with elements with same end flags
in the same transation.
Reset the overlap flag as described by 7c84d41416 ("netfilter:
nft_set_rbtree: Detect partial overlaps on insertion").
Fixes: 7c84d41416 ("netfilter: nft_set_rbtree: Detect partial overlaps on insertion")
Fixes: d2df92e98a ("netfilter: nft_set_rbtree: handle element re-addition after deletion")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Reviewed-by: Stefano Brivio <sbrivio@redhat.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Extend nft_set_do_lookup() to use direct calls when retpoline feature
is enabled.
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
This patch extends the set infrastructure to add a special catch-all set
element. If the lookup fails to find an element (or range) in the set,
then the catch-all element is selected. Users can specify a mapping,
expression(s) and timeout to be attached to the catch-all element.
This patch adds a catchall list to the set, this list might contain more
than one single catch-all element (e.g. in case that the catch-all
element is removed and a new one is added in the same transaction).
However, most of the time, there will be either one element or no
elements at all in this list.
The catch-all element is identified via NFT_SET_ELEM_CATCHALL flag and
such special element has no NFTA_SET_ELEM_KEY attribute. There is a new
nft_set_elem_catchall object that stores a reference to the dummy
catch-all element (catchall->elem) whose layout is the same of the set
element type to reuse the existing set element codebase.
The set size does not apply to the catch-all element, users can define a
catch-all element even if the set is full.
The check for valid set element flags hava been updates to report
EOPNOTSUPP in case userspace requests flags that are not supported when
using new userspace nftables and old kernel.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Getting creative with nft and omitting the interval_overlap()
check from the set_overlap() function, without omitting
set_overlap() altogether, led to the observation of a partial
overlap that wasn't detected, and would actually result in
replacement of the end element of an existing interval.
This is due to the fact that we'll return -EEXIST on a matching,
pre-existing start element, instead of -ENOTEMPTY, and the error
is cleared by API if NLM_F_EXCL is not given. At this point, we
can insert a matching start, and duplicate the end element as long
as we don't end up into other intervals.
For instance, inserting interval 0 - 2 with an existing 0 - 3
interval would result in a single 0 - 2 interval, and a dangling
'3' end element. This is because nft will proceed after inserting
the '0' start element as no error is reported, and no further
conflicting intervals are detected on insertion of the end element.
This needs a different approach as it's a local condition that can
be detected by looking for duplicate ends coming from left and
right, separately. Track those and directly report -ENOTEMPTY on
duplicated end elements for a matching start.
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Checks for partial overlaps on insertion assume that end elements
are always descendant nodes of their corresponding start, because
they are inserted later. However, this is not the case if a
previous delete operation caused a tree rotation as part of
rebalancing.
Taking the issue reported by Andreas Fischer as an example, if we
omit delete operations, the existing procedure works because,
equivalently, we are inserting a start item with value 40 in the
this region of the red-black tree with single-sized intervals:
overlap flag
10 (start)
/ \ false
20 (start)
/ \ false
30 (start)
/ \ false
60 (start)
/ \ false
50 (end)
/ \ false
20 (end)
/ \ false
40 (start)
if we now delete interval 30 - 30, the tree can be rearranged in
a way similar to this (note the rotation involving 50 - 50):
overlap flag
10 (start)
/ \ false
20 (start)
/ \ false
25 (start)
/ \ false
70 (start)
/ \ false
50 (end)
/ \ true (from rule a1.)
50 (start)
/ \ true
40 (start)
and we traverse interval 50 - 50 from the opposite direction
compared to what was expected.
To deal with those cases, add a start-before-start rule, b4.,
that covers traversal of existing intervals from the right.
We now need to restrict start-after-end rule b3. to cases
where there are no occurring nodes between existing start and
end elements, because addition of rule b4. isn't sufficient to
ensure that the pre-existing end element we encounter while
descending the tree corresponds to a start element of an
interval that we already traversed entirely.
Different types of overlap detection on trees with rotations
resulting from re-balancing will be covered by nft test case
sets/0044interval_overlap_1.
Reported-by: Andreas Fischer <netfilter@d9c.eu>
Bugzilla: https://bugzilla.netfilter.org/show_bug.cgi?id=1449
Cc: <stable@vger.kernel.org> # 5.6.x
Fixes: 7c84d41416 ("netfilter: nft_set_rbtree: Detect partial overlaps on insertion")
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
A sequence counter write side critical section must be protected by some
form of locking to serialize writers. A plain seqcount_t does not
contain the information of which lock must be held when entering a write
side critical section.
Use the new seqcount_rwlock_t data type, which allows to associate a
rwlock with the sequence counter. This enables lockdep to verify that
the rwlock used for writer serialization is held when the write side
critical section is entered.
If lockdep is disabled this lock association is compiled out and has
neither storage size nor runtime overhead.
Signed-off-by: Ahmed S. Darwish <a.darwish@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20200720155530.1173732-16-a.darwish@linutronix.de
While checking the validity of insertion in __nft_rbtree_insert(),
we currently ignore conflicting elements and intervals only if they
are not active within the next generation.
However, if we consider expired elements and intervals as
potentially conflicting and overlapping, we'll return error for
entries that should be added instead. This is particularly visible
with garbage collection intervals that are comparable with the
element timeout itself, as reported by Mike Dillinger.
Other than the simple issue of denying insertion of valid entries,
this might also result in insertion of a single element (opening or
closing) out of a given interval. With single entries (that are
inserted as intervals of size 1), this leads in turn to the creation
of new intervals. For example:
# nft add element t s { 192.0.2.1 }
# nft list ruleset
[...]
elements = { 192.0.2.1-255.255.255.255 }
Always ignore expired elements active in the next generation, while
checking for conflicts.
It might be more convenient to introduce a new macro that covers
both inactive and expired items, as this type of check also appears
quite frequently in other set back-ends. This is however beyond the
scope of this fix and can be deferred to a separate patch.
Other than the overlap detection cases introduced by commit
7c84d41416 ("netfilter: nft_set_rbtree: Detect partial overlaps
on insertion"), we also have to cover the original conflict check
dealing with conflicts between two intervals of size 1, which was
introduced before support for timeout was introduced. This won't
return an error to the user as -EEXIST is masked by nft if
NLM_F_EXCL is not given, but would result in a silent failure
adding the entry.
Reported-by: Mike Dillinger <miked@softtalker.com>
Cc: <stable@vger.kernel.org> # 5.6.x
Fixes: 8d8540c4f5 ("netfilter: nft_set_rbtree: add timeout support")
Fixes: 7c84d41416 ("netfilter: nft_set_rbtree: Detect partial overlaps on insertion")
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Acked-by: Phil Sutter <phil@nwl.cc>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Expired intervals would still match and be dumped to user space until
garbage collection wiped them out. Make sure they stop matching and
disappear (from users' perspective) as soon as they expire.
Fixes: 8d8540c4f5 ("netfilter: nft_set_rbtree: add timeout support")
Signed-off-by: Phil Sutter <phil@nwl.cc>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Case a1. for overlap detection in __nft_rbtree_insert() is not a valid
one: start-after-start is not needed to detect any type of interval
overlap and it actually results in a false positive if, while
descending the tree, this is the only step we hit after starting from
the root.
This introduced a regression, as reported by Pablo, in Python tests
cases ip/ip.t and ip/numgen.t:
ip/ip.t: ERROR: line 124: add rule ip test-ip4 input ip hdrlength vmap { 0-4 : drop, 5 : accept, 6 : continue } counter: This rule should not have failed.
ip/numgen.t: ERROR: line 7: add rule ip test-ip4 pre dnat to numgen inc mod 10 map { 0-5 : 192.168.10.100, 6-9 : 192.168.20.200}: This rule should not have failed.
Drop case a1. and renumber others, so that they are a bit clearer. In
order for these diagrams to be readily understandable, a bigger rework
is probably needed, such as an ASCII art of the actual rbtree (instead
of a flattened version).
Shell script test sets/0044interval_overlap_0 should cover all
possible cases for false negatives, so I consider that test case still
sufficient after this change.
v2: Fix comments for cases a3. and b3.
Reported-by: Pablo Neira Ayuso <pablo@netfilter.org>
Fixes: 7c84d41416 ("netfilter: nft_set_rbtree: Detect partial overlaps on insertion")
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Overlapping header include additions in macsec.c
A bug fix in 'net' overlapping with the removal of 'version'
string in ena_netdev.c
Overlapping test additions in selftests Makefile
Overlapping PCI ID table adjustments in iwlwifi driver.
Signed-off-by: David S. Miller <davem@davemloft.net>
...and return -ENOTEMPTY to the front-end in this case, instead of
proceeding. Currently, nft takes care of checking for these cases
and not sending them to the kernel, but if we drop the set_overlap()
call in nft we can end up in situations like:
# nft add table t
# nft add set t s '{ type inet_service ; flags interval ; }'
# nft add element t s '{ 1 - 5 }'
# nft add element t s '{ 6 - 10 }'
# nft add element t s '{ 4 - 7 }'
# nft list set t s
table ip t {
set s {
type inet_service
flags interval
elements = { 1-3, 4-5, 6-7 }
}
}
This change has the primary purpose of making the behaviour
consistent with nft_set_pipapo, but is also functional to avoid
inconsistent behaviour if userspace sends overlapping elements for
any reason.
v2: When we meet the same key data in the tree, as start element while
inserting an end element, or as end element while inserting a start
element, actually check that the existing element is active, before
resetting the overlap flag (Pablo Neira Ayuso)
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Replace negations of nft_rbtree_interval_end() with a new helper,
nft_rbtree_interval_start(), wherever this helps to visualise the
problem at hand, that is, for all the occurrences except for the
comparison against given flags in __nft_rbtree_get().
This gets especially useful in the next patch.
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
They do not need to be writeable anymore.
v2: remove left-over __read_mostly annotation in set_pipapo.c (Stefano)
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Introduce a new nested netlink attribute, NFTA_SET_DESC_CONCAT, used
to specify the length of each field in a set concatenation.
This allows set implementations to support concatenation of multiple
ranged items, as they can divide the input key into matching data for
every single field. Such set implementations would be selected as
they specify support for NFT_SET_INTERVAL and allow desc->field_count
to be greater than one. Explicitly disallow this for nft_set_rbtree.
In order to specify the interval for a set entry, userspace would
include in NFTA_SET_DESC_CONCAT attributes field lengths, and pass
range endpoints as two separate keys, represented by attributes
NFTA_SET_ELEM_KEY and NFTA_SET_ELEM_KEY_END.
While at it, export the number of 32-bit registers available for
packet matching, as nftables will need this to know the maximum
number of field lengths that can be specified.
For example, "packets with an IPv4 address between 192.0.2.0 and
192.0.2.42, with destination port between 22 and 25", can be
expressed as two concatenated elements:
NFTA_SET_ELEM_KEY: 192.0.2.0 . 22
NFTA_SET_ELEM_KEY_END: 192.0.2.42 . 25
and NFTA_SET_DESC_CONCAT attribute would contain:
NFTA_LIST_ELEM
NFTA_SET_FIELD_LEN: 4
NFTA_LIST_ELEM
NFTA_SET_FIELD_LEN: 2
v4: No changes
v3: Complete rework, NFTA_SET_DESC_CONCAT instead of NFTA_SET_SUBKEY
v2: No changes
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
The existing rbtree implementation might store consecutive elements
where the closing element and the opening element might overlap, eg.
[ a, a+1) [ a+1, a+2)
This patch removes the optimization for non-anonymous sets in the exact
matching case, where it is assumed to stop searching in case that the
closing element is found. Instead, invalidate candidate interval and
keep looking further in the tree.
The lookup/get operation might return false, while there is an element
in the rbtree. Moreover, the get operation returns true as if a+2 would
be in the tree. This happens with named sets after several set updates.
The existing lookup optimization (that only works for the anonymous
sets) might not reach the opening [ a+1,... element if the closing
...,a+1) is found in first place when walking over the rbtree. Hence,
walking the full tree in that case is needed.
This patch fixes the lookup and get operations.
Fixes: e701001e7c ("netfilter: nft_rbtree: allow adjacent intervals with dynamic updates")
Fixes: ba0e4d9917 ("netfilter: nf_tables: get set elements via netlink")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Sparse rightly complains about undeclared symbols.
CHECK net/netfilter/nft_set_hash.c
net/netfilter/nft_set_hash.c:647:21: warning: symbol 'nft_set_rhash_type' was not declared. Should it be static?
net/netfilter/nft_set_hash.c:670:21: warning: symbol 'nft_set_hash_type' was not declared. Should it be static?
net/netfilter/nft_set_hash.c:690:21: warning: symbol 'nft_set_hash_fast_type' was not declared. Should it be static?
CHECK net/netfilter/nft_set_bitmap.c
net/netfilter/nft_set_bitmap.c:296:21: warning: symbol 'nft_set_bitmap_type' was not declared. Should it be static?
CHECK net/netfilter/nft_set_rbtree.c
net/netfilter/nft_set_rbtree.c:470:21: warning: symbol 'nft_set_rbtree_type' was not declared. Should it be static?
Include nf_tables_core.h rather than nf_tables.h to pick up the additional definitions.
Signed-off-by: Valdis Kletnieks <valdis.kletnieks@vt.edu>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Based on 2 normalized pattern(s):
this program is free software you can redistribute it and or modify
it under the terms of the gnu general public license version 2 as
published by the free software foundation
this program is free software you can redistribute it and or modify
it under the terms of the gnu general public license version 2 as
published by the free software foundation #
extracted by the scancode license scanner the SPDX license identifier
GPL-2.0-only
has been chosen to replace the boilerplate/reference in 4122 file(s).
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Enrico Weigelt <info@metux.net>
Reviewed-by: Kate Stewart <kstewart@linuxfoundation.org>
Reviewed-by: Allison Randal <allison@lohutok.net>
Cc: linux-spdx@vger.kernel.org
Link: https://lkml.kernel.org/r/20190604081206.933168790@linutronix.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Otherwise, we hit bogus ENOENT when removing elements.
Fixes: e701001e7c ("netfilter: nft_rbtree: allow adjacent intervals with dynamic updates")
Reported-by: Václav Zindulka <vaclav.zindulka@tlapnet.cz>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Allow to find closest matching for the right side of an interval (end
flag set on) so we allow lookups in inner ranges, eg. 10-20 in 5-25.
Fixes: ba0e4d9917 ("netfilter: nf_tables: get set elements via netlink")
Reported-by: Phil Sutter <phil@nwl.cc>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
This patch disallows rbtree with single elements, which is causing
problems with the recent timeout support. Before this patch, you
could opt out individual set representations per module, which is
just adding extra complexity.
Fixes: 8d8540c4f5e0("netfilter: nft_set_rbtree: add timeout support")
Reported-by: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Pablo Neira Ayuso says:
====================
Netfilter/IPVS fixes for net
The following patchset contains Netfilter/IPVS fixes for your net tree:
1) Reject non-null terminated helper names from xt_CT, from Gao Feng.
2) Fix KASAN splat due to out-of-bound access from commit phase, from
Alexey Kodanev.
3) Missing conntrack hook registration on IPVS FTP helper, from Julian
Anastasov.
4) Incorrect skbuff allocation size in bridge nft_reject, from Taehee Yoo.
5) Fix inverted check on packet xmit to non-local addresses, also from
Julian.
6) Fix ebtables alignment compat problems, from Alin Nastac.
7) Hook mask checks are not correct in xt_set, from Serhey Popovych.
8) Fix timeout listing of element in ipsets, from Jozsef.
9) Cap maximum timeout value in ipset, also from Jozsef.
10) Don't allow family option for hash:mac sets, from Florent Fourcot.
11) Restrict ebtables to work with NFPROTO_BRIDGE targets only, this
Florian.
12) Another bug reported by KASAN in the rbtree set backend, from
Taehee Yoo.
13) Missing __IPS_MAX_BIT update doesn't include IPS_OFFLOAD_BIT.
From Gao Feng.
14) Missing initialization of match/target in ebtables, from Florian
Westphal.
15) Remove useless nft_dup.h file in include path, from C. Labbe.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Drop nft_set_type's ability to act as a container of multiple backend
implementations it chooses from. Instead consolidate the whole selection
logic in nft_select_set_ops() and the actual backend provided estimate()
callback.
This turns nf_tables_set_types into a list containing all available
backends which is traversed when selecting one matching userspace
requested criteria.
Also, this change allows to embed nft_set_ops structure into
nft_set_type and pull flags field into the latter as it's only used
during selection phase.
A crucial part of this change is to make sure the new layout respects
hash backend constraints formerly enforced by nft_hash_select_ops()
function: This is achieved by introduction of a specific estimate()
callback for nft_hash_fast_ops which returns false for key lengths != 4.
In turn, nft_hash_estimate() is changed to return false for key lengths
== 4 so it won't be chosen by accident. Also, both callbacks must return
false for unbounded sets as their size estimate depends on a known
maximum element count.
Note that this patch partially reverts commit 4f2921ca21 ("netfilter:
nf_tables: meter: pick a set backend that supports updates") by making
nft_set_ops_candidate() not explicitly look for an update callback but
make NFT_SET_EVAL a regular backend feature flag which is checked along
with the others. This way all feature requirements are checked in one
go.
Signed-off-by: Phil Sutter <phil@nwl.cc>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
This patch adds a new get operation to look up for specific elements in
a set via netlink interface. You can also use it to check if an interval
already exists.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
switch to lockless lockup. write side now also increments sequence
counter. On lookup, sample counter value and only take the lock
if we did not find a match and the counter has changed.
This avoids need to write to private area in normal (lookup) cases.
In case we detect a writer (seqretry is true) we fall back to taking
the readlock.
The readlock is also used during dumps to ensure we get a consistent
tree walk.
Similar technique (rbtree+seqlock) was used by David Howells in rxrpc.
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Pablo Neira Ayuso says:
====================
Netfilter updates for net-next
The following patchset contains Netfilter updates for your net-next
tree. This batch contains connection tracking updates for the cleanup
iteration path, patches from Florian Westphal:
X) Skip unconfirmed conntracks in nf_ct_iterate_cleanup_net(), just set
dying bit to let the CPU release them.
X) Add nf_ct_iterate_destroy() to be used on module removal, to kill
conntrack from all namespace.
X) Restart iteration on hashtable resizing, since both may occur at
the same time.
X) Use the new nf_ct_iterate_destroy() to remove conntrack with NAT
mapping on module removal.
X) Use nf_ct_iterate_destroy() to remove conntrack entries helper
module removal, from Liping Zhang.
X) Use nf_ct_iterate_cleanup_net() to remove the timeout extension
if user requests this, also from Liping.
X) Add net_ns_barrier() and use it from FTP helper, so make sure
no concurrent namespace removal happens at the same time while
the helper module is being removed.
X) Use NFPROTO_MAX in layer 3 conntrack protocol array, to reduce
module size. Same thing in nf_tables.
Updates for the nf_tables infrastructure:
X) Prepare usage of the extended ACK reporting infrastructure for
nf_tables.
X) Remove unnecessary forward declaration in nf_tables hash set.
X) Skip set size estimation if number of element is not specified.
X) Changes to accomodate a (faster) unresizable hash set implementation,
for anonymous sets and dynamic size fixed sets with no timeouts.
X) Faster lookup function for unresizable hash table for 2 and 4
bytes key.
And, finally, a bunch of asorted small updates and cleanups:
X) Do not hold reference to netdev from ipt_CLUSTER, instead subscribe
to device events and look up for index from the packet path, this
is fixing an issue that is present since the very beginning, patch
from Xin Long.
X) Use nf_register_net_hook() in ipt_CLUSTER, from Florian Westphal.
X) Use ebt_invalid_target() whenever possible in the ebtables tree,
from Gao Feng.
X) Calm down compilation warning in nf_dup infrastructure, patch from
stephen hemminger.
X) Statify functions in nftables rt expression, also from stephen.
X) Update Makefile to use canonical method to specify nf_tables-objs.
From Jike Song.
X) Use nf_conntrack_helpers_register() in amanda and H323.
X) Space cleanup for ctnetlink, from linzhang.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds the infrastructure to support several implementations of
the same set type. This selection will be based on the set description
and the features available for this set. This allow us to select set
backend implementation that will result in better performance numbers.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
This size estimation is ignored by the existing set backend selection
logic, since this estimation structure is stack allocated, set this to
~0 to make it easier to catch bugs in future changes.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
The existing code selects no next branch to be inspected when
re-inserting an inactive element into the rb-tree, looping endlessly.
This patch restricts the check for active elements to the EEXIST case
only.
Fixes: e701001e7c ("netfilter: nft_rbtree: allow adjacent intervals with dynamic updates")
Reported-by: Wolfgang Bumiller <w.bumiller@proxmox.com>
Tested-by: Wolfgang Bumiller <w.bumiller@proxmox.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Karel Rericha reported that in his test case, ICMP packets going through
boxes had normally about 5ms latency. But when running nft, actually
listing the sets with interval flags, latency would go up to 30-100ms.
This was observed when router throughput is from 600Mbps to 2Gbps.
This is because we use a single global spinlock to protect the whole
rbtree sets, so "dumping sets" will race with the "key lookup" inevitably.
But actually they are all _readers_, so it's ok to convert the spinlock
to rwlock to avoid competition between them. Also use per-set rwlock since
each set is independent.
Reported-by: Karel Rericha <karel@unitednetworks.cz>
Tested-by: Karel Rericha <karel@unitednetworks.cz>
Signed-off-by: Liping Zhang <zlpnobody@gmail.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
In case of adjacent ranges, we may indeed see either the high part of
the range in first place or the low part of it. Remove this incorrect
assumption, let's make sure we annotate the low part of the interval in
case of we have adjacent interva intervals so we hit a matching in
lookups.
Reported-by: Simon Hanisch <hanisch@wh2.tu-dresden.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
The space notation allows us to classify the set backend implementation
based on the amount of required memory. This provides an order of the
set representation scalability in terms of memory. The size field is
still left in place so use this if the userspace provides no explicit
number of elements, so we cannot calculate the real memory that this set
needs. This also helps us break ties in the set backend selection
routine, eg. two backend implementations provide the same performance.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Use lookup as field name instead, to prepare the introduction of the
memory class in a follow up patch.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Although semantics are similar to deactivate() with no implicit element
lookup, this is only called from the set flush path, so better rename
this to flush().
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
This patch adds support for set flushing, that consists of walking over
the set elements if the NFTA_SET_ELEM_LIST_ELEMENTS attribute is set.
This patch requires the following changes:
1) Add set->ops->deactivate_one() operation: This allows us to
deactivate an element from the set element walk path, given we can
skip the lookup that happens in ->deactivate().
2) Add a new nft_trans_alloc_gfp() function since we need to allocate
transactions using GFP_ATOMIC given the set walk path happens with
held rcu_read_lock.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Pablo Neira Ayuso