As Eric reported, the 'reason' field is not presented when trace the
kfree_skb event by perf:
$ perf record -e skb:kfree_skb -a sleep 10
$ perf script
ip_defrag 14605 [021] 221.614303: skb:kfree_skb:
skbaddr=0xffff9d2851242700 protocol=34525 location=0xffffffffa39346b1
reason:
The cause seems to be passing kernel address directly to TP_printk(),
which is not right. As the enum 'skb_drop_reason' is not exported to
user space through TRACE_DEFINE_ENUM(), perf can't get the drop reason
string from the 'reason' field, which is a number.
Therefore, we introduce the macro DEFINE_DROP_REASON(), which is used
to define the trace enum by TRACE_DEFINE_ENUM(). With the help of
DEFINE_DROP_REASON(), now we can remove the auto-generate that we
introduced in the commit ec43908dd5
("net: skb: use auto-generation to convert skb drop reason to string"),
and define the string array 'drop_reasons'.
Hmmmm...now we come back to the situation that have to maintain drop
reasons in both enum skb_drop_reason and DEFINE_DROP_REASON. But they
are both in dropreason.h, which makes it easier.
After this commit, now the format of kfree_skb is like this:
$ cat /tracing/events/skb/kfree_skb/format
name: kfree_skb
ID: 1524
format:
field:unsigned short common_type; offset:0; size:2; signed:0;
field:unsigned char common_flags; offset:2; size:1; signed:0;
field:unsigned char common_preempt_count; offset:3; size:1; signed:0;
field:int common_pid; offset:4; size:4; signed:1;
field:void * skbaddr; offset:8; size:8; signed:0;
field:void * location; offset:16; size:8; signed:0;
field:unsigned short protocol; offset:24; size:2; signed:0;
field:enum skb_drop_reason reason; offset:28; size:4; signed:0;
print fmt: "skbaddr=%p protocol=%u location=%p reason: %s", REC->skbaddr, REC->protocol, REC->location, __print_symbolic(REC->reason, { 1, "NOT_SPECIFIED" }, { 2, "NO_SOCKET" } ......
Fixes: ec43908dd5 ("net: skb: use auto-generation to convert skb drop reason to string")
Link: https://lore.kernel.org/netdev/CANn89i+bx0ybvE55iMYf5GJM48WwV1HNpdm9Q6t-HaEstqpCSA@mail.gmail.com/
Reported-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Menglong Dong <imagedong@tencent.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
While reading sysctl_tstamp_allow_data, it can be changed
concurrently. Thus, we need to add READ_ONCE() to its reader.
Fixes: b245be1f4d ("net-timestamp: no-payload only sysctl")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
skb_copy_bits() could fail, which requires a check on the return
value.
Signed-off-by: Li Zhong <floridsleeves@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pavel Begunkov says:
====================
io_uring zerocopy send
The patchset implements io_uring zerocopy send. It works with both registered
and normal buffers, mixing is allowed but not recommended. Apart from usual
request completions, just as with MSG_ZEROCOPY, io_uring separately notifies
the userspace when buffers are freed and can be reused (see API design below),
which is delivered into io_uring's Completion Queue. Those "buffer-free"
notifications are not necessarily per request, but the userspace has control
over it and should explicitly attaching a number of requests to a single
notification. The series also adds some internal optimisations when used with
registered buffers like removing page referencing.
From the kernel networking perspective there are two main changes. The first
one is passing ubuf_info into the network layer from io_uring (inside of an
in kernel struct msghdr). This allows extra optimisations, e.g. ubuf_info
caching on the io_uring side, but also helps to avoid cross-referencing
and synchronisation problems. The second part is an optional optimisation
removing page referencing for requests with registered buffers.
Benchmarking UDP with an optimised version of the selftest (see [1]), which
sends a bunch of requests, waits for completions and repeats. "+ flush" column
posts one additional "buffer-free" notification per request, and just "zc"
doesn't post buffer notifications at all.
NIC (requests / second):
IO size | non-zc | zc | zc + flush
4000 | 495134 | 606420 (+22%) | 558971 (+12%)
1500 | 551808 | 577116 (+4.5%) | 565803 (+2.5%)
1000 | 584677 | 592088 (+1.2%) | 560885 (-4%)
600 | 596292 | 598550 (+0.4%) | 555366 (-6.7%)
dummy (requests / second):
IO size | non-zc | zc | zc + flush
8000 | 1299916 | 2396600 (+84%) | 2224219 (+71%)
4000 | 1869230 | 2344146 (+25%) | 2170069 (+16%)
1200 | 2071617 | 2361960 (+14%) | 2203052 (+6%)
600 | 2106794 | 2381527 (+13%) | 2195295 (+4%)
Previously it also brought a massive performance speedup compared to the
msg_zerocopy tool (see [3]), which is probably not super interesting. There
is also an additional bunch of refcounting optimisations that was omitted from
the series for simplicity and as they don't change the picture drastically,
they will be sent as follow up, as well as flushing optimisations closing the
performance gap b/w two last columns.
For TCP on localhost (with hacks enabling localhost zerocopy) and including
additional overhead for receive:
IO size | non-zc | zc
1200 | 4174 | 4148
4096 | 7597 | 11228
Using a real NIC 1200 bytes, zc is worse than non-zc ~5-10%, maybe the
omitted optimisations will somewhat help, should look better for 4000,
but couldn't test properly because of setup problems.
Links:
liburing (benchmark + tests):
[1] https://github.com/isilence/liburing/tree/zc_v4
kernel repo:
[2] https://github.com/isilence/linux/tree/zc_v4
RFC v1:
[3] https://lore.kernel.org/io-uring/cover.1638282789.git.asml.silence@gmail.com/
RFC v2:
https://lore.kernel.org/io-uring/cover.1640029579.git.asml.silence@gmail.com/
Net patches based:
git@github.com:isilence/linux.git zc_v4-net-base
or
https://github.com/isilence/linux/tree/zc_v4-net-base
API design overview:
The series introduces an io_uring concept of notifactors. From the userspace
perspective it's an entity to which it can bind one or more requests and then
requesting to flush it. Flushing a notifier makes it impossible to attach new
requests to it, and instructs the notifier to post a completion once all
requests attached to it are completed and the kernel doesn't need the buffers
anymore.
Notifications are stored in notification slots, which should be registered as
an array in io_uring. Each slot stores only one notifier at any particular
moment. Flushing removes it from the slot and the slot automatically replaces
it with a new notifier. All operations with notifiers are done by specifying
an index of a slot it's currently in.
When registering a notification the userspace specifies a u64 tag for each
slot, which will be copied in notification completion entries as
cqe::user_data. cqe::res is 0 and cqe::flags is equal to wrap around u32
sequence number counting notifiers of a slot.
====================
Link: https://lore.kernel.org/r/cover.1657643355.git.asml.silence@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Some users like io_uring can do page pinning more efficiently, so we
want a way to delegate referencing to other subsystems. For that add
a new flag called SKBFL_MANAGED_FRAG_REFS. When set, skb doesn't hold
page references and upper layers are responsivle to managing page
lifetime.
It's allowed to convert skbs from managed to normal by calling
skb_zcopy_downgrade_managed(). The function will take all needed
page references and clear the flag. It's needed, for instance,
to avoid mixing managed modes.
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Add support for custom iov_iter handling to msghdr. The idea is that
in-kernel subsystems want control over how an SG is split.
Signed-off-by: David Ahern <dsahern@kernel.org>
[pavel: move callback into msghdr]
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
We don't want to list every single ubuf_info callback in
skb_orphan_frags(), add a flag controlling the behaviour.
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
We should not append MSG_ZEROCOPY requests to skbuff with non
MSG_ZEROCOPY ubuf_info, they might be not compatible.
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
TCP allocates 'fast clones' skbs for packets in tx queues.
Currently, __alloc_skb() initializes the companion fclone
field to SKB_FCLONE_CLONE, and leaves other fields untouched.
It makes sense to defer this init much later in skb_clone(),
because all fclone fields are copied and hot in cpu caches
at that time.
This removes one cache line miss in __alloc_skb(), cost seen
on an host with 256 cpus all competing on memory accesses.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Move the len fields manipulation in the skbs to a helper function.
There is a comment specifically requesting this and there are several
other areas in the code displaying the same pattern which can be
refactored.
This improves code readability.
Signed-off-by: Richard Gobert <richardbgobert@gmail.com>
Link: https://lore.kernel.org/r/20220622160853.GA6478@debian
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
NAPI cache skb_count is being checked twice without condition. Change to
checking the second time only if the first check is run.
Signed-off-by: Sieng Piaw Liew <liew.s.piaw@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Commit 6454eca81e ("net: Use lockdep_assert_in_softirq()
in napi_consume_skb()") added a check in napi_consume_skb()
which is a bit weak.
napi_consume_skb() and __napi_alloc_skb() should only
be used from BH context, not from hard irq or nmi context,
otherwise we could have races.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Remove this check from fast path unless CONFIG_DEBUG_NET=y
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
It is annoying to add new skb drop reasons to 'enum skb_drop_reason'
and TRACE_SKB_DROP_REASON in trace/event/skb.h, and it's easy to forget
to add the new reasons we added to TRACE_SKB_DROP_REASON.
TRACE_SKB_DROP_REASON is used to convert drop reason of type number
to string. For now, the string we passed to user space is exactly the
same as the name in 'enum skb_drop_reason' with a 'SKB_DROP_REASON_'
prefix. Therefore, we can use 'auto-generation' to generate these
drop reasons to string at build time.
The new source 'dropreason_str.c' will be auto generated during build
time, which contains the string array
'const char * const drop_reasons[]'.
Signed-off-by: Menglong Dong <imagedong@tencent.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
When user sets skb_defer_max to 1 the kick threshold is 0
(half of 1). If we increment queue length before the check
the kick will never happen, and the skb may get stranded.
This is likely harmless but can be avoided by moving the
increment after the check. This way skb_defer_max == 1
will always kick. Still a silly config to have, but
somehow that feels more correct.
While at it drop a comment which seems to be outdated
or confusing, and wrap the defer_count write with
a WRITE_ONCE() since it's read on the fast path
that avoids taking the lock.
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/20220518185522.2038683-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
commit 68822bdf76 ("net: generalize skb freeing
deferral to per-cpu lists") added another per-cpu
cache of skbs. It was expected to be small,
and an IPI was forced whenever the list reached 128
skbs.
We might need to be able to control more precisely
queue capacity and added latency.
An IPI is generated whenever queue reaches half capacity.
Default value of the new limit is 64.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
A cpu can observe sd->defer_count reaching 128,
and call smp_call_function_single_async()
Problem is that the remote CPU can clear sd->defer_count
before the IPI is run/acknowledged.
Other cpus can queue more packets and also decide
to call smp_call_function_single_async() while the pending
IPI was not yet delivered.
This is a common issue with smp_call_function_single_async().
Callers must ensure correct synchronization and serialization.
I triggered this issue while experimenting smaller threshold.
Performing the call to smp_call_function_single_async()
under sd->defer_lock protection did not solve the problem.
Commit 5a18ceca63 ("smp: Allow smp_call_function_single_async()
to insert locked csd") replaced an informative WARN_ON_ONCE()
with a return of -EBUSY, which is often ignored.
Test of CSD_FLAG_LOCK presence is racy anyway.
Fixes: 68822bdf76 ("net: generalize skb freeing deferral to per-cpu lists")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Sometimes, we may forget to reset skb drop reason to NOT_SPECIFIED after
we make it the return value of the functions with return type of enum
skb_drop_reason, such as tcp_inbound_md5_hash. Therefore, its value can
be SKB_NOT_DROPPED_YET(0), which is invalid for kfree_skb_reason().
So we check the range of drop reason in kfree_skb_reason() with
DEBUG_NET_WARN_ON_ONCE().
Reviewed-by: Jiang Biao <benbjiang@tencent.com>
Reviewed-by: Hao Peng <flyingpeng@tencent.com>
Signed-off-by: Menglong Dong <imagedong@tencent.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When clatd starts with ebpf offloaing, and NETIF_F_GRO_FRAGLIST is enable,
several skbs are gathered in skb_shinfo(skb)->frag_list. The first skb's
ipv6 header will be changed to ipv4 after bpf_skb_proto_6_to_4,
network_header\transport_header\mac_header have been updated as ipv4 acts,
but other skbs in frag_list didnot update anything, just ipv6 packets.
udp_queue_rcv_skb will call skb_segment_list to traverse other skbs in
frag_list and make sure right udp payload is delivered to user space.
Unfortunately, other skbs in frag_list who are still ipv6 packets are
updated like the first skb and will have wrong transport header length.
e.g.before bpf_skb_proto_6_to_4,the first skb and other skbs in frag_list
has the same network_header(24)& transport_header(64), after
bpf_skb_proto_6_to_4, ipv6 protocol has been changed to ipv4, the first
skb's network_header is 44,transport_header is 64, other skbs in frag_list
didnot change.After skb_segment_list, the other skbs in frag_list has
different network_header(24) and transport_header(44), so there will be 20
bytes different from original,that is difference between ipv6 header and
ipv4 header. Just change transport_header to be the same with original.
Actually, there are two solutions to fix it, one is traversing all skbs
and changing every skb header in bpf_skb_proto_6_to_4, the other is
modifying frag_list skb's header in skb_segment_list. Considering
efficiency, adopt the second one--- when the first skb and other skbs in
frag_list has different network_header length, restore them to make sure
right udp payload is delivered to user space.
Signed-off-by: Lina Wang <lina.wang@mediatek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
msg_zerocopy_alloc is only used by msg_zerocopy_realloc; remove the
export and make static in skbuff.c
Signed-off-by: David Ahern <dsahern@kernel.org>
Acked-by: Jonathan Lemon <jonathan.lemon@gmail.com>
Link: https://lore.kernel.org/r/20220504170947.18773-1-dsahern@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
It's expensive to make a copy of 40B struct iov_iter to the point it
was taking 0.2-0.5% of all cycles in my tests. iov_iter_revert() should
be fine as it's a simple case without nested reverts/truncates.
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://lore.kernel.org/r/a7e1690c00c5dfe700c30eb9a8a81ec59f6545dd.1650884401.git.asml.silence@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
skb_zerocopy_iter_dgram() is a small proxy function, inline it. For
that, move __zerocopy_sg_from_iter into linux/skbuff.h
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Daniel Borkmann says:
====================
pull-request: bpf-next 2022-04-27
We've added 85 non-merge commits during the last 18 day(s) which contain
a total of 163 files changed, 4499 insertions(+), 1521 deletions(-).
The main changes are:
1) Teach libbpf to enhance BPF verifier log with human-readable and relevant
information about failed CO-RE relocations, from Andrii Nakryiko.
2) Add typed pointer support in BPF maps and enable it for unreferenced pointers
(via probe read) and referenced ones that can be passed to in-kernel helpers,
from Kumar Kartikeya Dwivedi.
3) Improve xsk to break NAPI loop when rx queue gets full to allow for forward
progress to consume descriptors, from Maciej Fijalkowski & Björn Töpel.
4) Fix a small RCU read-side race in BPF_PROG_RUN routines which dereferenced
the effective prog array before the rcu_read_lock, from Stanislav Fomichev.
5) Implement BPF atomic operations for RV64 JIT, and add libbpf parsing logic
for USDT arguments under riscv{32,64}, from Pu Lehui.
6) Implement libbpf parsing of USDT arguments under aarch64, from Alan Maguire.
7) Enable bpftool build for musl and remove nftw with FTW_ACTIONRETVAL usage
so it can be shipped under Alpine which is musl-based, from Dominique Martinet.
8) Clean up {sk,task,inode} local storage trace RCU handling as they do not
need to use call_rcu_tasks_trace() barrier, from KP Singh.
9) Improve libbpf API documentation and fix error return handling of various
API functions, from Grant Seltzer.
10) Enlarge offset check for bpf_skb_{load,store}_bytes() helpers given data
length of frags + frag_list may surpass old offset limit, from Liu Jian.
11) Various improvements to prog_tests in area of logging, test execution
and by-name subtest selection, from Mykola Lysenko.
12) Simplify map_btf_id generation for all map types by moving this process
to build time with help of resolve_btfids infra, from Menglong Dong.
13) Fix a libbpf bug in probing when falling back to legacy bpf_probe_read*()
helpers; the probing caused always to use old helpers, from Runqing Yang.
14) Add support for ARCompact and ARCv2 platforms for libbpf's PT_REGS
tracing macros, from Vladimir Isaev.
15) Cleanup BPF selftests to remove old & unneeded rlimit code given kernel
switched to memcg-based memory accouting a while ago, from Yafang Shao.
16) Refactor of BPF sysctl handlers to move them to BPF core, from Yan Zhu.
17) Fix BPF selftests in two occasions to work around regressions caused by latest
LLVM to unblock CI until their fixes are worked out, from Yonghong Song.
18) Misc cleanups all over the place, from various others.
* https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next: (85 commits)
selftests/bpf: Add libbpf's log fixup logic selftests
libbpf: Fix up verifier log for unguarded failed CO-RE relos
libbpf: Simplify bpf_core_parse_spec() signature
libbpf: Refactor CO-RE relo human description formatting routine
libbpf: Record subprog-resolved CO-RE relocations unconditionally
selftests/bpf: Add CO-RE relos and SEC("?...") to linked_funcs selftests
libbpf: Avoid joining .BTF.ext data with BPF programs by section name
libbpf: Fix logic for finding matching program for CO-RE relocation
libbpf: Drop unhelpful "program too large" guess
libbpf: Fix anonymous type check in CO-RE logic
bpf: Compute map_btf_id during build time
selftests/bpf: Add test for strict BTF type check
selftests/bpf: Add verifier tests for kptr
selftests/bpf: Add C tests for kptr
libbpf: Add kptr type tag macros to bpf_helpers.h
bpf: Make BTF type match stricter for release arguments
bpf: Teach verifier about kptr_get kfunc helpers
bpf: Wire up freeing of referenced kptr
bpf: Populate pairs of btf_id and destructor kfunc in btf
bpf: Adapt copy_map_value for multiple offset case
...
====================
Link: https://lore.kernel.org/r/20220427224758.20976-1-daniel@iogearbox.net
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Logic added in commit f35f821935 ("tcp: defer skb freeing after socket
lock is released") helped bulk TCP flows to move the cost of skbs
frees outside of critical section where socket lock was held.
But for RPC traffic, or hosts with RFS enabled, the solution is far from
being ideal.
For RPC traffic, recvmsg() has to return to user space right after
skb payload has been consumed, meaning that BH handler has no chance
to pick the skb before recvmsg() thread. This issue is more visible
with BIG TCP, as more RPC fit one skb.
For RFS, even if BH handler picks the skbs, they are still picked
from the cpu on which user thread is running.
Ideally, it is better to free the skbs (and associated page frags)
on the cpu that originally allocated them.
This patch removes the per socket anchor (sk->defer_list) and
instead uses a per-cpu list, which will hold more skbs per round.
This new per-cpu list is drained at the end of net_action_rx(),
after incoming packets have been processed, to lower latencies.
In normal conditions, skbs are added to the per-cpu list with
no further action. In the (unlikely) cases where the cpu does not
run net_action_rx() handler fast enough, we use an IPI to raise
NET_RX_SOFTIRQ on the remote cpu.
Also, we do not bother draining the per-cpu list from dev_cpu_dead()
This is because skbs in this list have no requirement on how fast
they should be freed.
Note that we can add in the future a small per-cpu cache
if we see any contention on sd->defer_lock.
Tested on a pair of hosts with 100Gbit NIC, RFS enabled,
and /proc/sys/net/ipv4/tcp_rmem[2] tuned to 16MB to work around
page recycling strategy used by NIC driver (its page pool capacity
being too small compared to number of skbs/pages held in sockets
receive queues)
Note that this tuning was only done to demonstrate worse
conditions for skb freeing for this particular test.
These conditions can happen in more general production workload.
10 runs of one TCP_STREAM flow
Before:
Average throughput: 49685 Mbit.
Kernel profiles on cpu running user thread recvmsg() show high cost for
skb freeing related functions (*)
57.81% [kernel] [k] copy_user_enhanced_fast_string
(*) 12.87% [kernel] [k] skb_release_data
(*) 4.25% [kernel] [k] __free_one_page
(*) 3.57% [kernel] [k] __list_del_entry_valid
1.85% [kernel] [k] __netif_receive_skb_core
1.60% [kernel] [k] __skb_datagram_iter
(*) 1.59% [kernel] [k] free_unref_page_commit
(*) 1.16% [kernel] [k] __slab_free
1.16% [kernel] [k] _copy_to_iter
(*) 1.01% [kernel] [k] kfree
(*) 0.88% [kernel] [k] free_unref_page
0.57% [kernel] [k] ip6_rcv_core
0.55% [kernel] [k] ip6t_do_table
0.54% [kernel] [k] flush_smp_call_function_queue
(*) 0.54% [kernel] [k] free_pcppages_bulk
0.51% [kernel] [k] llist_reverse_order
0.38% [kernel] [k] process_backlog
(*) 0.38% [kernel] [k] free_pcp_prepare
0.37% [kernel] [k] tcp_recvmsg_locked
(*) 0.37% [kernel] [k] __list_add_valid
0.34% [kernel] [k] sock_rfree
0.34% [kernel] [k] _raw_spin_lock_irq
(*) 0.33% [kernel] [k] __page_cache_release
0.33% [kernel] [k] tcp_v6_rcv
(*) 0.33% [kernel] [k] __put_page
(*) 0.29% [kernel] [k] __mod_zone_page_state
0.27% [kernel] [k] _raw_spin_lock
After patch:
Average throughput: 73076 Mbit.
Kernel profiles on cpu running user thread recvmsg() looks better:
81.35% [kernel] [k] copy_user_enhanced_fast_string
1.95% [kernel] [k] _copy_to_iter
1.95% [kernel] [k] __skb_datagram_iter
1.27% [kernel] [k] __netif_receive_skb_core
1.03% [kernel] [k] ip6t_do_table
0.60% [kernel] [k] sock_rfree
0.50% [kernel] [k] tcp_v6_rcv
0.47% [kernel] [k] ip6_rcv_core
0.45% [kernel] [k] read_tsc
0.44% [kernel] [k] _raw_spin_lock_irqsave
0.37% [kernel] [k] _raw_spin_lock
0.37% [kernel] [k] native_irq_return_iret
0.33% [kernel] [k] __inet6_lookup_established
0.31% [kernel] [k] ip6_protocol_deliver_rcu
0.29% [kernel] [k] tcp_rcv_established
0.29% [kernel] [k] llist_reverse_order
v2: kdoc issue (kernel bots)
do not defer if (alloc_cpu == smp_processor_id()) (Paolo)
replace the sk_buff_head with a single-linked list (Jakub)
add a READ_ONCE()/WRITE_ONCE() for the lockless read of sd->defer_list
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Paolo Abeni <pabeni@redhat.com>
Link: https://lore.kernel.org/r/20220422201237.416238-1-eric.dumazet@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Both pskb_may_pull() and skb_clone_writable()'s length parameters are of
type unsigned int already. Therefore, change this function's write_len
param to unsigned int type.
Signed-off-by: Liu Jian <liujian56@huawei.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Song Liu <songliubraving@fb.com>
Link: https://lore.kernel.org/bpf/20220416105801.88708-3-liujian56@huawei.com
Fix a use-after-free when using page_pool with page fragments. We
encountered this problem during normal RX in the hns3 driver:
(1) Initially we have three descriptors in the RX queue. The first one
allocates PAGE1 through page_pool, and the other two allocate one
half of PAGE2 each. Page references look like this:
RX_BD1 _______ PAGE1
RX_BD2 _______ PAGE2
RX_BD3 _________/
(2) Handle RX on the first descriptor. Allocate SKB1, eventually added
to the receive queue by tcp_queue_rcv().
(3) Handle RX on the second descriptor. Allocate SKB2 and pass it to
netif_receive_skb():
netif_receive_skb(SKB2)
ip_rcv(SKB2)
SKB3 = skb_clone(SKB2)
SKB2 and SKB3 share a reference to PAGE2 through
skb_shinfo()->dataref. The other ref to PAGE2 is still held by
RX_BD3:
SKB2 ---+- PAGE2
SKB3 __/ /
RX_BD3 _________/
(3b) Now while handling TCP, coalesce SKB3 with SKB1:
tcp_v4_rcv(SKB3)
tcp_try_coalesce(to=SKB1, from=SKB3) // succeeds
kfree_skb_partial(SKB3)
skb_release_data(SKB3) // drops one dataref
SKB1 _____ PAGE1
\____
SKB2 _____ PAGE2
/
RX_BD3 _________/
In skb_try_coalesce(), __skb_frag_ref() takes a page reference to
PAGE2, where it should instead have increased the page_pool frag
reference, pp_frag_count. Without coalescing, when releasing both
SKB2 and SKB3, a single reference to PAGE2 would be dropped. Now
when releasing SKB1 and SKB2, two references to PAGE2 will be
dropped, resulting in underflow.
(3c) Drop SKB2:
af_packet_rcv(SKB2)
consume_skb(SKB2)
skb_release_data(SKB2) // drops second dataref
page_pool_return_skb_page(PAGE2) // drops one pp_frag_count
SKB1 _____ PAGE1
\____
PAGE2
/
RX_BD3 _________/
(4) Userspace calls recvmsg()
Copies SKB1 and releases it. Since SKB3 was coalesced with SKB1, we
release the SKB3 page as well:
tcp_eat_recv_skb(SKB1)
skb_release_data(SKB1)
page_pool_return_skb_page(PAGE1)
page_pool_return_skb_page(PAGE2) // drops second pp_frag_count
(5) PAGE2 is freed, but the third RX descriptor was still using it!
In our case this causes IOMMU faults, but it would silently corrupt
memory if the IOMMU was disabled.
Change the logic that checks whether pp_recycle SKBs can be coalesced.
We still reject differing pp_recycle between 'from' and 'to' SKBs, but
in order to avoid the situation described above, we also reject
coalescing when both 'from' and 'to' are pp_recycled and 'from' is
cloned.
The new logic allows coalescing a cloned pp_recycle SKB into a page
refcounted one, because in this case the release (4) will drop the right
reference, the one taken by skb_try_coalesce().
Fixes: 53e0961da1 ("page_pool: add frag page recycling support in page pool")
Suggested-by: Alexander Duyck <alexanderduyck@fb.com>
Signed-off-by: Jean-Philippe Brucker <jean-philippe@linaro.org>
Reviewed-by: Yunsheng Lin <linyunsheng@huawei.com>
Reviewed-by: Alexander Duyck <alexanderduyck@fb.com>
Acked-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
Acked-by: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
To report reasons of skb drops, introduce the function
kfree_skb_list_reason() and make kfree_skb_list() an inline call to
it. This function will be used in the next commit in
__dev_xmit_skb().
Signed-off-by: Menglong Dong <imagedong@tencent.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In __skb_tstamp_tx(), it may clone the egress skb and queues the clone to
the sk_error_queue. The outgoing skb may have the mono delivery_time
while the (rcv) timestamp is expected for the clone, so the
skb->mono_delivery_time bit needs to be cleared from the clone.
This patch adds the skb->mono_delivery_time clearing to the existing
__net_timestamp() and use it in __skb_tstamp_tx().
The __net_timestamp() fast path usage in dev.c is changed to directly
call ktime_get_real() since the mono_delivery_time bit is not set at
that point.
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Right now, skb->tstamp is reset to 0 whenever the skb is forwarded.
If skb->tstamp has the mono delivery_time, clearing it can hurt
the performance when it finally transmits out to fq@phy-dev.
The earlier patch added a skb->mono_delivery_time bit to
flag the skb->tstamp carrying the mono delivery_time.
This patch adds skb_clear_tstamp() helper which keeps
the mono delivery_time and clears everything else.
The delivery_time clearing will be postponed until the stack knows the
skb will be delivered locally. It will be done in a latter patch.
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The truesize for a UDP GRO packet is added by main skb and skbs in main
skb's frag_list:
skb_gro_receive_list
p->truesize += skb->truesize;
The commit 53475c5dd8 ("net: fix use-after-free when UDP GRO with
shared fraglist") introduced a truesize increase for frag_list skbs.
When uncloning skb, it will call pskb_expand_head and trusesize for
frag_list skbs may increase. This can occur when allocators uses
__netdev_alloc_skb and not jump into __alloc_skb. This flow does not
use ksize(len) to calculate truesize while pskb_expand_head uses.
skb_segment_list
err = skb_unclone(nskb, GFP_ATOMIC);
pskb_expand_head
if (!skb->sk || skb->destructor == sock_edemux)
skb->truesize += size - osize;
If we uses increased truesize adding as delta_truesize, it will be
larger than before and even larger than previous total truesize value
if skbs in frag_list are abundant. The main skb truesize will become
smaller and even a minus value or a huge value for an unsigned int
parameter. Then the following memory check will drop this abnormal skb.
To avoid this error we should use the original truesize to segment the
main skb.
Fixes: 53475c5dd8 ("net: fix use-after-free when UDP GRO with shared fraglist")
Signed-off-by: lena wang <lena.wang@mediatek.com>
Acked-by: Paolo Abeni <pabeni@redhat.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/1646133431-8948-1-git-send-email-lena.wang@mediatek.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
syzbot found another way to trigger the infamous WARN_ON_ONCE(delta < len)
in skb_try_coalesce() [1]
I was able to root cause the issue to kfence.
When kfence is in action, the following assertion is no longer true:
int size = xxxx;
void *ptr1 = kmalloc(size, gfp);
void *ptr2 = kmalloc(size, gfp);
if (ptr1 && ptr2)
ASSERT(ksize(ptr1) == ksize(ptr2));
We attempted to fix these issues in the blamed commits, but forgot
that TCP was possibly shifting data after skb_unclone_keeptruesize()
has been used, notably from tcp_retrans_try_collapse().
So we not only need to keep same skb->truesize value,
we also need to make sure TCP wont fill new tailroom
that pskb_expand_head() was able to get from a
addr = kmalloc(...) followed by ksize(addr)
Split skb_unclone_keeptruesize() into two parts:
1) Inline skb_unclone_keeptruesize() for the common case,
when skb is not cloned.
2) Out of line __skb_unclone_keeptruesize() for the 'slow path'.
WARNING: CPU: 1 PID: 6490 at net/core/skbuff.c:5295 skb_try_coalesce+0x1235/0x1560 net/core/skbuff.c:5295
Modules linked in:
CPU: 1 PID: 6490 Comm: syz-executor161 Not tainted 5.17.0-rc4-syzkaller-00229-g4f12b742eb2b #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
RIP: 0010:skb_try_coalesce+0x1235/0x1560 net/core/skbuff.c:5295
Code: bf 01 00 00 00 0f b7 c0 89 c6 89 44 24 20 e8 62 24 4e fa 8b 44 24 20 83 e8 01 0f 85 e5 f0 ff ff e9 87 f4 ff ff e8 cb 20 4e fa <0f> 0b e9 06 f9 ff ff e8 af b2 95 fa e9 69 f0 ff ff e8 95 b2 95 fa
RSP: 0018:ffffc900063af268 EFLAGS: 00010293
RAX: 0000000000000000 RBX: 00000000ffffffd5 RCX: 0000000000000000
RDX: ffff88806fc05700 RSI: ffffffff872abd55 RDI: 0000000000000003
RBP: ffff88806e675500 R08: 00000000ffffffd5 R09: 0000000000000000
R10: ffffffff872ab659 R11: 0000000000000000 R12: ffff88806dd554e8
R13: ffff88806dd9bac0 R14: ffff88806dd9a2c0 R15: 0000000000000155
FS: 00007f18014f9700(0000) GS:ffff8880b9c00000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000000020002000 CR3: 000000006be7a000 CR4: 00000000003506f0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Call Trace:
<TASK>
tcp_try_coalesce net/ipv4/tcp_input.c:4651 [inline]
tcp_try_coalesce+0x393/0x920 net/ipv4/tcp_input.c:4630
tcp_queue_rcv+0x8a/0x6e0 net/ipv4/tcp_input.c:4914
tcp_data_queue+0x11fd/0x4bb0 net/ipv4/tcp_input.c:5025
tcp_rcv_established+0x81e/0x1ff0 net/ipv4/tcp_input.c:5947
tcp_v4_do_rcv+0x65e/0x980 net/ipv4/tcp_ipv4.c:1719
sk_backlog_rcv include/net/sock.h:1037 [inline]
__release_sock+0x134/0x3b0 net/core/sock.c:2779
release_sock+0x54/0x1b0 net/core/sock.c:3311
sk_wait_data+0x177/0x450 net/core/sock.c:2821
tcp_recvmsg_locked+0xe28/0x1fd0 net/ipv4/tcp.c:2457
tcp_recvmsg+0x137/0x610 net/ipv4/tcp.c:2572
inet_recvmsg+0x11b/0x5e0 net/ipv4/af_inet.c:850
sock_recvmsg_nosec net/socket.c:948 [inline]
sock_recvmsg net/socket.c:966 [inline]
sock_recvmsg net/socket.c:962 [inline]
____sys_recvmsg+0x2c4/0x600 net/socket.c:2632
___sys_recvmsg+0x127/0x200 net/socket.c:2674
__sys_recvmsg+0xe2/0x1a0 net/socket.c:2704
do_syscall_x64 arch/x86/entry/common.c:50 [inline]
do_syscall_64+0x35/0xb0 arch/x86/entry/common.c:80
entry_SYSCALL_64_after_hwframe+0x44/0xae
Fixes: c4777efa75 ("net: add and use skb_unclone_keeptruesize() helper")
Fixes: 097b9146c0 ("net: fix up truesize of cloned skb in skb_prepare_for_shift()")
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Marco Elver <elver@google.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
We have multiple places where this helper is convenient,
and plan using it in the following patch.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Whenever one of these functions pull all data from an skb in a frag_list,
use consume_skb() instead of kfree_skb() to avoid polluting drop
monitoring.
Fixes: 6fa01ccd88 ("skbuff: Add pskb_extract() helper function")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/20220220154052.1308469-1-eric.dumazet@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
UDP sendmsg() can be lockless, this is causing all kinds
of data races.
This patch converts sk->sk_tskey to remove one of these races.
BUG: KCSAN: data-race in __ip_append_data / __ip_append_data
read to 0xffff8881035d4b6c of 4 bytes by task 8877 on cpu 1:
__ip_append_data+0x1c1/0x1de0 net/ipv4/ip_output.c:994
ip_make_skb+0x13f/0x2d0 net/ipv4/ip_output.c:1636
udp_sendmsg+0x12bd/0x14c0 net/ipv4/udp.c:1249
inet_sendmsg+0x5f/0x80 net/ipv4/af_inet.c:819
sock_sendmsg_nosec net/socket.c:705 [inline]
sock_sendmsg net/socket.c:725 [inline]
____sys_sendmsg+0x39a/0x510 net/socket.c:2413
___sys_sendmsg net/socket.c:2467 [inline]
__sys_sendmmsg+0x267/0x4c0 net/socket.c:2553
__do_sys_sendmmsg net/socket.c:2582 [inline]
__se_sys_sendmmsg net/socket.c:2579 [inline]
__x64_sys_sendmmsg+0x53/0x60 net/socket.c:2579
do_syscall_x64 arch/x86/entry/common.c:50 [inline]
do_syscall_64+0x44/0xd0 arch/x86/entry/common.c:80
entry_SYSCALL_64_after_hwframe+0x44/0xae
write to 0xffff8881035d4b6c of 4 bytes by task 8880 on cpu 0:
__ip_append_data+0x1d8/0x1de0 net/ipv4/ip_output.c:994
ip_make_skb+0x13f/0x2d0 net/ipv4/ip_output.c:1636
udp_sendmsg+0x12bd/0x14c0 net/ipv4/udp.c:1249
inet_sendmsg+0x5f/0x80 net/ipv4/af_inet.c:819
sock_sendmsg_nosec net/socket.c:705 [inline]
sock_sendmsg net/socket.c:725 [inline]
____sys_sendmsg+0x39a/0x510 net/socket.c:2413
___sys_sendmsg net/socket.c:2467 [inline]
__sys_sendmmsg+0x267/0x4c0 net/socket.c:2553
__do_sys_sendmmsg net/socket.c:2582 [inline]
__se_sys_sendmmsg net/socket.c:2579 [inline]
__x64_sys_sendmmsg+0x53/0x60 net/socket.c:2579
do_syscall_x64 arch/x86/entry/common.c:50 [inline]
do_syscall_64+0x44/0xd0 arch/x86/entry/common.c:80
entry_SYSCALL_64_after_hwframe+0x44/0xae
value changed: 0x0000054d -> 0x0000054e
Reported by Kernel Concurrency Sanitizer on:
CPU: 0 PID: 8880 Comm: syz-executor.5 Not tainted 5.17.0-rc2-syzkaller-00167-gdcb85f85fa6f-dirty #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Fixes: 09c2d251b7 ("net-timestamp: add key to disambiguate concurrent datagrams")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Willem de Bruijn <willemb@google.com>
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Introduce the interface kfree_skb_reason(), which is able to pass
the reason why the skb is dropped to 'kfree_skb' tracepoint.
Add the 'reason' field to 'trace_kfree_skb', therefor user can get
more detail information about abnormal skb with 'drop_monitor' or
eBPF.
All drop reasons are defined in the enum 'skb_drop_reason', and
they will be print as string in 'kfree_skb' tracepoint in format
of 'reason: XXX'.
( Maybe the reasons should be defined in a uapi header file, so that
user space can use them? )
Signed-off-by: Menglong Dong <imagedong@tencent.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
- Add support for Foxconn MT7922A
- Add support for Realtek RTL8852AE
- Rework HCI event handling to use skb_pull_data
-----BEGIN PGP SIGNATURE-----
iQJNBAABCAA3FiEE7E6oRXp8w05ovYr/9JCA4xAyCykFAmHMzl8ZHGx1aXoudm9u
LmRlbnR6QGludGVsLmNvbQAKCRD0kIDjEDILKSpcD/0dP8yVmlPztQwV/O/DjIHc
rVCplr+pzdOcuW4qXDsKhDt9y5E6KgIhIqnksHV3fM58yY1L6sm9n0Qpihys0mIS
GHQ7VtqCE8WMpifkCTqdKV34Guy9BP0GGQpWTh7jNTeiwHT1t5pjA6ajYE3qaSS/
z4opVFz/eGqinPLPxjP7GBP4VqaLG0QMNo3pSz0hi+E/krIwnjT+VoKNKteDkHf9
klmHrGF2OYtXC8VezvquFgmUgoN+yL/63Ci1DlkF3mRUEc6jMHFMs7BiQov2eENt
Hq9E1ye5Hlf5IedtJ80QdkLvW/JgE1jZKATB7x7E09j9QBP2gxJRP9z6sSJiVIVm
53MRRYaA9jOPCfaVPKGMp+4NQMa7F2gahoI1k7LVS0ksc0RGC/nixtdiZWSjwU6E
cYiQOk5CD3hK+lPSYBvjMOnm+h4w4FZ5W70g5VKKnAGCUBvERSwh2qmUdpG4j1vq
4oeg64DsLP5n7MWbHn7zbtviiJVbMg+a0D4GlExlxu06OOx7yra3s+LbcHoZbG4B
T3ivhx9oX+ho7fG3rjBPrENoZB62n48x/mzxsLyFHt/m3uSgsXVSCVIm9dwwsdBM
P2OzX/dSx/3AhnH0wNPtENAoU4/ETuMOKuFCoU9pL0cjJbQ3M9FipoZakzF/B1se
18+x0pq2o4PT9W1JUPMfvA==
=AVll
-----END PGP SIGNATURE-----
Merge tag 'for-net-next-2021-12-29' of git://git.kernel.org/pub/scm/linux/kernel/git/bluetooth/bluetooth-next
Luiz Augusto von Dentz says:
====================
bluetooth-next pull request for net-next:
- Add support for Foxconn MT7922A
- Add support for Realtek RTL8852AE
- Rework HCI event handling to use skb_pull_data
* tag 'for-net-next-2021-12-29' of git://git.kernel.org/pub/scm/linux/kernel/git/bluetooth/bluetooth-next: (62 commits)
Bluetooth: MGMT: Fix spelling mistake "simultanous" -> "simultaneous"
Bluetooth: vhci: Set HCI_QUIRK_VALID_LE_STATES
Bluetooth: MGMT: Fix LE simultaneous roles UUID if not supported
Bluetooth: hci_sync: Add check simultaneous roles support
Bluetooth: hci_sync: Wait for proper events when connecting LE
Bluetooth: hci_sync: Add support for waiting specific LE subevents
Bluetooth: hci_sync: Add hci_le_create_conn_sync
Bluetooth: hci_event: Use skb_pull_data when processing inquiry results
Bluetooth: hci_sync: Push sync command cancellation to workqueue
Bluetooth: hci_qca: Stop IBS timer during BT OFF
Bluetooth: btusb: Add support for Foxconn MT7922A
Bluetooth: btintel: Add missing quirks and msft ext for legacy bootloader
Bluetooth: btusb: Add two more Bluetooth parts for WCN6855
Bluetooth: L2CAP: Fix using wrong mode
Bluetooth: hci_sync: Fix not always pausing advertising when necessary
Bluetooth: mgmt: Make use of mgmt_send_event_skb in MGMT_EV_DEVICE_CONNECTED
Bluetooth: mgmt: Make use of mgmt_send_event_skb in MGMT_EV_DEVICE_FOUND
Bluetooth: mgmt: Introduce mgmt_alloc_skb and mgmt_send_event_skb
Bluetooth: btusb: Return error code when getting patch status failed
Bluetooth: btusb: Handle download_firmware failure cases
...
====================
Link: https://lore.kernel.org/r/20211229211258.2290966-1-luiz.dentz@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
When printing netdev features %pNF already takes care of the 0x prefix,
remove the explicit one.
Fixes: 6413139dfc ("skbuff: increase verbosity when dumping skb data")
Signed-off-by: Gal Pressman <gal@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Like skb_pull but returns the original data pointer before pulling the
data after performing a check against sbk->len.
This allows to change code that does "struct foo *p = (void *)skb->data;"
which is hard to audit and error prone, to:
p = skb_pull_data(skb, sizeof(*p));
if (!p)
return;
Which is both safer and cleaner.
Acked-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
In preparation for FORTIFY_SOURCE performing compile-time and run-time
field bounds checking for memcpy(), memmove(), and memset(), avoid
intentionally writing across neighboring fields.
Replace the existing empty member position markers "headers_start" and
"headers_end" with a struct_group(). This will allow memcpy() and sizeof()
to more easily reason about sizes, and improve readability.
"pahole" shows no size nor member offset changes to struct sk_buff.
"objdump -d" shows no object code changes (outside of WARNs affected by
source line number changes).
Signed-off-by: Kees Cook <keescook@chromium.org>
Reviewed-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Reviewed-by: Jason A. Donenfeld <Jason@zx2c4.com> # drivers/net/wireguard/*
Link: https://lore.kernel.org/lkml/20210728035006.GD35706@embeddedor
Signed-off-by: David S. Miller <davem@davemloft.net>
net/core/gro.c will contain all core gro functions,
to shrink net/core/skbuff.c and net/core/dev.c
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This helper is used once, no need to keep it in fat net/core/skbuff.c
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
include/linux/netdevice.h became too big, move gro stuff
into include/net/gro.h
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Move sk_is_tcp() to include/net/sock.h and use it where we can.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Track skbs containing only zerocopy data and avoid charging them to
kernel memory to correctly account the memory utilization for
msg_zerocopy. All of the data in such skbs is held in user pages which
are already accounted to user. Before this change, they are charged
again in kernel in __zerocopy_sg_from_iter. The charging in kernel is
excessive because data is not being copied into skb frags. This
excessive charging can lead to kernel going into memory pressure
state which impacts all sockets in the system adversely. Mark pure
zerocopy skbs with a SKBFL_PURE_ZEROCOPY flag and remove
charge/uncharge for data in such skbs.
Initially, an skb is marked pure zerocopy when it is empty and in
zerocopy path. skb can then change from a pure zerocopy skb to mixed
data skb (zerocopy and copy data) if it is at tail of write queue and
there is room available in it and non-zerocopy data is being sent in
the next sendmsg call. At this time sk_mem_charge is done for the pure
zerocopied data and the pure zerocopy flag is unmarked. We found that
this happens very rarely on workloads that pass MSG_ZEROCOPY.
A pure zerocopy skb can later be coalesced into normal skb if they are
next to each other in queue but this patch prevents coalescing from
happening. This avoids complexity of charging when skb downgrades from
pure zerocopy to mixed. This is also rare.
In sk_wmem_free_skb, if it is a pure zerocopy skb, an sk_mem_uncharge
for SKB_TRUESIZE(skb_end_offset(skb)) is done for sk_mem_charge in
tcp_skb_entail for an skb without data.
Testing with the msg_zerocopy.c benchmark between two hosts(100G nics)
with zerocopy showed that before this patch the 'sock' variable in
memory.stat for cgroup2 that tracks sum of sk_forward_alloc,
sk_rmem_alloc and sk_wmem_queued is around 1822720 and with this
change it is 0. This is due to no charge to sk_forward_alloc for
zerocopy data and shows memory utilization for kernel is lowered.
With this commit we don't see the warning we saw in previous commit
which resulted in commit 84882cf72c.
Signed-off-by: Talal Ahmad <talalahmad@google.com>
Acked-by: Arjun Roy <arjunroy@google.com>
Acked-by: Soheil Hassas Yeganeh <soheil@google.com>
Signed-off-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
While commit 097b9146c0 ("net: fix up truesize of cloned
skb in skb_prepare_for_shift()") fixed immediate issues found
when KFENCE was enabled/tested, there are still similar issues,
when tcp_trim_head() hits KFENCE while the master skb
is cloned.
This happens under heavy networking TX workloads,
when the TX completion might be delayed after incoming ACK.
This patch fixes the WARNING in sk_stream_kill_queues
when sk->sk_mem_queued/sk->sk_forward_alloc are not zero.
Fixes: d3fb45f370 ("mm, kfence: insert KFENCE hooks for SLAB")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Marco Elver <elver@google.com>
Link: https://lore.kernel.org/r/20211102004555.1359210-1-eric.dumazet@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Menglong Dong
Kuniyuki Iwashima
lily
Jakub Kicinski
Pavel Begunkov
David Ahern
Eric Dumazet
Richard Gobert
Sieng Piaw Liew
Lina Wang
Liu Jian
Jean-Philippe Brucker
Martin KaFai Lau
lena wang
Tom Rix
Gal Pressman
Luiz Augusto von Dentz
Kees Cook
Talal Ahmad