| Age | Commit message (Collapse) | Author |
|---|
|
TLS is a relatively poor fit for strparser. We pause the input
every time a message is received, wait for a read which will
decrypt the message, start the parser, repeat. strparser is
built to delineate the messages, wrap them in individual skbs
and let them float off into the stack or a different socket.
TLS wants the data pages and nothing else. There's no need
for TLS to keep cloning (and occasionally skb_unclone()'ing)
the TCP rx queue.
This patch uses a pre-allocated skb and attaches the skbs
from the TCP rx queue to it as frags. TLS is careful never
to modify the input skb without CoW'ing / detaching it first.
Since we call TCP rx queue cleanup directly we also get back
the benefit of skb deferred free.
Overall this results in a 6% gain in my benchmarks.
Acked-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Wrap the remaining skb_cow_data() into a helper, so it's easier
to replace down the lane. The new version will change the skb
so make sure relevant pointers get reloaded after the call.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
The non-zero-copy path assumes a full skb with decrypted contents.
This means the device offload would have to CoW the data. Try
to keep the zero-copy status instead, copy the data to user space.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
In the future we'll want to reuse the input skb in case of
zero-copy so we shouldn't always free darg.skb. Move the
freeing of darg.skb into the non-zc cases. All cases will
now free ctx->recv_pkt (inside let tls_rx_rec_done()).
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
After recent changes the SW side of tls_rx_one_record() can
be nicely encapsulated in its own function. Move the pad handling
as well. This will be useful for ->zc handling in tls_decrypt_device().
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
To allow for the logic to change later wrap accesses
which interrogate the input skb in helper functions.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Eric reports we should release the socket lock if the entire
"grab reader lock" operation has failed. The callers assume
they don't have to release it or otherwise unwind.
Reported-by: Eric Dumazet <edumazet@google.com>
Reported-by: syzbot+16e72110feb2b653ef27@syzkaller.appspotmail.com
Fixes: 4cbc325ed6b4 ("tls: rx: allow only one reader at a time")
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/20220720203701.2179034-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
No conflicts.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Socket destruction flow and tls_device_down function sync against each
other using tls_device_lock and the context refcount, to guarantee the
device resources are freed via tls_dev_del() by the end of
tls_device_down.
In the following unfortunate flow, this won't happen:
- refcount is decreased to zero in tls_device_sk_destruct.
- tls_device_down starts, skips the context as refcount is zero, going
all the way until it flushes the gc work, and returns without freeing
the device resources.
- only then, tls_device_queue_ctx_destruction is called, queues the gc
work and frees the context's device resources.
Solve it by decreasing the refcount in the socket's destruction flow
under the tls_device_lock, for perfect synchronization. This does not
slow down the common likely destructor flow, in which both the refcount
is decreased and the spinlock is acquired, anyway.
Fixes: e8f69799810c ("net/tls: Add generic NIC offload infrastructure")
Reviewed-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Reviewed-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
We currently CoW Rx skbs whenever we can't decrypt to a user
space buffer. The skbs can be enormous (64kB) and CoW does
a linear alloc which has a strong chance of failing under
memory pressure. Or even without, skb_cow_data() assumes
GFP_ATOMIC.
Allocate a new frag'd skb and decrypt into it. We finally
take advantage of the decrypted skb getting returned via
darg.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
The "zero-copy" path in SW TLS will engage either for no skbs or
for all but last. If the recvmsg parameters are right and the
socket can do ZC we'll ZC until the iterator can't fit a full
record at which point we'll decrypt one more record and copy
over the necessary bits to fill up the request.
The only reason we hold onto the ZC skbs which went thru the async
path until the end of recvmsg() is to count bytes. We need an accurate
count of zc'ed bytes so that we can calculate how much of the non-zc'd
data to copy. To allow freeing input skbs on the ZC path count only
how much of the list we'll need to consume.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
Async crypto currently benefits from the fact that we decrypt
in place. When we allow input and output to be different skbs
we will have to hang onto the input while we move to the next
record. Clone the inputs and keep them on a list.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
Async crypto TLS Rx currently waits for crypto to be done
in order to strip the TLS header and tailer. Simplify
the code by moving the pointers immediately, since only
TLS 1.2 is supported here there is no message padding.
This simplifies the decryption into a new skb in the next
patch as we don't have to worry about input vs output
skb in the decrypt_done() handler any more.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
Instead of using ctx->recv_pkt after decryption read the skb
from darg.skb. This moves the decision of what the "output skb"
is to the decrypt handlers. For now after decrypt handler returns
successfully ctx->recv_pkt is simply moved to darg.skb, but it
will change soon.
Note that tls_decrypt_sg() cannot clear the ctx->recv_pkt
because it gets called to re-encrypt (i.e. by the device offload).
So we need an awkward temporary if() in tls_rx_one_record().
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
Callers always pass ctx->recv_pkt into decrypt_skb_update(),
and it propagates it to its callees. This may give someone
the false impression that those functions can accept any valid
skb containing a TLS record. That's not the case, the record
sequence number is read from the context, and they can only
take the next record coming out of the strp.
Let the functions get the skb from the context instead of
passing it in. This will also make it cleaner to return
a different skb than ctx->recv_pkt as the decrypted one
later on.
Since we're touching the definition of decrypt_skb_update()
use this as an opportunity to rename it.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
I already forgot to transform darg from input to output
semantics once on the NIC inline crypto fastpath. To
avoid this happening again create a device equivalent
of decrypt_internal(). A function responsible for decryption
and transforming darg.
While at it rename decrypt_internal() to a hopefully slightly
more meaningful name.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
We no longer allow a decrypted skb to remain linked to ctx->recv_pkt.
Anything on the list is decrypted, anything on ctx->recv_pkt needs
to be decrypted.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
Detach the skb from ctx->recv_pkt after decryption is done,
even if we can't consume it.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
I thought that having the skb either always on the ctx->rx_list
or ctx->recv_pkt will simplify the handling, as we would not
have to remember to flip it from one to the other on exit paths.
This became a little harder to justify with the fix for BPF
sockmaps. Subsequent changes will make the situation even worse.
Queue the skbs only when really needed.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
recvmsg() in TLS gets data from the skb list (rx_list) or fresh
skbs we read from TCP via strparser. The former holds skbs which were
already decrypted for peek or decrypted and partially consumed.
tls_wait_data() only notices appearance of fresh skbs coming out
of TCP (or psock). It is possible, if there is a concurrent call
to peek() and recv() that the peek() will move the data from input
to rx_list without recv() noticing. recv() will then read data out
of order or never wake up.
This is not a practical use case/concern, but it makes the self
tests less reliable. This patch solves the problem by allowing
only one reader in.
Because having multiple processes calling read()/peek() is not
normal avoid adding a lock and try to fast-path the single reader
case.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
include/net/sock.h
310731e2f161 ("net: Fix data-races around sysctl_mem.")
e70f3c701276 ("Revert "net: set SK_MEM_QUANTUM to 4096"")
https://lore.kernel.org/all/20220711120211.7c8b7cba@canb.auug.org.au/
net/ipv4/fib_semantics.c
747c14307214 ("ip: fix dflt addr selection for connected nexthop")
d62607c3fe45 ("net: rename reference+tracking helpers")
net/tls/tls.h
include/net/tls.h
3d8c51b25a23 ("net/tls: Check for errors in tls_device_init")
587903142308 ("tls: create an internal header")
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Add missing error checks in tls_device_init.
Fixes: e8f69799810c ("net/tls: Add generic NIC offload infrastructure")
Reported-by: Jakub Kicinski <kuba@kernel.org>
Reviewed-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Link: https://lore.kernel.org/r/20220714070754.1428-1-tariqt@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Maxim reports do_tls_getsockopt_no_pad() will
always return an error. Indeed looks like refactoring
gone wrong - remove err and use value.
Reported-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Fixes: 88527790c079 ("tls: rx: add sockopt for enabling optimistic decrypt with TLS 1.3")
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
As discussed with Maxim add a counter for true NoPad violations.
This should help deployments catch unexpected padded records vs
just control records which always need re-encryption.
https: //lore.kernel.org/all/b111828e6ac34baad9f4e783127eba8344ac252d.camel@nvidia.com/
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
MIN -> MIB
Fixes: 88527790c079 ("tls: rx: add sockopt for enabling optimistic decrypt with TLS 1.3")
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
tls_wait_data() sets the return code as an output parameter
and always returns ctx->recv_pkt on success.
Return the error code directly and let the caller read the skb
from the context. Use positive return code to indicate ctx->recv_pkt
is ready.
While touching the definition of the function rename it.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
include/net/tls.h is getting a little long, and is probably hard
for driver authors to navigate. Split out the internals into a
header which will live under net/tls/. While at it move some
static inlines with a single user into the source files, add
a few tls_ prefixes and fix spelling of 'proccess'.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Jump to the free() call, instead of having to remember
to free the memory in multiple places.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
The max size of iv + aad + tail is 22B. That's smaller
than a single sg entry (32B). Don't bother with the
memory packing, just create a struct which holds the
max size of those members.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
AAD size is either 5 or 13. Really no point complicating
the code for the 8B of difference. This will also let us
turn the chunked up buffer into a sane struct.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
No conflicts.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
This reverts commit 284b4d93daee56dff3e10029ddf2e03227f50dbf.
When using TLS device offload and coming from tls_device_reencrypt()
flow, -EBADMSG error in tls_do_decryption() should not be counted
towards the TLSTlsDecryptError counter.
Move the counter increase back to the decrypt_internal() call site in
decrypt_skb_update().
This also fixes an issue where:
if (n_sgin < 1)
return -EBADMSG;
Errors in decrypt_internal() were not counted after the cited patch.
Fixes: 284b4d93daee ("tls: rx: move counting TlsDecryptErrors for sync")
Cc: Jakub Kicinski <kuba@kernel.org>
Reviewed-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Gal Pressman <gal@nvidia.com>
Reviewed-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
We continuously hold the socket lock during large reads and writes.
This may inflate RTT and negatively impact TCP performance.
Flush the backlog periodically. I tried to pick a flush period (128kB)
which gives significant benefit but the max Bps rate is not yet visibly
impacted.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
Since optimisitic decrypt may add extra load in case of retries
require socket owner to explicitly opt-in.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
We currently don't support decrypt to user buffer with TLS 1.3
because we don't know the record type and how much padding
record contains before decryption. In practice data records
are by far most common and padding gets used rarely so
we can assume data record, no padding, and if we find out
that wasn't the case - retry the crypto in place (decrypt
to skb).
To safeguard from user overwriting content type and padding
before we can check it attach a 1B sg entry where last byte
of the record will land.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
To make future patches easier to review make data_len
contain the length of the data, without the tail.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
Most drivers use "skb_transport_offset(skb) + tcp_hdrlen(skb)"
to compute headers length for a TCP packet, but others
use more convoluted (but equivalent) ways.
Add skb_tcp_all_headers() and skb_inner_tcp_all_headers()
helpers to harmonize this a bit.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
Commit 8a59f9d1e3d4 ("sock: Introduce sk->sk_prot->psock_update_sk_prot()")
has moved the inet_csk_has_ulp(sk) check from sk_psock_init() to
the new tcp_bpf_update_proto() function. I'm guessing that this
was done to allow creating psocks for non-inet sockets.
Unfortunately the destruction path for psock includes the ULP
unwind, so we need to fail the sk_psock_init() itself.
Otherwise if ULP is already present we'll notice that later,
and call tcp_update_ulp() with the sk_proto of the ULP
itself, which will most likely result in the ULP looping
its callbacks.
Fixes: 8a59f9d1e3d4 ("sock: Introduce sk->sk_prot->psock_update_sk_prot()")
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Reviewed-by: John Fastabend <john.fastabend@gmail.com>
Reviewed-by: Jakub Sitnicki <jakub@cloudflare.com>
Tested-by: Jakub Sitnicki <jakub@cloudflare.com>
Link: https://lore.kernel.org/r/20220620191353.1184629-2-kuba@kernel.org
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
This reverts commit 69135c572d1f84261a6de2a1268513a7e71753e2.
This commit was just papering over the issue, ULP should not
get ->update() called with its own sk_prot. Each ULP would
need to add this check.
Fixes: 69135c572d1f ("net/tls: fix tls_sk_proto_close executed repeatedly")
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Reviewed-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/r/20220620191353.1184629-1-kuba@kernel.org
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
After setting the sock ktls, update ctx->sk_proto to sock->sk_prot by
tls_update(), so now ctx->sk_proto->close is tls_sk_proto_close(). When
close the sock, tls_sk_proto_close() is called for sock->sk_prot->close
is tls_sk_proto_close(). But ctx->sk_proto->close() will be executed later
in tls_sk_proto_close(). Thus tls_sk_proto_close() executed repeatedly
occurred. That will trigger the following bug.
=================================================================
KASAN: null-ptr-deref in range [0x0000000000000010-0x0000000000000017]
RIP: 0010:tls_sk_proto_close+0xd8/0xaf0 net/tls/tls_main.c:306
Call Trace:
<TASK>
tls_sk_proto_close+0x356/0xaf0 net/tls/tls_main.c:329
inet_release+0x12e/0x280 net/ipv4/af_inet.c:428
__sock_release+0xcd/0x280 net/socket.c:650
sock_close+0x18/0x20 net/socket.c:1365
Updating a proto which is same with sock->sk_prot is incorrect. Add proto
and sock->sk_prot equality check at the head of tls_update() to fix it.
Fixes: 95fa145479fb ("bpf: sockmap/tls, close can race with map free")
Reported-by: syzbot+29c3c12f3214b85ad081@syzkaller.appspotmail.com
Signed-off-by: Ziyang Xuan <william.xuanziyang@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
To embrace possible future optimizations of TLS, rename zerocopy
sendfile definitions to more generic ones:
* setsockopt: TLS_TX_ZEROCOPY_SENDFILE- > TLS_TX_ZEROCOPY_RO
* sock_diag: TLS_INFO_ZC_SENDFILE -> TLS_INFO_ZC_RO_TX
RO stands for readonly and emphasizes that the application shouldn't
modify the data being transmitted with zerocopy to avoid potential
disconnection.
Fixes: c1318b39c7d3 ("tls: Add opt-in zerocopy mode of sendfile()")
Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Link: https://lore.kernel.org/r/20220608153425.3151146-1-maximmi@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Artem points out that skb may try to take over the skb and
queue it to its own list. Unlink the skb before calling out.
Fixes: b1a2c1786330 ("tls: rx: clear ctx->recv_pkt earlier")
Reported-by: Artem Savkov <asavkov@redhat.com>
Tested-by: Artem Savkov <asavkov@redhat.com>
Link: https://lore.kernel.org/r/20220518205644.2059468-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
TLS device offload copies sendfile data to a bounce buffer before
transmitting. It allows to maintain the valid MAC on TLS records when
the file contents change and a part of TLS record has to be
retransmitted on TCP level.
In many common use cases (like serving static files over HTTPS) the file
contents are not changed on the fly. In many use cases breaking the
connection is totally acceptable if the file is changed during
transmission, because it would be received corrupted in any case.
This commit allows to optimize performance for such use cases to
providing a new optional mode of TLS sendfile(), in which the extra copy
is skipped. Removing this copy improves performance significantly, as
TLS and TCP sendfile perform the same operations, and the only overhead
is TLS header/trailer insertion.
The new mode can only be enabled with the new socket option named
TLS_TX_ZEROCOPY_SENDFILE on per-socket basis. It preserves backwards
compatibility with existing applications that rely on the copying
behavior.
The new mode is safe, meaning that unsolicited modifications of the file
being sent can't break integrity of the kernel. The worst thing that can
happen is sending a corrupted TLS record, which is in any case not
forbidden when using regular TCP sockets.
Sockets other than TLS device offload are not affected by the new socket
option. The actual status of zerocopy sendfile can be queried with
sock_diag.
Performance numbers in a single-core test with 24 HTTPS streams on
nginx, under 100% CPU load:
* non-zerocopy: 33.6 Gbit/s
* zerocopy: 79.92 Gbit/s
CPU: Intel(R) Xeon(R) Platinum 8380 CPU @ 2.30GHz
Signed-off-by: Boris Pismenny <borisp@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: Jakub Kicinski <kuba@kernel.org>
Link: https://lore.kernel.org/r/20220518092731.1243494-1-maximmi@nvidia.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
No conflicts.
Build issue in drivers/net/ethernet/sfc/ptp.c
54fccfdd7c66 ("sfc: efx_default_channel_type APIs can be static")
49e6123c65da ("net: sfc: fix memory leak due to ptp channel")
https://lore.kernel.org/all/20220510130556.52598fe2@canb.auug.org.au/
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
The commit cited below claims to fix a use-after-free condition after
tls_device_down. Apparently, the description wasn't fully accurate. The
context stayed alive, but ctx->netdev became NULL, and the offload was
torn down without a proper fallback, so a bug was present, but a
different kind of bug.
Due to misunderstanding of the issue, the original patch dropped the
refcount_dec_and_test line for the context to avoid the alleged
premature deallocation. That line has to be restored, because it matches
the refcount_inc_not_zero from the same function, otherwise the contexts
that survived tls_device_down are leaked.
This patch fixes the described issue by restoring refcount_dec_and_test.
After this change, there is no leak anymore, and the fallback to
software kTLS still works.
Fixes: c55dcdd435aa ("net/tls: Fix use-after-free after the TLS device goes down and up")
Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Link: https://lore.kernel.org/r/20220512091830.678684-1-maximmi@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
include/linux/netdevice.h
net/core/dev.c
6510ea973d8d ("net: Use this_cpu_inc() to increment net->core_stats")
794c24e9921f ("net-core: rx_otherhost_dropped to core_stats")
https://lore.kernel.org/all/20220428111903.5f4304e0@canb.auug.org.au/
drivers/net/wan/cosa.c
d48fea8401cf ("net: cosa: fix error check return value of register_chrdev()")
89fbca3307d4 ("net: wan: remove support for COSA and SRP synchronous serial boards")
https://lore.kernel.org/all/20220428112130.1f689e5e@canb.auug.org.au/
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Calling tls_append_frag when max_open_record_len == record->len might
add an empty fragment to the TLS record if the call happens to be on the
page boundary. Normally tls_append_frag coalesces the zero-sized
fragment to the previous one, but not if it's on page boundary.
If a resync happens then, the mlx5 driver posts dump WQEs in
tx_post_resync_dump, and the empty fragment may become a data segment
with byte_count == 0, which will confuse the NIC and lead to a CQE
error.
This commit fixes the described issue by skipping tls_append_frag on
zero size to avoid adding empty fragments. The fix is not in the driver,
because an empty fragment is hardly the desired behavior.
Fixes: e8f69799810c ("net/tls: Add generic NIC offload infrastructure")
Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Link: https://lore.kernel.org/r/20220426154949.159055-1-maximmi@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
When NIC takes care of crypto (or the record has already
been decrypted) we forget to update darg->async. ->async
is supposed to mean whether record is async capable on
input and whether record has been queued for async crypto
on output.
Reported-by: Gal Pressman <gal@nvidia.com>
Fixes: 3547a1f9d988 ("tls: rx: use async as an in-out argument")
Tested-by: Gal Pressman <gal@nvidia.com>
Link: https://lore.kernel.org/r/20220425233309.344858-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Logic added in commit f35f821935d8 ("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>
|
|
TLS 1.3 and ChaChaPoly don't carry IV in the packet.
The code before this change would copy out iv_size
worth of whatever followed the TLS header in the packet
and then for TLS 1.3 | ChaCha overwrite that with
the sequence number. Waste of cycles especially
with TLS 1.2 being close to dead and TLS 1.3 being
the common case.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
