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: NJakub 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: e8f69799 ("net/tls: Add generic NIC offload infrastructure")
Reviewed-by: NMaxim Mikityanskiy <maximmi@nvidia.com>
Signed-off-by: NTariq Toukan <tariqt@nvidia.com>
Reviewed-by: NJakub Kicinski <kuba@kernel.org>
Signed-off-by: NDavid 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: NJakub Kicinski <kuba@kernel.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add missing error checks in tls_device_init.

Fixes: e8f69799 ("net/tls: Add generic NIC offload infrastructure")
Reported-by: NJakub Kicinski <kuba@kernel.org>
Reviewed-by: NMaxim Mikityanskiy <maximmi@nvidia.com>
Signed-off-by: NTariq Toukan <tariqt@nvidia.com>
Link: https://lore.kernel.org/r/20220714070754.1428-1-tariqt@nvidia.comSigned-off-by: NJakub 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: NJakub 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: NBoris Pismenny <borisp@nvidia.com>
Signed-off-by: NTariq Toukan <tariqt@nvidia.com>
Signed-off-by: NMaxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: NJakub Kicinski <kuba@kernel.org>
Link: https://lore.kernel.org/r/20220518092731.1243494-1-maximmi@nvidia.comSigned-off-by: NPaolo Abeni <pabeni@redhat.com>

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: c55dcdd4 ("net/tls: Fix use-after-free after the TLS device goes down and up")
Signed-off-by: NMaxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: NTariq Toukan <tariqt@nvidia.com>
Link: https://lore.kernel.org/r/20220512091830.678684-1-maximmi@nvidia.comSigned-off-by: NJakub 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: e8f69799 ("net/tls: Add generic NIC offload infrastructure")
Signed-off-by: NMaxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: NTariq Toukan <tariqt@nvidia.com>
Link: https://lore.kernel.org/r/20220426154949.159055-1-maximmi@nvidia.comSigned-off-by: NJakub Kicinski <kuba@kernel.org>

Instead of tls_device poking into internals of the message
return 1 from tls_device_decrypted() if the device handled
the decryption.
Signed-off-by: NJakub Kicinski <kuba@kernel.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Similar justification to previous change, the information
about decryption status belongs in the skb.
Signed-off-by: NJakub Kicinski <kuba@kernel.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

It is known that priority setting HW offload when set tls TX/RX offload
by setsockopt(). Check netdevice whether support NETIF_F_HW_TLS_TX or
not at the later stages in the whole tls_set_device_offload() process,
some memory allocations have been done before that. We must release those
memory and return error if we judge the netdevice not support
NETIF_F_HW_TLS_TX. It is redundant.

Move NETIF_F_HW_TLS_TX judgement forward, and move start_marker_record
and offload_ctx memory allocation back slightly. Thus, we can get
simpler exception handling process.
Signed-off-by: NZiyang Xuan <william.xuanziyang@huawei.com>
Signed-off-by: NJakub Kicinski <kuba@kernel.org>

This is a prerequisite patch, the next one is enabling recycling of
skbs and fragments. Add an extra argument on __skb_frag_unref() to
handle recycling, and update the current users of the function with that.
Signed-off-by: NMatteo Croce <mcroce@microsoft.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When a netdev with active TLS offload goes down, tls_device_down is
called to stop the offload and tear down the TLS context. However, the
socket stays alive, and it still points to the TLS context, which is now
deallocated. If a netdev goes up, while the connection is still active,
and the data flow resumes after a number of TCP retransmissions, it will
lead to a use-after-free of the TLS context.

This commit addresses this bug by keeping the context alive until its
normal destruction, and implements the necessary fallbacks, so that the
connection can resume in software (non-offloaded) kTLS mode.

On the TX side tls_sw_fallback is used to encrypt all packets. The RX
side already has all the necessary fallbacks, because receiving
non-decrypted packets is supported. The thing needed on the RX side is
to block resync requests, which are normally produced after receiving
non-decrypted packets.

The necessary synchronization is implemented for a graceful teardown:
first the fallbacks are deployed, then the driver resources are released
(it used to be possible to have a tls_dev_resync after tls_dev_del).

A new flag called TLS_RX_DEV_DEGRADED is added to indicate the fallback
mode. It's used to skip the RX resync logic completely, as it becomes
useless, and some objects may be released (for example, resync_async,
which is allocated and freed by the driver).

Fixes: e8f69799 ("net/tls: Add generic NIC offload infrastructure")
Signed-off-by: NMaxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: NTariq Toukan <tariqt@nvidia.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

RCU synchronization is guaranteed to finish in finite time, unlike a
busy loop that polls a flag. This patch is a preparation for the bugfix
in the next patch, where the same synchronize_net() call will also be
used to sync with the TX datapath.
Signed-off-by: NMaxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: NTariq Toukan <tariqt@nvidia.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

record is being initialized to ctx->open_record but this is never
read as record is overwritten later on.  Remove the redundant
initialization.

Cleans up the following clang-analyzer warning:

net/tls/tls_device.c:421:26: warning: Value stored to 'record' during
its initialization is never read [clang-analyzer-deadcode.DeadStores].
Reported-by: NAbaci Robot <abaci@linux.alibaba.com>
Signed-off-by: NJiapeng Chong <jiapeng.chong@linux.alibaba.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

s/beggining/beginning/
Reported-by: NHulk Robot <hulkci@huawei.com>
Signed-off-by: NWang Hai <wanghai38@huawei.com>
Acked-by: NRandy Dunlap <rdunlap@infradead.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In the tls_dev_event handler, ignore tlsdev_ops requirement for bond
interfaces, they do not exist as the interaction is done directly with
the lower device.

Also, make the validate function pass when it's called with the upper
bond interface.
Signed-off-by: NTariq Toukan <tariqt@nvidia.com>
Reviewed-by: NBoris Pismenny <borisp@nvidia.com>
Signed-off-by: NJakub Kicinski <kuba@kernel.org>

Do not call the tls_dev_ops of upper devices. Instead, ask them
for the proper lowest device and communicate with it directly.
Signed-off-by: NTariq Toukan <tariqt@nvidia.com>
Reviewed-by: NBoris Pismenny <borisp@nvidia.com>
Signed-off-by: NJakub Kicinski <kuba@kernel.org>

Recent changes made to remove AES constants started using protocol
aware salt_size. ctx->prot_info's salt_size is filled in tls sw case,
but not in tls offload mode, and was working so far because of the
hard coded value was used.

Fixes: 6942a284 ("net/tls: make inline helpers protocol-aware")
Signed-off-by: NRohit Maheshwari <rohitm@chelsio.com>
Link: https://lore.kernel.org/r/20201201090752.27355-1-rohitm@chelsio.comSigned-off-by: NJakub Kicinski <kuba@kernel.org>

Inline functions defined in tls.h have a lot of AES-specific
constants. Remove these constants and change argument to struct
tls_prot_info to have an access to cipher type in later patches
Signed-off-by: NVadim Fedorenko <vfedorenko@novek.ru>
Signed-off-by: NJakub Kicinski <kuba@kernel.org>

tls_device_offload_cleanup_rx doesn't clear tls_ctx->netdev after
calling tls_dev_del if TLX TX offload is also enabled. Clearing
tls_ctx->netdev gets postponed until tls_device_gc_task. It leaves a
time frame when tls_device_down may get called and call tls_dev_del for
RX one extra time, confusing the driver, which may lead to a crash.

This patch corrects this racy behavior by adding a flag to prevent
tls_device_down from calling tls_dev_del the second time.

Fixes: e8f69799 ("net/tls: Add generic NIC offload infrastructure")
Signed-off-by: NMaxim Mikityanskiy <maximmi@mellanox.com>
Signed-off-by: NSaeed Mahameed <saeedm@nvidia.com>
Link: https://lore.kernel.org/r/20201125221810.69870-1-saeedm@nvidia.comSigned-off-by: NJakub Kicinski <kuba@kernel.org>

In async_resync mode, we log the TCP seq of records until the async request
is completed.  Later, in case one of the logged seqs matches the resync
request, we return it, together with its record serial number.  Before this
fix, we mistakenly returned the serial number of the current record
instead.

Fixes: ed9b7646 ("net/tls: Add asynchronous resync")
Signed-off-by: NTariq Toukan <tariqt@nvidia.com>
Reviewed-by: NBoris Pismenny <borisp@nvidia.com>
Link: https://lore.kernel.org/r/20201115131448.2702-1-tariqt@nvidia.comSigned-off-by: NJakub Kicinski <kuba@kernel.org>

At first when sendpage gets called, if there is more data, 'more' in
tls_push_data() gets set which later sets pending_open_record_frags, but
when there is no more data in file left, and last time tls_push_data()
gets called, pending_open_record_frags doesn't get reset. And later when
2 bytes of encrypted alert comes as sendmsg, it first checks for
pending_open_record_frags, and since this is set, it creates a record with
0 data bytes to encrypt, meaning record length is prepend_size + tag_size
only, which causes problem.
 We should set/reset pending_open_record_frags based on more bit.

Fixes: e8f69799 ("net/tls: Add generic NIC offload infrastructure")
Signed-off-by: NRohit Maheshwari <rohitm@chelsio.com>
Signed-off-by: NJakub Kicinski <kuba@kernel.org>

When MSG_OOB is specified to tls_device_sendpage() the mapped page is
never unmapped.

Hold off mapping the page until after the flags are checked and the page
is actually needed.

Fixes: e8f69799 ("net/tls: Add generic NIC offload infrastructure")
Signed-off-by: NIra Weiny <ira.weiny@intel.com>
Reviewed-by: NJakub Kicinski <kuba@kernel.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch adds support for asynchronous resynchronization in tls_device.
Async resync follows two distinct stages:

1. The NIC driver indicates that it would like to resync on some TLS
record within the received packet (P), but the driver does not
know (yet) which of the TLS records within the packet.
At this stage, the NIC driver will query the device to find the exact
TCP sequence for resync (tcpsn), however, the driver does not wait
for the device to provide the response.

2. Eventually, the device responds, and the driver provides the tcpsn
within the resync packet to KTLS. Now, KTLS can check the tcpsn against
any processed TLS records within packet P, and also against any record
that is processed in the future within packet P.

The asynchronous resync path simplifies the device driver, as it can
save bits on the packet completion (32-bit TCP sequence), and pass this
information on an asynchronous command instead.
Signed-off-by: NBoris Pismenny <borisp@mellanox.com>
Signed-off-by: NTariq Toukan <tariqt@mellanox.com>
Signed-off-by: NSaeed Mahameed <saeedm@mellanox.com>

This reverts commit b3ae2459.
Revert the force resync API.
Not in use. To be replaced by a better async resync API downstream.
Signed-off-by: NBoris Pismenny <borisp@mellanox.com>
Signed-off-by: NTariq Toukan <tariqt@mellanox.com>
Reviewed-by: NMaxim Mikityanskiy <maximmi@mellanox.com>
Signed-off-by: NSaeed Mahameed <saeedm@mellanox.com>

This patch adds a field to the tls rx offload context which enables
drivers to force a send_resync call.

This field can be used by drivers to request a resync at the next
possible tls record. It is beneficial for hardware that provides the
resync sequence number asynchronously. In such cases, the packet that
triggered the resync does not contain the information required for a
resync. Instead, the driver requests resync for all the following
TLS record until the asynchronous notification with the resync request
TCP sequence arrives.

A following series for mlx5e ConnectX-6DX TLS RX offload support will
use this mechanism.
Signed-off-by: NBoris Pismenny <borisp@mellanox.com>
Signed-off-by: NTariq Toukan <tariqt@mellanox.com>
Reviewed-by: NMaxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: NSaeed Mahameed <saeedm@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

sockmap performs lockless writes to sk->sk_prot on the following paths:

tcp_bpf_{recvmsg|sendmsg} / sock_map_unref
  sk_psock_put
    sk_psock_drop
      sk_psock_restore_proto
        WRITE_ONCE(sk->sk_prot, proto)

To prevent load/store tearing [1], and to make tooling aware of intentional
shared access [2], we need to annotate other sites that access sk_prot with
READ_ONCE/WRITE_ONCE macros.

Change done with Coccinelle with following semantic patch:

@@
expression E;
identifier I;
struct sock *sk;
identifier sk_prot =~ "^sk_prot$";
@@
(
 E =
-sk->sk_prot
+READ_ONCE(sk->sk_prot)
|
-sk->sk_prot = E
+WRITE_ONCE(sk->sk_prot, E)
|
-sk->sk_prot
+READ_ONCE(sk->sk_prot)
 ->I
)
Signed-off-by: NJakub Sitnicki <jakub@cloudflare.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Current code doesn't check if tcp sequence number is starting from (/after)
1st record's start sequnce number. It only checks if seq number is before
1st record's end sequnce number. This problem will always be a possibility
in re-transmit case. If a record which belongs to a requested seq number is
already deleted, tls_get_record will start looking into list and as per the
check it will look if seq number is before the end seq of 1st record, which
will always be true and will return 1st record always, it should in fact
return NULL.
As part of the fix, start looking each record only if the sequence number
lies in the list else return NULL.
There is one more check added, driver look for the start marker record to
handle tcp packets which are before the tls offload start sequence number,
hence return 1st record if the record is tls start marker and seq number is
before the 1st record's starting sequence number.

Fixes: e8f69799 ("net/tls: Add generic NIC offload infrastructure")
Signed-off-by: NRohit Maheshwari <rohitm@chelsio.com>
Reviewed-by: NJakub Kicinski <kuba@kernel.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

There is currently no way for driver to reliably check that
the socket it has looked up is in fact RX offloaded. Add
a helper. This allows drivers to catch misbehaving firmware.
Signed-off-by: NJakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

ENOTSUPP is not available in userspace, for example:

  setsockopt failed, 524, Unknown error 524
Signed-off-by: NValentin Vidic <vvidic@valentin-vidic.from.hr>
Acked-by: NJakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

TLS TX needs to release and re-acquire the socket lock if send buffer
fills up.

TLS SW TX path currently depends on only allowing one thread to enter
the function by the abuse of sk_write_pending. If another writer is
already waiting for memory no new ones are allowed in.

This has two problems:
 - writers don't wake other threads up when they leave the kernel;
   meaning that this scheme works for single extra thread (second
   application thread or delayed work) because memory becoming
   available will send a wake up request, but as Mallesham and
   Pooja report with larger number of threads it leads to threads
   being put to sleep indefinitely;
 - the delayed work does not get _scheduled_ but it may _run_ when
   other writers are present leading to crashes as writers don't
   expect state to change under their feet (same records get pushed
   and freed multiple times); it's hard to reliably bail from the
   work, however, because the mere presence of a writer does not
   guarantee that the writer will push pending records before exiting.

Ensuring wakeups always happen will make the code basically open
code a mutex. Just use a mutex.

The TLS HW TX path does not have any locking (not even the
sk_write_pending hack), yet it uses a per-socket sg_tx_data
array to push records.

Fixes: a42055e8 ("net/tls: Add support for async encryption of records for performance")
Reported-by: NMallesham  Jatharakonda <mallesh537@gmail.com>
Reported-by: NPooja Trivedi <poojatrivedi@gmail.com>
Signed-off-by: NJakub Kicinski <jakub.kicinski@netronome.com>
Reviewed-by: NSimon Horman <simon.horman@netronome.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

sk_write_pending being not zero does not guarantee that partial
record will be pushed. If the thread waiting for memory times out
the pending record may get stuck.

In case of tls_device there is no path where parial record is
set and writer present in the first place. Partial record is
set only in tls_push_sg() and tls_push_sg() will return an
error immediately. All tls_device callers of tls_push_sg()
will return (and not wait for memory) if it failed.

Fixes: a42055e8 ("net/tls: Add support for async encryption of records for performance")
Signed-off-by: NJakub Kicinski <jakub.kicinski@netronome.com>
Reviewed-by: NSimon Horman <simon.horman@netronome.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Avoid unnecessary pointer chasing and calculations, callers already
have most of the state tls_device_decrypted() needs.
Signed-off-by: NJakub Kicinski <jakub.kicinski@netronome.com>
Reviewed-by: NDirk van der Merwe <dirk.vandermerwe@netronome.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Make sure GCC realizes it's unlikely that allocations will fail.
Signed-off-by: NJakub Kicinski <jakub.kicinski@netronome.com>
Reviewed-by: NDirk van der Merwe <dirk.vandermerwe@netronome.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Tell GCC sk->err is not likely to be set.
Signed-off-by: NJakub Kicinski <jakub.kicinski@netronome.com>
Reviewed-by: NDirk van der Merwe <dirk.vandermerwe@netronome.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add a statistic for number of RX resyncs sent down to the NIC.
Signed-off-by: NJakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add a tracepoint to the TLS offload's fast path. This tracepoint
can be used to track the decrypted and encrypted status of received
records. Records decrypted by the device should have decrypted set
to 1, records which have neither decrypted nor decrypted set are
partially decrypted, require re-encryption and therefore are most
expensive to deal with.
Signed-off-by: NJakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add tracing of device-related interaction to aid performance
analysis, especially around resync:

 tls:tls_device_offload_set
 tls:tls_device_rx_resync_send
 tls:tls_device_rx_resync_nh_schedule
 tls:tls_device_rx_resync_nh_delay
 tls:tls_device_tx_resync_req
 tls:tls_device_tx_resync_send
Signed-off-by: NJakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Unlike normal TCP code TLS has to touch the cache lines
it copies into to fill header info. On memory-heavy workloads
having non temporal stores and normal accesses targeting
the same cache line leads to significant overhead.

Measured 3% overhead running 3600 round robin connections
with additional memory heavy workload.
Signed-off-by: NJakub Kicinski <jakub.kicinski@netronome.com>
Reviewed-by: NDirk van der Merwe <dirk.vandermerwe@netronome.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>