This patch implements GSI transactions.  A GSI transaction is a
structure that represents a single request (consisting of one or
more TREs) sent to the GSI hardware.  The last TRE in a transaction
includes a flag requesting that the GSI interrupt the AP to notify
that it has completed.

TREs are executed and completed strictly in order.  For this reason,
the completion of a single TRE implies that all previous TREs (in
particular all of those "earlier" in a transaction) have completed.

Whenever there is a need to send a request (a set of TREs) to the
IPA, a GSI transaction is allocated, specifying the number of TREs
that will be required.  Details of the request (e.g. transfer offsets
and length) are represented by in a Linux scatterlist array that is
incorporated in the transaction structure.

Once all commands (TREs) are added to a transaction it is committed.
When the hardware signals that the request has completed, a callback
function allows for cleanup or followup activity to be performed
before the transaction is freed.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch provides interface functions supplied by the IPA layer
that are called from the GSI layer.  One function is called when a
GSI transaction has completed.  The others allow the GSI layer to
inform the IPA layer when the hardware has been told it has new TREs
to execute, and when the hardware has indicated transactions have
completed.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch includes "gsi.c", which implements the generic software
interface (GSI) for IPA.  The generic software interface abstracts
channels, which provide a means of transferring data either from the
AP to the IPA, or from the IPA to the AP.  A ring buffer of "transfer
elements" (TREs) is used to describe data transfers to perform.  The
AP writes a doorbell register associated with a channel to let it know
it has added new entries (for an AP->IPA channel) or has finished
processing entries (for an IPA->AP channel).

Each channel also has an event ring buffer, used by the IPA to
communicate information about events related to a channel (for
example, the completion of TREs).  The IPA writes its own doorbell
register, which triggers an interrupt on the AP, to signal that
new event information has arrived.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

The Generic Software Interface is a layer of the IPA driver that
abstracts the underlying hardware.  The next patch includes the
main code for GSI (including some additional documentation).  This
patch just includes three GSI header files.

  - "gsi.h" is the top-level GSI header file.  This structure is
    is embedded within the IPA structure.  The main abstraction
    implemented by the GSI code is the channel, and this header
    exposes several operations that can be performed on a GSI channel.

  - "gsi_private.h" exposes some definitions that are intended to be
    private, used only by the main GSI code and the GSI transaction
    code (defined in an upcoming patch).

  - Like "ipa_reg.h", "gsi_reg.h" defines the offsets of the 32-bit
    registers used by the GSI layer, along with masks that define the
    position and width of fields less than 32 bits located within
    these registers.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch incorporates three source files (and their headers).  They're
grouped into one patch mainly for the purpose of making the number and
size of patches in this series somewhat reasonable.

  - "ipa_clock.c" and "ipa_clock.h" implement clocking for the IPA device.
    The IPA has a single core clock managed by the common clock framework.
    In addition, the IPA has three buses whose bandwidth is managed by the
    Linux interconnect framework.  At this time the core clock and all
    three buses are either on or off; we don't yet do any more fine-grained
    management than that.  The core clock and interconnects are enabled
    and disabled as a unit, using a unified clock-like abstraction,
    ipa_clock_get()/ipa_clock_put().

  - "ipa_interrupt.c" and "ipa_interrupt.h" implement IPA interrupts.
    There are two hardware IRQs used by the IPA driver (the other is
    the GSI interrupt, described in a separate patch).  Several types
    of interrupt are handled by the IPA IRQ handler; these are not part
    of data/fast path.

  - The IPA has a region of local memory that is accessible by the AP
    (and modem).  Within that region are areas with certain defined
    purposes.  "ipa_mem.c" and "ipa_mem.h" define those regions, and
    implement their initialization.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch defines configuration data that is used to specify some
of the details of IPA hardware supported by the driver.  It is built
as Device Tree match data, discovered at boot time.  The driver
supports the Qualcomm SDM845 SoC.  Data for the Qualcomm SC7180 is
also defined here, but it is not yet completely supported.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch includes three source files that represent some basic "main
program" code for the IPA driver.  They are:
  - "ipa.h" defines the top-level IPA structure which represents an IPA
     device throughout the code.
  - "ipa_main.c" contains the platform driver probe function, along with
    some general code used during initialization.
  - "ipa_reg.h" defines the offsets of the 32-bit registers used for the
    IPA device, along with masks that define the position and width of
    fields within these registers.
  - "version.h" defines some symbolic IPA version numbers.

Each file includes some documentation that provides a little more
overview of how the code is organized and used.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Add the binding definitions for the "qcom,ipa" device tree node.

Signed-off-by: Alex Elder <elder@linaro.org>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Set up a subdev in the q6v5 modem remoteproc driver that generates
event notifications for the IPA driver to use for initialization and
recovery following a modem shutdown or crash.

A pair of new functions provides a way for the IPA driver to register
and deregister a notification callback function that will be called
whenever modem events (about to boot, running, about to shut down,
etc.) occur.  A void pointer value (provided by the IPA driver at
registration time) and an event type are supplied to the callback
function.

One event, MODEM_REMOVING, is signaled whenever the q6v5 driver is
about to remove the notification subdevice.  It requires the IPA
driver de-register its callback.

This sub-device is only used by the modem subsystem (MSS) driver,
so the code that adds the new subdev and allows registration and
deregistration of the notifier is found in "qcom_q6v6_mss.c".

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Sascha Hauer says:

====================
QorIQ DPAA: Use random MAC address when none is given

Use random MAC addresses when they are not provided in the device tree.
Tested on LS1046ARDB.

Changes in v3:
 addressed all MAC types, removed some redundant code in dtsec in
 the process
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

If there is no valid MAC address in the device tree, use a random
MAC address.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Madalin Bucur <madalin.bucur@oss.nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Allow the initialization of the MAC to be performed even if the
device tree does not provide a valid MAC address. Later a random
MAC address should be assigned by the Ethernet driver.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Madalin Bucur <madalin.bucur@oss.nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Reuse the set_mac_address() in the init() function.

Signed-off-by: Madalin Bucur <madalin.bucur@oss.nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Michael Chan says:

====================
bnxt_en: Updates.

This series includes simplification and improvement of NAPI polling
logic in bnxt_poll_p5().  The improvements will prevent starving the
async events from firmware if we are in continuous NAPI polling.
The rest of the patches include cleanups, a better return code for
firmware busy, and to clear devlink port type more properly.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

Similar to other drivers, properly clear the devlink port type when
removing the device before unregistration.

Cc: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Vasundhara Volam <vasundhara-v.volam@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

If firmware command returns error code as HWRM_ERR_CODE_BUSY, which
means it cannot handle the command due to a conflicting command
from another function, convert it to -EAGAIN.  If it is an ethtool
operation, this error code will be returned to userspace.

Signed-off-by: Vasundhara Volam <vasundhara-v.volam@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Return code is not needed in some of these functions, as the return
code from firmware message is ignored. Remove the unused rc variable
and also convert functions to void.

Signed-off-by: Vasundhara Volam <vasundhara-v.volam@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

As part of converting error code in firmware message to standard
code, checking for firmware return code is removed in most of the
places. Remove the assignment of return code where the function
can directly return.

Signed-off-by: Vasundhara Volam <vasundhara-v.volam@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The driver stores a copy of the DCB settings that have been applied to
the firmware.  After firmware reset, the firmware settings are gone and
will revert back to default.  Clear the driver's copy so that if there
is a DCBNL request to get the settings, the driver will retrieve the
current settings from the firmware.  lldpad keeps the DCB settings in
userspace and will re-apply the settings if it is running.

Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

When we are in continuous NAPI polling mode, the current code in
bnxt_poll_p5() will only process the completion rings and will not
process the NQ until interrupt is re-enabled.  Tis logic works and
will not cause RX or TX starvation, but async events in the NQ may
be delayed for the duration of continuous NAPI polling.  These
async events may be firmware or VF events.

Continue to handle the NQ after we are done polling the completion
rings.  This actually simplies the code in bnxt_poll_p5().

Acknowledge the NQ so these async events will not overflow.

Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Simplify the function by removing tha 'all' parameter.  In the current
code, the caller has to specify whether to update/arm both completion
rings with the 'all' parameter.

Instead of this, we can just update/arm all the completion rings
that have been polled.  By setting cpr->had_work_done earlier in
__bnxt_poll_work(), we know which completion ring has been polled
and can just update/arm all the completion rings with
cpr->had_work_done set.

This simplifies the function with one less parameter and works just
as well.

Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

In bnxt_poll_p5(), the logic polls for up to 2 completion rings (RX and
TX) for work.  In the current code, if we reach budget polling the
first completion ring, we will stop.  If the other completion ring
has work to do, we will handle it when NAPI calls us back.

This is not optimal.  We potentially leave an unproceesed entry in
the NQ.  When we are finally done with NAPI polling and re-enable
interrupt, the remaining entry in the NQ will cause interrupt to
be triggered immediately for no reason.

Modify the code in bnxt_poll_p5() to keep looping until all NQ
entries are handled even if the first completion ring has reached
budget.

Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Convert zones_lock spinlock to zones_mutex mutex,
and struct (tcf_ct_flow_table)->ref to a refcount,
so that control path can use regular GFP_KERNEL allocations
from standard process context. This is more robust
in case of memory pressure.

The refcount is needed because tcf_ct_flow_table_put() can
be called from RCU callback, thus in BH context.

The issue was spotted by syzbot, as rhashtable_init()
was called with a spinlock held, which is bad since GFP_KERNEL
allocations can sleep.

Note to developers : Please make sure your patches are tested
with CONFIG_DEBUG_ATOMIC_SLEEP=y

BUG: sleeping function called from invalid context at mm/slab.h:565
in_atomic(): 1, irqs_disabled(): 0, non_block: 0, pid: 9582, name: syz-executor610
2 locks held by syz-executor610/9582:
 #0: ffffffff8a34eb80 (rtnl_mutex){+.+.}, at: rtnl_lock net/core/rtnetlink.c:72 [inline]
 #0: ffffffff8a34eb80 (rtnl_mutex){+.+.}, at: rtnetlink_rcv_msg+0x3f9/0xad0 net/core/rtnetlink.c:5437
 #1: ffffffff8a3961b8 (zones_lock){+...}, at: spin_lock_bh include/linux/spinlock.h:343 [inline]
 #1: ffffffff8a3961b8 (zones_lock){+...}, at: tcf_ct_flow_table_get+0xa3/0x1700 net/sched/act_ct.c:67
Preemption disabled at:
[<0000000000000000>] 0x0
CPU: 0 PID: 9582 Comm: syz-executor610 Not tainted 5.6.0-rc3-syzkaller #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Call Trace:
 __dump_stack lib/dump_stack.c:77 [inline]
 dump_stack+0x188/0x20d lib/dump_stack.c:118
 ___might_sleep.cold+0x1f4/0x23d kernel/sched/core.c:6798
 slab_pre_alloc_hook mm/slab.h:565 [inline]
 slab_alloc_node mm/slab.c:3227 [inline]
 kmem_cache_alloc_node_trace+0x272/0x790 mm/slab.c:3593
 __do_kmalloc_node mm/slab.c:3615 [inline]
 __kmalloc_node+0x38/0x60 mm/slab.c:3623
 kmalloc_node include/linux/slab.h:578 [inline]
 kvmalloc_node+0x61/0xf0 mm/util.c:574
 kvmalloc include/linux/mm.h:645 [inline]
 kvzalloc include/linux/mm.h:653 [inline]
 bucket_table_alloc+0x8b/0x480 lib/rhashtable.c:175
 rhashtable_init+0x3d2/0x750 lib/rhashtable.c:1054
 nf_flow_table_init+0x16d/0x310 net/netfilter/nf_flow_table_core.c:498
 tcf_ct_flow_table_get+0xe33/0x1700 net/sched/act_ct.c:82
 tcf_ct_init+0xba4/0x18a6 net/sched/act_ct.c:1050
 tcf_action_init_1+0x697/0xa20 net/sched/act_api.c:945
 tcf_action_init+0x1e9/0x2f0 net/sched/act_api.c:1001
 tcf_action_add+0xdb/0x370 net/sched/act_api.c:1411
 tc_ctl_action+0x366/0x456 net/sched/act_api.c:1466
 rtnetlink_rcv_msg+0x44e/0xad0 net/core/rtnetlink.c:5440
 netlink_rcv_skb+0x15a/0x410 net/netlink/af_netlink.c:2478
 netlink_unicast_kernel net/netlink/af_netlink.c:1303 [inline]
 netlink_unicast+0x537/0x740 net/netlink/af_netlink.c:1329
 netlink_sendmsg+0x882/0xe10 net/netlink/af_netlink.c:1918
 sock_sendmsg_nosec net/socket.c:652 [inline]
 sock_sendmsg+0xcf/0x120 net/socket.c:672
 ____sys_sendmsg+0x6b9/0x7d0 net/socket.c:2343
 ___sys_sendmsg+0x100/0x170 net/socket.c:2397
 __sys_sendmsg+0xec/0x1b0 net/socket.c:2430
 do_syscall_64+0xf6/0x790 arch/x86/entry/common.c:294
 entry_SYSCALL_64_after_hwframe+0x49/0xbe
RIP: 0033:0x4403d9
Code: 18 89 d0 c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 00 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 0f 83 fb 13 fc ff c3 66 2e 0f 1f 84 00 00 00 00
RSP: 002b:00007ffd719af218 EFLAGS: 00000246 ORIG_RAX: 000000000000002e
RAX: ffffffffffffffda RBX: 00000000004002c8 RCX: 00000000004403d9
RDX: 0000000000000000 RSI: 0000000020000300 RDI: 0000000000000003
RBP: 00000000006ca018 R08: 0000000000000005 R09: 00000000004002c8
R10: 0000000000000008 R11: 00000000000

Fixes: c34b961a

 ("net/sched: act_ct: Create nf flow table per zone")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Paul Blakey <paulb@mellanox.com>
Cc: Jiri Pirko <jiri@mellanox.com>
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The rmnet_vnd_setup(), which is the callback of ->ndo_start_xmit() is
allowed to call concurrently because it uses RCU protected data.
So, it doesn't need tx lock.

Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Taehee Yoo says:

====================
bareudp: several code cleanup for bareudp module

This patchset is to cleanup bareudp module code.

1. The first patch is to add module alias
In the current bareudp code, there is no module alias.
So, RTNL couldn't load bareudp module automatically.

2. The second patch is to add extack message.
The extack error message is useful for noticing specific errors
when command is failed.

3. The third patch is to remove unnecessary udp_encap_enable().
In the bareudp_socket_create(), udp_encap_enable() is called.
But, the it's already called in the setup_udp_tunnel_sock().
So, it could be removed.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

In the current code, udp_encap_enable() is called in
bareudp_socket_create().
But, setup_udp_tunnel_sock() internally calls udp_encap_enable().
So, udp_encap_enable() is unnecessary.

Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

When bareudp netlink command fails, it doesn't print any error message.
So, users couldn't know the exact reason.
In order to tell the exact reason to the user, the extack error message
is used in this patch.

Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

In the current bareudp code, there is no module alias.
So, RTNL couldn't load bareudp module automatically.

Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Rohit Maheshwari says:

====================
cxgb4/chcr: ktls tx ofld support on T6 adapter

This series of patches add support for kernel tls offload in Tx direction,
over Chelsio T6 NICs. SKBs marked as decrypted will be treated as tls plain
text packets and then offloaded to encrypt using network device (chelsio T6
adapter).

This series is broken down as follows:

Patch 1 defines a new macro and registers tls_dev_add and tls_dev_del
callbacks. When tls_dev_add gets called we send a connection request to
our hardware and to make HW understand about tls offload. Its a partial
connection setup and only ipv4 part is done.

Patch 2 handles the HW response of the connection request and then we
request to update TCB and handle it's HW response as well. Also we save
crypto key locally. Only supporting TLS_CIPHER_AES_GCM_128_KEY_SIZE.

Patch 3 handles tls marked skbs (decrypted bit set) and sends it to ULD for
crypto handling. This code has a minimal portion of tx handler, to handle
only one complete record per skb.

Patch 4 hanldes partial end part of records. Also added logic to handle
multiple records in one single skb. It also adds support to send out tcp
option(/s) if exists in skb. If a record is partial but has end part of a
record, we'll fetch complete record and then only send it to HW to generate
HASH on complete record.

Patch 5 handles partial first or middle part of record, it uses AES_CTR to
encrypt the partial record. If we are trying to send middle record, it's
start should be 16 byte aligned, so we'll fetch few earlier bytes from the
record and then send it to HW for encryption.

Patch 6 enables ipv6 support and also includes ktls startistics.

v1->v2:
- mark tcb state to close in tls_dev_del.
- u_ctx is now picked from adapter structure.
- clear atid in case of failure.
- corrected ULP_CRYPTO_KTLS_INLINE value.
- optimized tcb update using control queue.
- state machine handling when earlier states received.
- chcr_write_cpl_set_tcb_ulp  function is shifted to patch3.
- un-necessary updating left variable.

v2->v3:
- add empty line after variable declaration.
- local variable declaration in reverse christmas tree ordering.

v3->v4:
- replaced kfree_skb with dev_kfree_skb_any.
- corrected error message reported by kbuild test robot <lkp@intel.com>
- mss calculation logic.
- correct place for Alloc skb check.
- Replaced atomic_t with atomic64_t
- added few more statistics counters.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

Adding ipv6 support and ktls related statistics.

v1->v2:
- added blank lines at 2 places.

v3->v4:
- Replaced atomic_t with atomic64_t
- added few necessary stat counters.

Signed-off-by: Rohit Maheshwari <rohitm@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch contains handling of first part or middle part of the record.
When we get a middle record, we will fetch few already sent bytes to
make packet start 16 byte aligned.
And if the packet has only the header part, we don't need to send it for
packet encryption, send that packet as a plaintext.

v1->v2:
- un-necessary updating left variable.

v3->v4:
- replaced kfree_skb with dev_kfree_skb_any.

Signed-off-by: Rohit Maheshwari <rohitm@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

TCP segment can chop a record in any order. Record can either be
complete or it can be partial (first part which contains header,
middle part which doesn't have header or TAG, and the end part
which contains TAG. This patch handles partial end part of a tx
record. In case of partial end part's, driver will send complete
record to HW, so that HW will calculate GHASH (TAG) of complete
packet.
Also added support to handle multiple records in a segment.

v1->v2:
- miner change in calling chcr_write_cpl_set_tcb_ulp.
- no need of checking return value of chcr_ktls_write_tcp_options.

v3->v4:
- replaced kfree_skb with dev_kfree_skb_any.

Signed-off-by: Rohit Maheshwari <rohitm@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Added tx handling in this patch. This includes handling of segments
contain single complete record.

v1->v2:
- chcr_write_cpl_set_tcb_ulp is added in this patch.

v3->v4:
- mss calculation logic.
- replaced kfree_skb with dev_kfree_skb_any.
- corrected error message reported by kbuild test robot <lkp@intel.com>

Signed-off-by: Rohit Maheshwari <rohitm@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

As part of this patch generated and saved crypto keys, handled HW
response of act_open_req and set_tcb_req. Defined connection state
update.

v1->v2:
- optimized tcb update using control queue.
- state machine handling when earlier states received.

v2->v3:
- Added one empty line after function declaration.

Signed-off-by: Rohit Maheshwari <rohitm@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

A new macro is defined to enable ktls tx offload support on Chelsio
T6 adapter. And if this macro is enabled, cxgb4 will send mailbox to
enable or disable ktls settings on HW.
In chcr, enabled tx offload flag in netdev and registered tls_dev_add
and tls_dev_del.

v1->v2:
- mark tcb state to close in tls_dev_del.
- u_ctx is now picked from adapter structure.
- clear atid in case of failure.
- corrected ULP_CRYPTO_KTLS_INLINE value.

v2->v3:
- add empty line after variable declaration.
- local variable declaration in reverse christmas tree ordering.

Signed-off-by: Rohit Maheshwari <rohitm@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Jiri Pirko says:

====================
net: allow user specify TC action HW stats type

Currently, when user adds a TC action and the action gets offloaded,
the user expects the HW stats to be counted and included in stats dump.
However, since drivers may implement different types of counting, there
is no way to specify which one the user is interested in.

For example for mlx5, only delayed counters are available as the driver
periodically polls for updated stats.

In case of mlxsw, the counters are queried on dump time. However, the
HW resources for this type of counters is quite limited (couple of
thousands). This limits the amount of supported offloaded filters
significantly. Without counter assigned, the HW is capable to carry
millions of those.

On top of that, mlxsw HW is able to support delayed counters as well in
greater numbers. That is going to be added in a follow-up patch.

This patchset allows user to specify one of the following types of HW
stats for added action:
immediate - queried during dump time
delayed - polled from HW periodically or sent by HW in async manner
disabled - no stats needed

Note that if "hw_stats" option is not passed, user does not care about
the type, just expects any type of stats.

Examples:
$ tc filter add dev enp0s16np28 ingress proto ip handle 1 pref 1 flower skip_sw dst_ip 192.168.1.1 action drop hw_stats disabled
$ tc -s filter show dev enp0s16np28 ingress
filter protocol ip pref 1 flower chain 0
filter protocol ip pref 1 flower chain 0 handle 0x1
  eth_type ipv4
  dst_ip 192.168.1.1
  skip_sw
  in_hw in_hw_count 2
        action order 1: gact action drop
         random type none pass val 0
         index 1 ref 1 bind 1 installed 7 sec used 2 sec
        Action statistics:
        Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
        backlog 0b 0p requeues 0
        hw_stats disabled

$ tc filter add dev enp0s16np28 ingress proto ip handle 1 pref 1 flower skip_sw dst_ip 192.168.1.1 action drop hw_stats immediate
$ tc -s filter show dev enp0s16np28 ingress
filter protocol ip pref 1 flower chain 0
filter protocol ip pref 1 flower chain 0 handle 0x1
  eth_type ipv4
  dst_ip 192.168.1.1
  skip_sw
  in_hw in_hw_count 2
        action order 1: gact action drop
         random type none pass val 0
         index 1 ref 1 bind 1 installed 11 sec used 4 sec
        Action statistics:
        Sent 102 bytes 1 pkt (dropped 1, overlimits 0 requeues 0)
        Sent software 0 bytes 0 pkt
        Sent hardware 102 bytes 1 pkt
        backlog 0b 0p requeues 0
        hw_stats immediate
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

Currently, user who is adding an action expects HW to report stats,
however it does not have exact expectations about the stats types.
That is aligned with TCA_ACT_HW_STATS_TYPE_ANY.

Allow user to specify the type of HW stats for an action and require it.

Pass the information down to flow_offload layer.

Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Introduce new type for disabled HW stats and allow the value in
mlxsw offload.

Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Set a flag in case rule counter was created. Only query the device for
stats of a rule, which has the valid counter assigned.

Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Introduce new type for delayed HW stats and allow the value in
mlx5 offload.

Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>