t-queue
Tony Nguyen says:
====================
100GbE Intel Wired LAN Driver Updates 2022-04-07
Alexander Lobakin says:
This hunts down several places around packet templates/dummies for
switch rules which are either repetitive, fragile or just not
really readable code.
It's a common need to add new packet templates and to review such
changes as well, try to simplify both with the help of a pair
macros and aliases.
ice_find_dummy_packet() became very complex at this point with tons
of nested if-elses. It clearly showed this approach does not scale,
so convert its logics to the simple mask-match + static const array.
bloat-o-meter is happy about that (built w/ LLVM 13):
add/remove: 0/1 grow/shrink: 1/1 up/down: 2/-1058 (-1056)
Function old new delta
ice_fill_adv_dummy_packet 289 291 +2
ice_adv_add_update_vsi_list 201 - -201
ice_add_adv_rule 2950 2093 -857
Total: Before=414512, After=413456, chg -0.25%
add/remove: 53/52 grow/shrink: 0/0 up/down: 4660/-3988 (672)
RO Data old new delta
ice_dummy_pkt_profiles - 672 +672
Total: Before=37895, After=38567, chg +1.77%
Diffstat also looks nice, and adding new packet templates now takes
less lines.
We'll probably come out with dynamic template crafting in a while,
but for now let's improve what we have currently.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Trade text size for rodata size and replace tons of nested if-elses
to the const mask match based structs. The almost entire
ice_find_dummy_packet() now becomes just one plain while-increment
loop. The order in ice_dummy_pkt_profiles[] should be same with the
if-elses order previously, as masks become less and less strict
through the array to follow the original code flow.
Apart from removing 80 locs of 4-level if-elses, it brings a solid
text size optimization:
add/remove: 0/1 grow/shrink: 1/1 up/down: 2/-1058 (-1056)
Function old new delta
ice_fill_adv_dummy_packet 289 291 +2
ice_adv_add_update_vsi_list 201 - -201
ice_add_adv_rule 2950 2093 -857
Total: Before=414512, After=413456, chg -0.25%
add/remove: 53/52 grow/shrink: 0/0 up/down: 4660/-3988 (672)
RO Data old new delta
ice_dummy_pkt_profiles - 672 +672
Total: Before=37895, After=38567, chg +1.77%
Signed-off-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Reviewed-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com>
Tested-by: Marcin Szycik <marcin.szycik@linux.intel.com>
Tested-by: Sandeep Penigalapati <sandeep.penigalapati@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Declarations of dummy/template packet headers and offsets can be
minified to improve readability and simplify adding new templates.
Move all the repetitive constructions into two macros and let them
do the name and type expansions.
Linewrap removal is yet another positive side effect.
Signed-off-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Reviewed-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com>
Tested-by: Marcin Szycik <marcin.szycik@linux.intel.com>
Tested-by: Sandeep Penigalapati <sandeep.penigalapati@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
ice_find_dummy_packet() contains a lot of boilerplate code and a
nice room for copy-paste mistakes.
Instead of passing 3 separate pointers back and forth to get packet
template (dummy) params, directly return a structure containing
them. Then, use a macro to compose compound literals and avoid code
duplication on return path.
Now, dummy packet type/name is needed only once to return a full
correct triple pkt-pkt_len-offsets, and those are all one-liners.
dummy_ipv4_gtpu_ipv4_packet_offsets is just moved around and renamed
(as well as dummy_ipv6_gtp_packet_offsets) with no function changes.
Signed-off-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Reviewed-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com>
Tested-by: Marcin Szycik <marcin.szycik@linux.intel.com>
Tested-by: Sandeep Penigalapati <sandeep.penigalapati@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
A loop performing header modification according to the provided mask
in ice_fill_adv_dummy_packet() is very cryptic (and error-prone).
Replace two identical cast-deferences with a variable. Replace three
struct-member-array-accesses with a variable. Invert the condition,
reduce the indentation by one -> eliminate line wraps.
Signed-off-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Reviewed-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com>
Tested-by: Marcin Szycik <marcin.szycik@linux.intel.com>
Tested-by: Sandeep Penigalapati <sandeep.penigalapati@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
ice_adv_lkup_elem fields h_u and m_u are being accessed as raw u16
arrays in several places.
To reduce cast and braces burden, add permanent array-of-u16 aliases
with the same size as the `union ice_prot_hdr` itself via anonymous
unions to the actual struct declaration, and just access them
directly.
This:
- removes the need to cast the union to u16[] and then dereference
it each time -> reduces the horizon for potential bugs;
- improves -Warray-bounds coverage -- the array size is now known
at compilation time;
- addresses cppcheck complaints.
Signed-off-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Reviewed-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com>
Tested-by: Marcin Szycik <marcin.szycik@linux.intel.com>
Tested-by: Sandeep Penigalapati <sandeep.penigalapati@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Currently when XDP rings are created, each descriptor gets its DD bit
set, which turns out to be the wrong approach as it can lead to a
situation where more descriptors get cleaned than it was supposed to,
e.g. when AF_XDP busy poll is run with a large batch size. In this
situation, the driver would request for more buffers than it is able to
handle.
Fix this by not setting the DD bits in ice_xdp_alloc_setup_rings(). They
should be initialized to zero instead.
Fixes: 9610bd988d ("ice: optimize XDP_TX workloads")
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Tested-by: Shwetha Nagaraju <shwetha.nagaraju@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
ICE_DOWN is dedicated for pf->state. Check for ICE_VSI_DOWN being set on
vsi->state in ice_xsk_wakeup().
Fixes: 2d4238f556 ("ice: Add support for AF_XDP")
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Tested-by: Shwetha Nagaraju <shwetha.nagaraju@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Unfortunately, the ice driver doesn't respect the RCU critical section that
XSK wakeup is surrounded with. To fix this, add synchronize_rcu() calls to
paths that destroy resources that might be in use.
This was addressed in other AF_XDP ZC enabled drivers, for reference see
for example commit b3873a5be7 ("net/i40e: Fix concurrency issues
between config flow and XSK")
Fixes: efc2214b60 ("ice: Add support for XDP")
Fixes: 2d4238f556 ("ice: Add support for AF_XDP")
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Tested-by: Shwetha Nagaraju <shwetha.nagaraju@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Disable check for queue being enabled in ice_vc_dis_qs_msg, because
there could be a case when queues were created, but were not enabled.
We still need to delete those queues.
Normal workflow for VF looks like:
Enable path:
VIRTCHNL_OP_ADD_ETH_ADDR (opcode 10)
VIRTCHNL_OP_CONFIG_VSI_QUEUES (opcode 6)
VIRTCHNL_OP_ENABLE_QUEUES (opcode 8)
Disable path:
VIRTCHNL_OP_DISABLE_QUEUES (opcode 9)
VIRTCHNL_OP_DEL_ETH_ADDR (opcode 11)
The issue appears only in stress conditions when VF is enabled and
disabled very fast.
Eventually there will be a case, when queues are created by
VIRTCHNL_OP_CONFIG_VSI_QUEUES, but are not enabled by
VIRTCHNL_OP_ENABLE_QUEUES.
In turn, these queues are not deleted by VIRTCHNL_OP_DISABLE_QUEUES,
because there is a check whether queues are enabled in
ice_vc_dis_qs_msg.
When we bring up the VF again, we will see the "Failed to set LAN Tx queue
context" error during VIRTCHNL_OP_CONFIG_VSI_QUEUES step. This
happens because old 16 queues were not deleted and VF requests to create
16 more, but ice_sched_get_free_qparent in ice_ena_vsi_txq would fail to
find a parent node for first newly requested queue (because all nodes
are allocated to 16 old queues).
Testing Hints:
Just enable and disable VF fast enough, so it would be disabled before
reaching VIRTCHNL_OP_ENABLE_QUEUES.
while true; do
ip link set dev ens785f0v0 up
sleep 0.065 # adjust delay value for you machine
ip link set dev ens785f0v0 down
done
Fixes: 77ca27c417 ("ice: add support for virtchnl_queue_select.[tx|rx]_queues bitmap")
Signed-off-by: Anatolii Gerasymenko <anatolii.gerasymenko@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Alice Michael <alice.michael@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
When VF is freshly created, but not brought up, ring->txq_teid
value is by default set to 0.
But 0 is a valid TEID. On some platforms the Root Node of
Tx scheduler has a TEID = 0. This can cause issues as shown below.
The proper way is to set ring->txq_teid to ICE_INVAL_TEID (0xFFFFFFFF).
Testing Hints:
echo 1 > /sys/class/net/ens785f0/device/sriov_numvfs
ip link set dev ens785f0v0 up
ip link set dev ens785f0v0 down
If we have freshly created VF and quickly turn it on and off, so there
would be no time to reach VIRTCHNL_OP_CONFIG_VSI_QUEUES stage, then
VIRTCHNL_OP_DISABLE_QUEUES stage will fail with error:
[ 639.531454] disable queue 89 failed 14
[ 639.532233] Failed to disable LAN Tx queues, error: ICE_ERR_AQ_ERROR
[ 639.533107] ice 0000:02:00.0: Failed to stop Tx ring 0 on VSI 5
The reason for the fail is that we are trying to send AQ command to
delete queue 89, which has never been created and receive an "invalid
argument" error from firmware.
As this queue has never been created, it's teid and ring->txq_teid
have default value 0.
ice_dis_vsi_txq has a check against non-existent queues:
node = ice_sched_find_node_by_teid(pi->root, q_teids[i]);
if (!node)
continue;
But on some platforms the Root Node of Tx scheduler has a teid = 0.
Hence, ice_sched_find_node_by_teid finds a node with teid = 0 (it is
pi->root), and we go further to submit an erroneous request to firmware.
Fixes: 37bb839012 ("ice: Move common functions out of ice_main.c part 7/7")
Signed-off-by: Anatolii Gerasymenko <anatolii.gerasymenko@intel.com>
Reviewed-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Alice Michael <alice.michael@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Handling of all-multicast flag and associated multicast promiscuous
mode is broken in ice driver. When an user switches allmulticast
flag on or off the driver checks whether any VLANs are configured
over the interface (except default VLAN 0).
If any extra VLANs are registered it enables multicast promiscuous
mode for all these VLANs (including default VLAN 0) using
ICE_SW_LKUP_PROMISC_VLAN look-up type. In this situation all
multicast packets tagged with known VLAN ID or untagged are received
and multicast packets tagged with unknown VLAN ID ignored.
If no extra VLANs are registered (so only VLAN 0 exists) it enables
multicast promiscuous mode for VLAN 0 and uses ICE_SW_LKUP_PROMISC
look-up type. In this situation any multicast packets including
tagged ones are received.
The driver handles IFF_ALLMULTI in ice_vsi_sync_fltr() this way:
ice_vsi_sync_fltr() {
...
if (changed_flags & IFF_ALLMULTI) {
if (netdev->flags & IFF_ALLMULTI) {
if (vsi->num_vlans > 1)
ice_set_promisc(..., ICE_MCAST_VLAN_PROMISC_BITS);
else
ice_set_promisc(..., ICE_MCAST_PROMISC_BITS);
} else {
if (vsi->num_vlans > 1)
ice_clear_promisc(..., ICE_MCAST_VLAN_PROMISC_BITS);
else
ice_clear_promisc(..., ICE_MCAST_PROMISC_BITS);
}
}
...
}
The code above depends on value vsi->num_vlan that specifies number
of VLANs configured over the interface (including VLAN 0) and
this is problem because that value is modified in NDO callbacks
ice_vlan_rx_add_vid() and ice_vlan_rx_kill_vid().
Scenario 1:
1. ip link set ens7f0 allmulticast on
2. ip link add vlan10 link ens7f0 type vlan id 10
3. ip link set ens7f0 allmulticast off
4. ip link set ens7f0 allmulticast on
[1] In this scenario IFF_ALLMULTI is enabled and the driver calls
ice_set_promisc(..., ICE_MCAST_PROMISC_BITS) that installs
multicast promisc rule with non-VLAN look-up type.
[2] Then VLAN with ID 10 is added and vsi->num_vlan incremented to 2
[3] Command switches IFF_ALLMULTI off and the driver calls
ice_clear_promisc(..., ICE_MCAST_VLAN_PROMISC_BITS) but this
call is effectively NOP because it looks for multicast promisc
rules for VLAN 0 and VLAN 10 with VLAN look-up type but no such
rules exist. So the all-multicast remains enabled silently
in hardware.
[4] Command tries to switch IFF_ALLMULTI on and the driver calls
ice_clear_promisc(..., ICE_MCAST_PROMISC_BITS) but this call
fails (-EEXIST) because non-VLAN multicast promisc rule already
exists.
Scenario 2:
1. ip link add vlan10 link ens7f0 type vlan id 10
2. ip link set ens7f0 allmulticast on
3. ip link add vlan20 link ens7f0 type vlan id 20
4. ip link del vlan10 ; ip link del vlan20
5. ip link set ens7f0 allmulticast off
[1] VLAN with ID 10 is added and vsi->num_vlan==2
[2] Command switches IFF_ALLMULTI on and driver installs multicast
promisc rules with VLAN look-up type for VLAN 0 and 10
[3] VLAN with ID 20 is added and vsi->num_vlan==3 but no multicast
promisc rules is added for this new VLAN so the interface does
not receive MC packets from VLAN 20
[4] Both VLANs are removed but multicast rule for VLAN 10 remains
installed so interface receives multicast packets from VLAN 10
[5] Command switches IFF_ALLMULTI off and because vsi->num_vlan is 1
the driver tries to remove multicast promisc rule for VLAN 0
with non-VLAN look-up that does not exist.
All-multicast looks disabled from user point of view but it
is partially enabled in HW (interface receives all multicast
packets either untagged or tagged with VLAN ID 10)
To resolve these issues the patch introduces these changes:
1. Adds handling for IFF_ALLMULTI to ice_vlan_rx_add_vid() and
ice_vlan_rx_kill_vid() callbacks. So when VLAN is added/removed
and IFF_ALLMULTI is enabled an appropriate multicast promisc
rule for that VLAN ID is added/removed.
2. In ice_vlan_rx_add_vid() when first VLAN besides VLAN 0 is added
so (vsi->num_vlan == 2) and IFF_ALLMULTI is enabled then look-up
type for existing multicast promisc rule for VLAN 0 is updated
to ICE_MCAST_VLAN_PROMISC_BITS.
3. In ice_vlan_rx_kill_vid() when last VLAN besides VLAN 0 is removed
so (vsi->num_vlan == 1) and IFF_ALLMULTI is enabled then look-up
type for existing multicast promisc rule for VLAN 0 is updated
to ICE_MCAST_PROMISC_BITS.
4. Both ice_vlan_rx_{add,kill}_vid() have to run under ICE_CFG_BUSY
bit protection to avoid races with ice_vsi_sync_fltr() that runs
in ice_service_task() context.
5. Bit ICE_VSI_VLAN_FLTR_CHANGED is use-less and can be removed.
6. Error messages added to ice_fltr_*_vsi_promisc() helper functions
to avoid them in their callers
7. Small improvements to increase readability
Fixes: 5eda8afd6b ("ice: Add support for PF/VF promiscuous mode")
Signed-off-by: Ivan Vecera <ivecera@redhat.com>
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Signed-off-by: Alice Michael <alice.michael@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Commit 2ccc1c1ccc ("ice: Remove excess error variables") merged
the usage of 'status' and 'err' variables into single one in
function ice_set_mac_address(). Unfortunately this causes
a regression when call of ice_fltr_add_mac() returns -EEXIST because
this return value does not indicate an error in this case but
value of 'err' remains to be -EEXIST till the end of the function
and is returned to caller.
Prior mentioned commit this does not happen because return value of
ice_fltr_add_mac() was stored to 'status' variable first and
if it was -EEXIST then 'err' remains to be zero.
Fix the problem by reset 'err' to zero when ice_fltr_add_mac()
returns -EEXIST.
Fixes: 2ccc1c1ccc ("ice: Remove excess error variables")
Signed-off-by: Ivan Vecera <ivecera@redhat.com>
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Acked-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Signed-off-by: Alice Michael <alice.michael@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
VSI is set as default forwarding one when promisc mode is set for
PF interface, when PF is switched to switchdev mode or when VF
driver asks to enable allmulticast or promisc mode for the VF
interface (when vf-true-promisc-support priv flag is off).
The third case is buggy because in that case VSI associated with
VF remains as default one after VF removal.
Reproducer:
1. Create VF
echo 1 > sys/class/net/ens7f0/device/sriov_numvfs
2. Enable allmulticast or promisc mode on VF
ip link set ens7f0v0 allmulticast on
ip link set ens7f0v0 promisc on
3. Delete VF
echo 0 > sys/class/net/ens7f0/device/sriov_numvfs
4. Try to enable promisc mode on PF
ip link set ens7f0 promisc on
Although it looks that promisc mode on PF is enabled the opposite
is true because ice_vsi_sync_fltr() responsible for IFF_PROMISC
handling first checks if any other VSI is set as default forwarding
one and if so the function does not do anything. At this point
it is not possible to enable promisc mode on PF without re-probe
device.
To resolve the issue this patch clear default forwarding VSI
during ice_vsi_release() when the VSI to be released is the default
one.
Fixes: 01b5e89aab ("ice: Add VF promiscuous support")
Signed-off-by: Ivan Vecera <ivecera@redhat.com>
Reviewed-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com>
Reviewed-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Alice Michael <alice.michael@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Ice driver tries to always create XDP rings array to be
num_possible_cpus() sized, regardless of user's queue count setting that
can be changed via ethtool -L for example.
Currently, ice_tx_xsk_pool() calculates the qid by decrementing the
ring->q_index by the count of XDP queues, but ring->q_index is set to 'i
+ vsi->alloc_txq'.
When user did ethtool -L $IFACE combined 1, alloc_txq is 1, but
vsi->num_xdp_txq is still num_possible_cpus(). Then, ice_tx_xsk_pool()
will do OOB access and in the final result ring would not get xsk_pool
pointer assigned. Then, each ice_xsk_wakeup() call will fail with error
and it will not be possible to get into NAPI and do the processing from
driver side.
Fix this by decrementing vsi->alloc_txq instead of vsi->num_xdp_txq from
ring-q_index in ice_tx_xsk_pool() so the calculation is reflected to the
setting of ring->q_index.
Fixes: 22bf877e52 ("ice: introduce XDP_TX fallback path")
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20220328142123.170157-5-maciej.fijalkowski@intel.com
This can happen with big budget values and some breakage of re-filling
descriptors as we do not clear the entry that ntu is pointing at the end
of ice_alloc_rx_bufs_zc. So if ntc is at ntu then it might be the case
that status_error0 has an old, uncleared value and ntc would go over
with processing which would result in false results.
Break Rx loop when ntc == ntu to avoid broken behavior.
Fixes: 3876ff525d ("ice: xsk: Handle SW XDP ring wrap and bump tail more often")
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20220328142123.170157-4-maciej.fijalkowski@intel.com
The NIC Tx ring completion routine cleans entries from the ring in
batches. However, it processes one more batch than it is supposed
to. Note that this does not matter from a functionality point of view
since it will not find a set DD bit for the next batch and just exit
the loop. But from a performance perspective, it is faster to
terminate the loop before and not issue an expensive read over PCIe to
get the DD bit.
Fixes: 126cdfe100 ("ice: xsk: Improve AF_XDP ZC Tx and use batching API")
Signed-off-by: Magnus Karlsson <magnus.karlsson@intel.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20220328142123.170157-3-maciej.fijalkowski@intel.com
ice_send_event_to_aux() eventually descends to mutex_lock()
(-> might_sched()), so it must not be called under non-task
context. However, at least two fixes have happened already for the
bug splats occurred due to this function being called from atomic
context.
To make the emergency landings softer, bail out early when executed
in non-task context emitting a warn splat only once. This way we
trade some events being potentially lost for system stability and
avoid any related hangs and crashes.
Fixes: 348048e724 ("ice: Implement iidc operations")
Signed-off-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Tested-by: Michal Kubiak <michal.kubiak@intel.com>
Reviewed-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Acked-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
There's a kernel BUG splat on processing aux critical error
interrupts in ice_misc_intr():
[ 2100.917085] BUG: scheduling while atomic: swapper/15/0/0x00010000
...
[ 2101.060770] Call Trace:
[ 2101.063229] <IRQ>
[ 2101.065252] dump_stack+0x41/0x60
[ 2101.068587] __schedule_bug.cold.100+0x4c/0x58
[ 2101.073060] __schedule+0x6a4/0x830
[ 2101.076570] schedule+0x35/0xa0
[ 2101.079727] schedule_preempt_disabled+0xa/0x10
[ 2101.084284] __mutex_lock.isra.7+0x310/0x420
[ 2101.088580] ? ice_misc_intr+0x201/0x2e0 [ice]
[ 2101.093078] ice_send_event_to_aux+0x25/0x70 [ice]
[ 2101.097921] ice_misc_intr+0x220/0x2e0 [ice]
[ 2101.102232] __handle_irq_event_percpu+0x40/0x180
[ 2101.106965] handle_irq_event_percpu+0x30/0x80
[ 2101.111434] handle_irq_event+0x36/0x53
[ 2101.115292] handle_edge_irq+0x82/0x190
[ 2101.119148] handle_irq+0x1c/0x30
[ 2101.122480] do_IRQ+0x49/0xd0
[ 2101.125465] common_interrupt+0xf/0xf
[ 2101.129146] </IRQ>
...
As Andrew correctly mentioned previously[0], the following call
ladder happens:
ice_misc_intr() <- hardirq
ice_send_event_to_aux()
device_lock()
mutex_lock()
might_sleep()
might_resched() <- oops
Add a new PF state bit which indicates that an aux critical error
occurred and serve it in ice_service_task() in process context.
The new ice_pf::oicr_err_reg is read-write in both hardirq and
process contexts, but only 3 bits of non-critical data probably
aren't worth explicit synchronizing (and they're even in the same
byte [31:24]).
[0] https://lore.kernel.org/all/YeSRUVmrdmlUXHDn@lunn.ch
Fixes: 348048e724 ("ice: Implement iidc operations")
Signed-off-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Tested-by: Michal Kubiak <michal.kubiak@intel.com>
Acked-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Tony Nguyen says:
====================
100GbE Intel Wired LAN Driver Updates 2022-03-16
This series contains updates to gtp and ice driver.
Wojciech fixes smatch reported inconsistent indenting for gtp and ice.
Yang Yingliang fixes a couple of return value checks for GNSS to IS_PTR
instead of null.
Jacob adds support for trace events on tx timestamps.
* '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/next-queue:
ice: add trace events for tx timestamps
ice: fix return value check in ice_gnss.c
ice: Fix inconsistent indenting in ice_switch
gtp: Fix inconsistent indenting
====================
Link: https://lore.kernel.org/r/20220316204024.3201500-1-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
We've previously run into many issues related to the latency of a Tx
timestamp completion with the ice hardware. It can be difficult to
determine the root cause of a slow Tx timestamp. To aid in this,
introduce new trace events which capture timing data about when the
driver reaches certain points while processing a transmit timestamp
* ice_tx_tstamp_request: Trace when the stack initiates a new timestamp
request.
* ice_tx_tstamp_fw_req: Trace when the driver begins a read of the
timestamp register in the work thread.
* ice_tx_tstamp_fw_done: Trace when the driver finishes reading a
timestamp register in the work thread.
* ice_tx_tstamp_complete: Trace when the driver submits the skb back to
the stack with a completed Tx timestamp.
These trace events can be enabled using the standard trace event
subsystem exposed by the ice driver. If they are disabled, they become
no-ops with no run time cost.
The following is a simple GNU AWK script which can highlight one
potential way to use the trace events to capture latency data from the
trace buffer about how long the driver takes to process a timestamp:
-----
BEGIN {
PREC=256
}
# Detect requests
/tx_tstamp_request/ {
time=strtonum($4)
skb=$7
# Store the time of request for this skb
requests[skb] = time
printf("skb %s: idx %d at %.6f\n", skb, idx, time)
}
# Detect completions
/tx_tstamp_complete/ {
time=strtonum($4)
skb=$7
idx=$9
if (skb in requests) {
latency = (time - requests[skb]) * 1000
printf("skb %s: %.3f to complete\n", skb, latency)
if (latency > 4) {
printf(">>> HIGH LATENCY <<<\n")
}
printf("\n")
} else {
printf("!!! skb %s (idx %d) at %.6f\n", skb, idx, time)
}
}
-----
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Gurucharan <gurucharanx.g@intel.com> (A Contingent worker at Intel)
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
kthread_create_worker() and tty_alloc_driver() return ERR_PTR()
and never return NULL. The NULL test in the return value check
should be replaced with IS_ERR().
Fixes: 43113ff734 ("ice: add TTY for GNSS module for E810T device")
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Yang Yingliang <yangyingliang@huawei.com>
Tested-by: Gurucharan <gurucharanx.g@intel.com> (A Contingent worker at Intel)
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Fix the following warning as reported by smatch:
smatch warnings:
drivers/net/ethernet/intel/ice/ice_switch.c:5568 ice_find_dummy_packet() warn: inconsistent indenting
Fixes: 9a225f81f5 ("ice: Support GTP-U and GTP-C offload in switchdev")
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Wojciech Drewek <wojciech.drewek@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Currently fdir_fltr_lock is accessed in ice_vsi_release_all() function
after it is destroyed. Instead destroy mutex after ice_vsi_release_all.
Fixes: 40319796b7 ("ice: Add flow director support for channel mode")
Signed-off-by: Sudheer Mogilappagari <sudheer.mogilappagari@intel.com>
Tested-by: Bharathi Sreenivas <bharathi.sreenivas@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
It is possible to do NULL pointer dereference in routine that updates
Tx ring stats. Currently only stats and bytes are updated when ring
pointer is valid, but later on ring is accessed to propagate gathered Tx
stats onto VSI stats.
Change the existing logic to move to next ring when ring is NULL.
Fixes: e72bba2135 ("ice: split ice_ring onto Tx/Rx separate structs")
Reported-by: kernel test robot <lkp@intel.com>
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Acked-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com> (A Contingent worker at Intel)
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The ice_check_vf_init function takes both a PF and a VF pointer. Every
caller looks up the PF pointer from the VF structure. Some callers only
use of the PF pointer is call this function. Move the lookup inside
ice_check_vf_init and drop the unnecessary argument.
Cleanup the callers to drop the now unnecessary local variables. In
particular, replace the local PF pointer with a HW structure pointer in
ice_vc_get_vf_res_msg which simplifies a few accesses to the HW
structure in that function.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Just as we moved the generic virtualization library logic into
ice_vf_lib.c, move the virtchnl message handling into ice_virtchnl.c
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Before we move the virtchnl message handling from ice_sriov.c into
ice_virtchnl.c, cleanup some long line warnings to avoid checkpatch.pl
complaints.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The ice_reset_vf function performs actions which must be taken only
while holding the VF configuration lock. Some flows already acquired the
lock, while other flows must acquire it just for the reset function. Add
the ICE_VF_RESET_LOCK flag to the function so that it can handle taking
and releasing the lock instead at the appropriate scope.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
In some cases of resetting a VF, the PF would like to first notify the
VF that a reset is impending. This is currently done via
ice_vc_notify_vf_reset. A wrapper to ice_reset_vf, ice_vf_reset_vf, is
used to call this function first before calling ice_reset_vf.
In fact, every single call to ice_vc_notify_vf_reset occurs just prior
to a call to ice_vc_reset_vf.
Now that ice_reset_vf has flags, replace this separate call with an
ICE_VF_RESET_NOTIFY flag. This removes an unnecessary exported function
of ice_vc_notify_vf_reset, and also makes there be a single function to
reset VFs (ice_reset_vf).
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The ice_reset_vf function takes a boolean parameter which indicates
whether or not the reset is due to a VFLR event.
This is somewhat confusing to read because readers must interpret what
"true" and "false" mean when seeing a line of code like
"ice_reset_vf(vf, false)".
We will want to add another toggle to the ice_reset_vf in a following
change. To avoid proliferating many arguments, convert this function to
take flags instead. ICE_VF_RESET_VFLR will indicate if this is a VFLR
reset. A value of 0 indicates no flags.
One could argue that "ice_reset_vf(vf, 0)" is no more readable than
"ice_reset_vf(vf, false)".. However, this type of flags interface is
somewhat common and using 0 to mean "no flags" makes sense in this
context. We could bother to add a define for "ICE_VF_RESET_PLAIN" or
something similar, but this can be confusing since its not an actual bit
flag.
This paves the way to add another flag to the function in a following
change.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The ice_reset_vf function returns a boolean value indicating whether or
not the VF reset. This is a bit confusing since it means that callers
need to know how to interpret the return value when needing to indicate
an error.
Refactor the function and call sites to report a regular error code. We
still report success (i.e. return 0) in cases where the reset is in
progress or is disabled.
Existing callers don't care because they do not check the return value.
We keep the error code anyways instead of a void return because we
expect future code which may care about or at least report the error
value.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The ice_reset_all_vfs function returns true if any VFs were reset, and
false otherwise. However, no callers check the return value.
Drop this return value and make the function void since the callers do
not care about this.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The ice_reset_all_vfs function takes a parameter to handle whether its
operating after a VFLR event or not. This is not necessary as every
caller always passes true. Simplify the interface by removing the
parameter.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Now that the reset functions do not rely on Single Root specific
behavior, move the ice_reset_vf, ice_reset_all_vfs, and
ice_vf_rebuild_host_cfg functions and their dependent helper functions
out of ice_sriov.c and into ice_vf_lib.c
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
We're about to move ice_reset_vf out of ice_sriov.c and into
ice_vf_lib.c
One of the dev_err statements has a checkpatch.pl violation due to
putting the vf->vf_id on the same line as the dev_err. Fix this style
issue first before moving the code.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The ice driver currently supports virtualization using Single Root IOV,
with code in the ice_sriov.c file. In the future, we plan to also
implement support for Scalable IOV, which uses slightly different
hardware implementations for some functionality.
To eventually allow this, we introduce a new ice_vf_ops structure which
will contain the basic operations that are different between the two IOV
implementations. This primarily includes logic for how to handle the VF
reset registers, as well as what to do before and after rebuilding the
VF's VSI.
Implement these ops structures and call the ops table instead of
directly calling the SR-IOV specific function. This will allow us to
easily add the Scalable IOV implementation in the future. Additionally,
it helps separate the generalized VF logic from SR-IOV specifics. This
change allows us to move the reset logic out of ice_sriov.c and into
ice_vf_lib.c without placing any Single Root specific details into the
generic file.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
If we fail to clear the malicious VF indication after a VF reset, the
dev_dbg message which is printed uses the local variable 'i' when it
meant to use vf->vf_id. Fix this.
Fixes: 0891c89674 ("ice: warn about potentially malicious VFs")
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Introduce the ice_vf_lib.c file along with the ice_vf_lib.h and
ice_vf_lib_private.h header files.
These files will house the generic VF structures and access functions.
Move struct ice_vf and its dependent definitions into this new header
file.
The ice_vf_lib.c is compiled conditionally on CONFIG_PCI_IOV. Some of
its functionality is required by all driver files. However, some of its
functionality will only be required by other files also conditionally
compiled based on CONFIG_PCI_IOV.
Declaring these functions used only in CONFIG_PCI_IOV files in
ice_vf_lib.h is verbose. This is because we must provide a fallback
implementation for each function in this header since it is included in
files which may not be compiled with CONFIG_PCI_IOV.
Instead, introduce a new ice_vf_lib_private.h header which verifies that
CONFIG_PCI_IOV is enabled. This header is intended to be directly
included in .c files which are CONFIG_PCI_IOV only. Add a #error
indication that will complain if the file ever gets included by another
C file on a kernel with CONFIG_PCI_IOV disabled. Add a comment
indicating the nature of the file and why it is useful.
This makes it so that we can easily define functions exposed from
ice_vf_lib.c into other virtualization files without needing to add
fallback implementations for every single function.
This begins the path to separate out generic code which will be reused
by other virtualization implementations from ice_sriov.h and ice_sriov.c
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The ice_vc_cfg_promiscuous_mode_msg function directly checks
ICE_VIRTCHNL_VF_CAP_PRIVILEGE, instead of using the existing helper
function ice_is_vf_trusted. Switch this to use the helper function so
that all trusted checks are consistent. This aids in any potential
future refactor by ensuring consistent code.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
When ice_eswitch_configure fails, print an error message to make it more
obvious why VF initialization did not succeed.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Sandeep Penigalapati <sandeep.penigalapati@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The ice_ena_vfs function and some of its sub-functions like
ice_set_per_vf_res use a "if (<function>) { <print error> ; <exit> }"
flow. This flow discards specialized errors reported by the called
function.
This style is generally not preferred as it makes tracing error sources
more difficult. It also means we cannot log the actual error received
properly.
Refactor several calls in the ice_ena_vfs function that do this to catch
the error in the 'ret' variable. Report this in the messages, and then
return the more precise error value.
Doing this reveals that ice_set_per_vf_res returns -EINVAL or -EIO in
places where -ENOSPC makes more sense. Fix these calls up to return the
more appropriate value.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The ice_set_vf_port_vlan function is located in ice_sriov.c very far
away from the other .ndo operations that it is similar to. Move this so
that its located near the other .ndo operation definitions.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The API to control the VSI spoof checking for a VF VSI has three
functions: enable, disable, and set. The set function takes the VSI and
the VF and decides whether to call enable or disable based on the
vf->spoofchk field.
In some flows, vf->spoofchk is not yet set, such as the function used to
control the setting for a VF. (vf->spoofchk is only updated after a
success).
Simplify this API by refactoring ice_vf_set_spoofchk_cfg to be
"ice_vsi_apply_spoofchk" which takes the boolean and allows all callers
to avoid having to determine whether to call enable or disable
themselves.
This matches the expected callers better, and will prevent the need to
export more than one function when this code must be called from another
file.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The ICE_MAX_VF_COUNT field is defined in ice_sriov.h. This count is true
for SR-IOV but will not be true for all VF implementations, such as when
the ice driver supports Scalable IOV.
Rename this definition to clearly indicate ICE_MAX_SRIOV_VFS.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
A few more macros exist in ice_sriov.h which are not used anywhere.
These can be safely removed. Note that ICE_VIRTCHNL_VF_CAP_L2 capability
is set but never checked anywhere in the driver. Thus it is also safe to
remove.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The vc_ops structure is used to allow different handlers for virtchnl
commands when the driver is in representor mode. The current
implementation uses a copy of the ops table in each VF, and modifies
this copy dynamically.
The usual practice in kernel code is to store the ops table in a
constant structure and point to different versions. This has a number of
advantages:
1. Reduced memory usage. Each VF merely points to the correct table,
so they're able to re-use the same constant lookup table in memory.
2. Consistency. It becomes more difficult to accidentally update or
edit only one op call. Instead, the code switches to the correct
able by a single pointer write. In general this is atomic, either
the pointer is updated or its not.
3. Code Layout. The VF structure can store a pointer to the table
without needing to have the full structure definition defined prior
to the VF structure definition. This will aid in future refactoring
of code by allowing the VF pointer to be kept in ice_vf_lib.h while
the virtchnl ops table can be maintained in ice_virtchnl.h
There is one major downside in the case of the vc_ops structure. Most of
the operations in the table are the same between the two current
implementations. This can appear to lead to duplication since each
implementation must now fill in the complete table. It could make
spotting the differences in the representor mode more challenging.
Unfortunately, methods to make this less error prone either add
complexity overhead (macros using CPP token concatenation) or don't work
on all compilers we support (constant initializer from another constant
structure).
The cost of maintaining two structures does not out weigh the benefits
of the constant table model.
While we're making these changes, go ahead and rename the structure and
implementations with "virtchnl" instead of "vc_vf_". This will more
closely align with the planned file renaming, and avoid similar names when
we later introduce a "vf ops" table for separating Scalable IOV and
Single Root IOV implementations.
Leave the accessor/assignment functions in order to avoid issues with
compiling with options disabled. The interface makes it easier to handle
when CONFIG_PCI_IOV is disabled in the kernel.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Sandeep Penigalapati <sandeep.penigalapati@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Several headers in the ice driver include ice.h even though they are
themselves included by that header. The most notable of these is
ice_common.h, but several other headers also do this.
Such a recursive inclusion is problematic as it forces headers to be
included in a strict order, otherwise compilation errors can result. The
circular inclusions do not trigger an endless loop due to standard
header inclusion guards, however other errors can occur.
For example, ice_flow.h defines ice_rss_hash_cfg, which is used by
ice_sriov.h as part of the definition of ice_vf_hash_ip_ctx.
ice_flow.h includes ice_acl.h, which includes ice_common.h, and which
finally includes ice.h. Since ice.h itself includes ice_sriov.h, this
creates a circular dependency.
The definition in ice_sriov.h requires things from ice_flow.h, but
ice_flow.h itself will lead to trying to load ice_sriov.h as part of its
process for expanding ice.h. The current code avoids this issue by
having an implicit dependency without the include of ice_flow.h.
If we were to fix that so that ice_sriov.h explicitly depends on
ice_flow.h the following pattern would occur:
ice_flow.h -> ice_acl.h -> ice_common.h -> ice.h -> ice_sriov.h
At this point, during the expansion of, the header guard for ice_flow.h
is already set, so when ice_sriov.h attempts to load the ice_flow.h
header it is skipped. Then, we go on to begin including the rest of
ice_sriov.h, including structure definitions which depend on
ice_rss_hash_cfg. This produces a compiler warning because
ice_rss_hash_cfg hasn't yet been included. Remember, we're just at the
start of ice_flow.h!
If the order of headers is incorrect (ice_flow.h is not implicitly
loaded first in all files which include ice_sriov.h) then we get the
same failure.
Removing this recursive inclusion requires fixing a few cases where some
headers depended on the header inclusions from ice.h. In addition, a few
other changes are also required.
Most notably, ice_hw_to_dev is implemented as a macro in ice_osdep.h,
which is the likely reason that ice_common.h includes ice.h at all. This
macro implementation requires the full definition of ice_pf in order to
properly compile.
Fix this by moving it to a function declared in ice_main.c, so that we
do not require all files to depend on the layout of the ice_pf
structure.
Note that this change only fixes circular dependencies, but it does not
fully resolve all implicit dependencies where one header may depend on
the inclusion of another. I tried to fix as many of the implicit
dependencies as I noticed, but fixing them all requires a somewhat
tedious analysis of each header and attempting to compile it separately.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com> (A Contingent worker at Intel)
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The ice_virtchnl_pf.c and ice_virtchnl_pf.h files are where most of the
code for implementing Single Root IOV virtualization resides. This code
includes support for bringing up and tearing down VFs, hooks into the
kernel SR-IOV netdev operations, and for handling virtchnl messages from
VFs.
In the future, we plan to support Scalable IOV in addition to Single
Root IOV as an alternative virtualization scheme. This implementation
will re-use some but not all of the code in ice_virtchnl_pf.c
To prepare for this future, we want to refactor and split up the code in
ice_virtchnl_pf.c into the following scheme:
* ice_vf_lib.[ch]
Basic VF structures and accessors. This is where scheme-independent
code will reside.
* ice_virtchnl.[ch]
Virtchnl message handling. This is where the bulk of the logic for
processing messages from VFs using the virtchnl messaging scheme will
reside. This is separated from ice_vf_lib.c because it is distinct
and has a bulk of the processing code.
* ice_sriov.[ch]
Single Root IOV implementation, including initialization and the
routines for interacting with SR-IOV based netdev operations.
* (future) ice_siov.[ch]
Scalable IOV implementation.
As a first step, lets assume that all of the code in
ice_virtchnl_pf.[ch] is for Single Root IOV. Rename this file to
ice_sriov.c and its header to ice_sriov.h
Future changes will further split out the code in these files following
the plan outlined here.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The ice_sriov.c file primarily contains code which handles the logic for
mailbox overflow detection and some other utility functions related to
the virtualization mailbox.
The bulk of the SR-IOV implementation is actually found in
ice_virtchnl_pf.c, and this file isn't strictly SR-IOV specific.
In the future, the ice driver will support an additional virtualization
scheme known as Scalable IOV, and the code in this file will be used
for this alternative implementation.
Rename this file (and its associated header) to ice_vf_mbx.c, so that we
can later re-use the ice_sriov.c file as the SR-IOV specific file.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Add support for creating filters for GTP-U and GTP-C in switchdev mode. Add
support for parsing GTP-specific options (QFI and PDU type) and TEID.
By default, a filter for GTP-U will be added. To add a filter for GTP-C,
specify enc_dst_port = 2123, e.g.:
tc filter add dev $GTP0 ingress prio 1 flower enc_key_id 1337 \
enc_dst_port 2123 action mirred egress redirect dev $VF1_PR
Note: GTP-U with outer IPv6 offload is not supported yet.
Note: GTP-U with no payload offload is not supported yet.
Signed-off-by: Marcin Szycik <marcin.szycik@linux.intel.com>
Reviewed-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com>
Tested-by: Sandeep Penigalapati <sandeep.penigalapati@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Checking only protocol ids while searching for correct FVs can lead to a
situation, when incorrect FV will be added to the list. Incorrect means
that FV has correct protocol id but incorrect offset.
Call ice_get_sw_fv_list with ice_prot_lkup_ext struct which contains all
protocol ids with offsets.
With this modification allocating and collecting protocol ids list is
not longer needed.
Signed-off-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com>
Tested-by: Sandeep Penigalapati <sandeep.penigalapati@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Tony Nguyen says:
====================
100GbE Intel Wired LAN Driver Updates 2022-03-09
This series contains updates to ice driver only.
Martyna implements switchdev filtering on inner EtherType field for
tunnels.
Marcin adds reporting of slowpath statistics for port representors.
Jonathan Toppins changes a non-fatal link error message from warning to
debug.
Maciej removes unnecessary checks in ice_clean_tx_irq().
Amritha adds support for ADQ to match outer destination MAC for tunnels.
* '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/next-queue:
ice: Add support for outer dest MAC for ADQ tunnels
ice: avoid XDP checks in ice_clean_tx_irq()
ice: change "can't set link" message to dbg level
ice: Add slow path offload stats on port representor in switchdev
ice: Add support for inner etype in switchdev
====================
Link: https://lore.kernel.org/r/20220309190315.1380414-1-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Commit 5dbbbd01cb ("ice: Avoid RTNL lock when re-creating
auxiliary device") changes a process of re-creation of aux device
so ice_plug_aux_dev() is called from ice_service_task() context.
This unfortunately opens a race window that can result in dead-lock
when interface has left LAG and immediately enters LAG again.
Reproducer:
```
#!/bin/sh
ip link add lag0 type bond mode 1 miimon 100
ip link set lag0
for n in {1..10}; do
echo Cycle: $n
ip link set ens7f0 master lag0
sleep 1
ip link set ens7f0 nomaster
done
```
This results in:
[20976.208697] Workqueue: ice ice_service_task [ice]
[20976.213422] Call Trace:
[20976.215871] __schedule+0x2d1/0x830
[20976.219364] schedule+0x35/0xa0
[20976.222510] schedule_preempt_disabled+0xa/0x10
[20976.227043] __mutex_lock.isra.7+0x310/0x420
[20976.235071] enum_all_gids_of_dev_cb+0x1c/0x100 [ib_core]
[20976.251215] ib_enum_roce_netdev+0xa4/0xe0 [ib_core]
[20976.256192] ib_cache_setup_one+0x33/0xa0 [ib_core]
[20976.261079] ib_register_device+0x40d/0x580 [ib_core]
[20976.266139] irdma_ib_register_device+0x129/0x250 [irdma]
[20976.281409] irdma_probe+0x2c1/0x360 [irdma]
[20976.285691] auxiliary_bus_probe+0x45/0x70
[20976.289790] really_probe+0x1f2/0x480
[20976.298509] driver_probe_device+0x49/0xc0
[20976.302609] bus_for_each_drv+0x79/0xc0
[20976.306448] __device_attach+0xdc/0x160
[20976.310286] bus_probe_device+0x9d/0xb0
[20976.314128] device_add+0x43c/0x890
[20976.321287] __auxiliary_device_add+0x43/0x60
[20976.325644] ice_plug_aux_dev+0xb2/0x100 [ice]
[20976.330109] ice_service_task+0xd0c/0xed0 [ice]
[20976.342591] process_one_work+0x1a7/0x360
[20976.350536] worker_thread+0x30/0x390
[20976.358128] kthread+0x10a/0x120
[20976.365547] ret_from_fork+0x1f/0x40
...
[20976.438030] task:ip state:D stack: 0 pid:213658 ppid:213627 flags:0x00004084
[20976.446469] Call Trace:
[20976.448921] __schedule+0x2d1/0x830
[20976.452414] schedule+0x35/0xa0
[20976.455559] schedule_preempt_disabled+0xa/0x10
[20976.460090] __mutex_lock.isra.7+0x310/0x420
[20976.464364] device_del+0x36/0x3c0
[20976.467772] ice_unplug_aux_dev+0x1a/0x40 [ice]
[20976.472313] ice_lag_event_handler+0x2a2/0x520 [ice]
[20976.477288] notifier_call_chain+0x47/0x70
[20976.481386] __netdev_upper_dev_link+0x18b/0x280
[20976.489845] bond_enslave+0xe05/0x1790 [bonding]
[20976.494475] do_setlink+0x336/0xf50
[20976.502517] __rtnl_newlink+0x529/0x8b0
[20976.543441] rtnl_newlink+0x43/0x60
[20976.546934] rtnetlink_rcv_msg+0x2b1/0x360
[20976.559238] netlink_rcv_skb+0x4c/0x120
[20976.563079] netlink_unicast+0x196/0x230
[20976.567005] netlink_sendmsg+0x204/0x3d0
[20976.570930] sock_sendmsg+0x4c/0x50
[20976.574423] ____sys_sendmsg+0x1eb/0x250
[20976.586807] ___sys_sendmsg+0x7c/0xc0
[20976.606353] __sys_sendmsg+0x57/0xa0
[20976.609930] do_syscall_64+0x5b/0x1a0
[20976.613598] entry_SYSCALL_64_after_hwframe+0x65/0xca
1. Command 'ip link ... set nomaster' causes that ice_plug_aux_dev()
is called from ice_service_task() context, aux device is created
and associated device->lock is taken.
2. Command 'ip link ... set master...' calls ice's notifier under
RTNL lock and that notifier calls ice_unplug_aux_dev(). That
function tries to take aux device->lock but this is already taken
by ice_plug_aux_dev() in step 1
3. Later ice_plug_aux_dev() tries to take RTNL lock but this is already
taken in step 2
4. Dead-lock
The patch fixes this issue by following changes:
- Bit ICE_FLAG_PLUG_AUX_DEV is kept to be set during ice_plug_aux_dev()
call in ice_service_task()
- The bit is checked in ice_clear_rdma_cap() and only if it is not set
then ice_unplug_aux_dev() is called. If it is set (in other words
plugging of aux device was requested and ice_plug_aux_dev() is
potentially running) then the function only clears the bit
- Once ice_plug_aux_dev() call (in ice_service_task) is finished
the bit ICE_FLAG_PLUG_AUX_DEV is cleared but it is also checked
whether it was already cleared by ice_clear_rdma_cap(). If so then
aux device is unplugged.
Signed-off-by: Ivan Vecera <ivecera@redhat.com>
Co-developed-by: Petr Oros <poros@redhat.com>
Signed-off-by: Petr Oros <poros@redhat.com>
Reviewed-by: Dave Ertman <david.m.ertman@intel.com>
Link: https://lore.kernel.org/r/20220310171641.3863659-1-ivecera@redhat.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
TC flower does not support matching on user specified
outer MAC address for tunnels. For ADQ tunnels, the driver
adds outer destination MAC address as lower netdev's
active unicast MAC address to filter out packets with
unrelated MAC address being delivered to ADQ VSIs.
Example:
- create tunnel device
ip l add $VXLAN_DEV type vxlan id $VXLAN_VNI dstport $VXLAN_PORT \
dev $PF
- add TC filter (in ADQ mode)
$tc filter add dev $VXLAN_DEV protocol ip parent ffff: flower \
dst_ip $INNER_DST_IP ip_proto tcp dst_port $INNER_DST_PORT \
enc_key_id $VXLAN_VNI hw_tc $ADQ_TC
Note: Filters with wild-card tunnel ID (when user does not
specify tunnel key) are also supported.
Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com>
Tested-by: Bharathi Sreenivas <bharathi.sreenivas@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Commit 9610bd988d ("ice: optimize XDP_TX workloads") introduced Tx IRQ
cleaning routine dedicated for XDP rings. Currently it is impossible to
call ice_clean_tx_irq() against XDP ring, so it is safe to drop
ice_ring_is_xdp() calls in there.
Reviewed-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Tested-by: Kiran Bhandare <kiranx.bhandare@intel.com> (A Contingent Worker at Intel)
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
In the case where the link is owned by manageability, the firmware is
not allowed to set the link state, so an error code is returned.
This however is non-fatal and there is nothing the operator can do,
so instead of confusing the operator with messages they can do nothing
about hide this message behind the debug log level.
Signed-off-by: Jonathan Toppins <jtoppins@redhat.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com> (A Contingent worker at Intel)
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Implement callbacks to check for stats and fetch port representor stats.
Stats are taken from RX/TX ring corresponding to port representor and show
the number of bytes/packets that were not offloaded.
To see slow path stats run:
ifstat -x cpu_hits -a
Signed-off-by: Marcin Szycik <marcin.szycik@linux.intel.com>
Tested-by: Sandeep Penigalapati <sandeep.penigalapati@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Enable support for adding TC rules that filter on the inner
EtherType field of tunneled packet headers.
Signed-off-by: Martyna Szapar-Mudlaw <martyna.szapar-mudlaw@intel.com>
Reviewed-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Tested-by: Sandeep Penigalapati <sandeep.penigalapati@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Change curr_link_speed advertised speed, due to
link_info.link_speed is not equal phy.curr_user_speed_req.
Without this patch it is impossible to set advertised
speed to same as link_speed.
Testing Hints: Try to set advertised speed
to 25G only with 25G default link (use ethtool -s 0x80000000)
Fixes: 48cb27f2fd ("ice: Implement handlers for ethtool PHY/link operations")
Signed-off-by: Grzegorz Siwik <grzegorz.siwik@intel.com>
Signed-off-by: Jedrzej Jagielski <jedrzej.jagielski@intel.com>
Tested-by: Gurucharan <gurucharanx.g@intel.com> (A Contingent worker at Intel)
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
ice_misc_intr() is an irq handler. It should not sleep.
Use GFP_ATOMIC instead of GFP_KERNEL when allocating some memory.
Fixes: 348048e724 ("ice: Implement iidc operations")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Tested-by: Leszek Kaliszczuk <leszek.kaliszczuk@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
When a bonded interface is destroyed, .ndo_change_mtu can be called
during the tear-down process while the RTNL lock is held. This is a
problem since the auxiliary driver linked to the LAN driver needs to be
notified of the MTU change, and this requires grabbing a device_lock on
the auxiliary_device's dev. Currently this is being attempted in the
same execution context as the call to .ndo_change_mtu which is causing a
dead-lock.
Move the notification of the changed MTU to a separate execution context
(watchdog service task) and eliminate the "before" notification.
Fixes: 348048e724 ("ice: Implement iidc operations")
Signed-off-by: Dave Ertman <david.m.ertman@intel.com>
Tested-by: Jonathan Toppins <jtoppins@redhat.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com> (A Contingent worker at Intel)
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The ice_vc_send_msg_to_vf function has logic to detect "failure"
responses being sent to a VF. If a VF is sent more than
ICE_DFLT_NUM_INVAL_MSGS_ALLOWED then the VF is marked as disabled.
Almost identical logic also existed in the i40e driver.
This logic was added to the ice driver in commit 1071a8358a ("ice:
Implement virtchnl commands for AVF support") which itself copied from
the i40e implementation in commit 5c3c48ac6b ("i40e: implement virtual
device interface").
Neither commit provides a proper explanation or justification of the
check. In fact, later commits to i40e changed the logic to allow
bypassing the check in some specific instances.
The "logic" for this seems to be that error responses somehow indicate a
malicious VF. This is not really true. The PF might be sending an error
for any number of reasons such as lack of resources, etc.
Additionally, this causes the PF to log an info message for every failed
VF response which may confuse users, and can spam the kernel log.
This behavior is not documented as part of any requirement for our
products and other operating system drivers such as the FreeBSD
implementation of our drivers do not include this type of check.
In fact, the change from dev_err to dev_info in i40e commit 18b7af57d9
("i40e: Lower some message levels") explains that these messages
typically don't actually indicate a real issue. It is quite likely that
a user who hits this in practice will be very confused as the VF will be
disabled without an obvious way to recover.
We already have robust malicious driver detection logic using actual
hardware detection mechanisms that detect and prevent invalid device
usage. Remove the logic since its not a documented requirement and the
behavior is not intuitive.
Fixes: 1071a8358a ("ice: Implement virtchnl commands for AVF support")
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Pragma unroll was introduced around GCC 8, whereas current xsk code in
ice that prepares loop_unrolled_for macro that is based on mentioned
pragma, compares GCC version against 4, which is wrong and Stephen
found this out by compiling kernel with GCC 5.4 [0].
Fix this mistake and check if GCC version is >= 8.
[0]: https://lore.kernel.org/netdev/20220307213659.47658125@canb.auug.org.au/
Fixes: 126cdfe100 ("ice: xsk: Improve AF_XDP ZC Tx and use batching API")
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Link: https://lore.kernel.org/r/20220307231353.56638-1-maciej.fijalkowski@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The ice driver stores VF structures in a simple array which is allocated
once at the time of VF creation. The VF structures are then accessed
from the array by their VF ID. The ID must be between 0 and the number
of allocated VFs.
Multiple threads can access this table:
* .ndo operations such as .ndo_get_vf_cfg or .ndo_set_vf_trust
* interrupts, such as due to messages from the VF using the virtchnl
communication
* processing such as device reset
* commands to add or remove VFs
The current implementation does not keep track of when all threads are
done operating on a VF and can potentially result in use-after-free
issues caused by one thread accessing a VF structure after it has been
released when removing VFs. Some of these are prevented with various
state flags and checks.
In addition, this structure is quite static and does not support a
planned future where virtualization can be more dynamic. As we begin to
look at supporting Scalable IOV with the ice driver (as opposed to just
supporting Single Root IOV), this structure is not sufficient.
In the future, VFs will be able to be added and removed individually and
dynamically.
To allow for this, and to better protect against a whole class of
use-after-free bugs, replace the VF storage with a combination of a hash
table and krefs to reference track all of the accesses to VFs through
the hash table.
A hash table still allows efficient look up of the VF given its ID, but
also allows adding and removing VFs. It does not require contiguous VF
IDs.
The use of krefs allows the cleanup of the VF memory to be delayed until
after all threads have released their reference (by calling ice_put_vf).
To prevent corruption of the hash table, a combination of RCU and the
mutex table_lock are used. Addition and removal from the hash table use
the RCU-aware hash macros. This allows simple read-only look ups that
iterate to locate a single VF can be fast using RCU. Accesses which
modify the hash table, or which can't take RCU because they sleep, will
hold the mutex lock.
By using this design, we have a stronger guarantee that the VF structure
can't be released until after all threads are finished operating on it.
We also pave the way for the more dynamic Scalable IOV implementation in
the future.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Before we switch the VF data structure storage mechanism to a hash,
introduce new accessor functions to define the new interface.
* ice_get_vf_by_id is a function used to obtain a reference to a VF from
the table based on its VF ID
* ice_has_vfs is used to quickly check if any VFs are configured
* ice_get_num_vfs is used to get an exact count of how many VFs are
configured
We can drop the old ice_validate_vf_id function, since every caller was
just going to immediately access the VF table to get a reference
anyways. This way we simply use the single ice_get_vf_by_id to both
validate the VF ID is within range and that there exists a VF with that
ID.
This change enables us to more easily convert the codebase to the hash
table since most callers now properly use the interface.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
We maintain a number of values for VFs within the ice_pf structure. This
includes the VF table, the number of allocated VFs, the maximum number
of supported SR-IOV VFs, the number of queue pairs per VF, the number of
MSI-X vectors per VF, and a bitmap of the VFs with detected MDD events.
We're about to add a few more variables to this list. Clean this up
first by extracting these members out into a new ice_vfs structure
defined in ice_virtchnl_pf.h
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The ice_for_each_vf macro is intended to be used to loop over all VFs.
The current implementation relies on an iterator that is the index into
the VF array in the PF structure. This forces all users to perform a
look up themselves.
This abstraction forces a lot of duplicate work on callers and leaks the
interface implementation to the caller. Replace this with an
implementation that includes the VF pointer the primary iterator. This
version simplifies callers which just want to iterate over every VF, as
they no longer need to perform their own lookup.
The "i" iterator value is replaced with a new unsigned int "bkt"
parameter, as this will match the necessary interface for replacing
the VF array with a hash table. For now, the bkt is the VF ID, but in
the future it will simply be the hash bucket index. Document that it
should not be treated as a VF ID.
This change aims to simplify switching from the array to a hash table. I
considered alternative implementations such as an xarray but decided
that the hash table was the simplest and most suitable implementation. I
also looked at methods to hide the bkt iterator entirely, but I couldn't
come up with a feasible solution that worked for hash table iterators.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
When removing VFs, the driver takes a weird approach of assigning
pf->num_alloc_vfs to 0 before iterating over the VFs using a temporary
variable.
This logic has been in the driver for a long time, and seems to have
been carried forward from i40e.
We want to refactor the way VFs are stored, and iterating over the data
structure without the ice_for_each_vf interface impedes this work.
The logic relies on implicitly using the num_alloc_vfs as a sort of
"safe guard" for accessing VF data.
While this sort of guard makes sense for Single Root IOV where all VFs
are added at once, the data structures don't work for VFs which can be
added and removed dynamically. We also have a separate state flag,
ICE_VF_DEINIT_IN_PROGRESS which is a stronger protection against
concurrent removal and access.
Avoid the custom tmp iteration and replace it with the standard
ice_for_each_vf iterator. Delay the assignment of num_alloc_vfs until
after this loop finishes.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The ice_vc_send_msg_to_vf function is used by the PF to send a response
to a VF. This function has overzealous checks to ensure its not passed a
NULL VF pointer and to ensure that the passed in struct ice_vf has a
valid vf_id sub-member.
These checks have existed since commit 1071a8358a ("ice: Implement
virtchnl commands for AVF support") and function as simple sanity
checks.
We are planning to refactor the ice driver to use a hash table along
with appropriate locks in a future refactor. This change will modify how
the ice_validate_vf_id function works. Instead of a simple >= check to
ensure the VF ID is between some range, it will check the hash table to
see if the specified VF ID is actually in the table. This requires that
the function properly lock the table to prevent race conditions.
The checks may seem ok at first glance, but they don't really provide
much benefit.
In order for ice_vc_send_msg_to_vf to have these checks fail, the
callers must either (1) pass NULL as the VF, (2) construct an invalid VF
pointer manually, or (3) be using a VF pointer which becomes invalid
after they obtain it properly using ice_get_vf_by_id.
For (1), a cursory glance over callers of ice_vc_send_msg_to_vf can show
that in most cases the functions already operate assuming their VF
pointer is valid, such as by derferencing vf->pf or other members.
They obtain the VF pointer by accessing the VF array using the VF ID,
which can never produce a NULL value (since its a simple address
operation on the array it will not be NULL.
The sole exception for (1) is that ice_vc_process_vf_msg will forward a
NULL VF pointer to this function as part of its goto error handler
logic. This requires some minor cleanup to simply exit immediately when
an invalid VF ID is detected (Rather than use the same error flow as
the rest of the function).
For (2), it is unexpected for a flow to construct a VF pointer manually
instead of accessing the VF array. Defending against this is likely to
just hide bad programming.
For (3), it is definitely true that VF pointers could become invalid,
for example if a thread is processing a VF message while the VF gets
removed. However, the correct solution is not to add additional checks
like this which do not guarantee to prevent the race. Instead we plan to
solve the root of the problem by preventing the possibility entirely.
This solution will require the change to a hash table with proper
locking and reference counts of the VF structures. When this is done,
ice_validate_vf_id will require locking of the hash table. This will be
problematic because all of the callers of ice_vc_send_msg_to_vf will
already have to take the lock to obtain the VF pointer anyways. With a
mutex, this leads to a double lock that could hang the kernel thread.
Avoid this by removing the checks which don't provide much value, so
that we can safely add the necessary protections properly.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
After removing all VFs, the driver clears the VFLR indication for VFs.
This has been in ice since the beginning of SR-IOV support in the ice
driver.
The implementation was copied from i40e, and the motivation for the VFLR
indication clearing is described in the commit f7414531a0 ("i40e:
acknowledge VFLR when disabling SR-IOV")
The commit explains that we need to clear the VFLR indication because
the virtual function undergoes a VFLR event. If we don't indicate that
it is complete it can cause an issue when VFs are re-enabled due to
a "phantom" VFLR.
The register block read was added under a pci_vfs_assigned check
originally. This was done because we added the check after calling
pci_disable_sriov. This was later moved to disable SRIOV earlier in the
flow so that the VF drivers could be torn down before we removed
functionality.
Move the VFLR acknowledge into the main loop that tears down VF
resources. This avoids using the tmp value for iterating over VFs
multiple times. The result will make it easier to refactor the VF array
in a future change.
It's possible we might want to modify this flow to also stop checking
pci_vfs_assigned. However, it seems reasonable to keep this change: we
should only clear the VFLR if we actually disabled SR-IOV.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The ice_mbx_clear_malvf function is used to clear the indication and
count of how many times a VF was detected as malicious. During
ice_free_vfs, we use this function to ensure that all removed VFs are
reset to a clean state.
The call currently is done at the end of ice_free_vfs() using a tmp
value to iterate over all of the entries in the bitmap.
This separate iteration using tmp is problematic for a planned refactor
of the VF array data structure. To avoid this, lets move the call
slightly higher into the function inside the loop where we teardown all
of the VFs. This avoids one use of the tmp value used for iteration.
We'll fix the other user in a future change.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
We are planning to replace the simple array structure tracking VFs with
a hash table. This change will also remove the "num_alloc_vfs" variable.
Instead, new access functions to use the hash table as the source of
truth will be introduced. These will generally be equivalent to existing
checks, except during VF initialization.
Specifically, ice_set_per_vf_res() cannot use the hash table as it will
be operating prior to VF structures being inserted into the hash table.
Instead of using pf->num_alloc_vfs, simply pass the num_vfs value in
from the caller.
Note that a sub-function of ice_set_per_vf_res, ice_determine_res, also
implicitly depends on pf->num_alloc_vfs. Replace ice_determine_res with
a simpler inline implementation based on rounddown_pow_of_two. Note that
we must explicitly check that the argument is non-zero since it does not
play well with zero as a value.
Instead of using the function and while loop, simply calculate the
number of queues we have available by dividing by num_vfs. Check if the
desired queues are available. If not, round down to the nearest power of
2 that fits within our available queues.
This matches the behavior of ice_determine_res but is easier to follow
as simple in-line logic. Remove ice_determine_res entirely.
With this change, we no longer depend on the pf->num_alloc_vfs during
the initialization phase of VFs. This will allow us to safely remove it
in a future planned refactor of the VF data structures.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The VSI structure contains a vf_id field used to associate a VSI with a
VF. This is used mainly for ICE_VSI_VF as well as partially for
ICE_VSI_CTRL associated with the VFs.
This API was designed with the idea that VFs are stored in a simple
array that was expected to be static throughout most of the driver's
life.
We plan on refactoring VF storage in a few key ways:
1) converting from a simple static array to a hash table
2) using krefs to track VF references obtained from the hash table
3) use RCU to delay release of VF memory until after all references
are dropped
This is motivated by the goal to ensure that the lifetime of VF
structures is accounted for, and prevent various use-after-free bugs.
With the existing vsi->vf_id, the reference tracking for VFs would
become somewhat convoluted, because each VSI maintains a vf_id field
which will then require performing a look up. This means all these flows
will require reference tracking and proper usage of rcu_read_lock, etc.
We know that the VF VSI will always be backed by a valid VF structure,
because the VSI is created during VF initialization and removed before
the VF is destroyed. Rely on this and store a reference to the VF in the
VSI structure instead of storing a VF ID. This will simplify the usage
and avoid the need to perform lookups on the hash table in the future.
For ICE_VSI_VF, it is expected that vsi->vf is always non-NULL after
ice_vsi_alloc succeeds. Because of this, use WARN_ON when checking if a
vsi->vf pointer is valid when dealing with VF VSIs. This will aid in
debugging code which violates this assumption and avoid more disastrous
panics.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The code for supporting eswitch mode and port representors on VFs uses
an unwind based cleanup flow when handling errors.
These flows are used to cleanup and get everything back to the state
prior to attempting to switch from legacy to representor mode or back.
The unwind iterations make sense, but complicate a plan to refactor the
VF array structure. In the future we won't have a clean method of
reversing an iteration of the VFs.
Instead, we can change the cleanup flow to just iterate over all VF
structures and clean up appropriately.
First notice that ice_repr_add_for_all_vfs and ice_repr_rem_from_all_vfs
have an additional step of re-assigning the VC ops. There is no good
reason to do this outside of ice_repr_add and ice_repr_rem. It can
simply be done as the last step of these functions.
Second, make sure ice_repr_rem is safe to call on a VF which does not
have a representor. Check if vf->repr is NULL first and exit early if
so.
Move ice_repr_rem_from_all_vfs above ice_repr_add_for_all_vfs so that we
can call it from the cleanup function.
In ice_eswitch.c, replace the unwind iteration with a call to
ice_eswitch_release_reprs. This will go through all of the VFs and
revert the VF back to the standard model without the eswitch mode.
To make this safe, ensure this function checks whether or not the
represent or has been moved. Rely on the metadata destination in
vf->repr->dst. This must be NULL if the representor has not been moved
to eswitch mode.
Ensure that we always re-assign this value back to NULL after freeing
it, and move the ice_eswitch_release_reprs so that it can be called from
the setup function.
With these changes, eswitch cleanup no longer uses an unwind flow that
is problematic for the planned VF data structure change.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Sandeep Penigalapati <sandeep.penigalapati@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Add a new ice_gnss.c file for holding the basic GNSS module functions.
If the device supports GNSS module, call the new ice_gnss_init and
ice_gnss_release functions where appropriate.
Implement basic functionality for reading the data from GNSS module
using TTY device.
Add I2C read AQ command. It is now required for controlling the external
physical connectors via external I2C port expander on E810-T adapters.
Future changes will introduce write functionality.
Signed-off-by: Karol Kolacinski <karol.kolacinski@intel.com>
Signed-off-by: Sudhansu Sekhar Mishra <sudhansu.mishra@intel.com>
Tested-by: Sunitha Mekala <sunithax.d.mekala@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Clang static analysis reports this issues
ice_common.c:5008:21: warning: The left expression of the compound
assignment is an uninitialized value. The computed value will
also be garbage
ldo->phy_type_low |= ((u64)buf << (i * 16));
~~~~~~~~~~~~~~~~~ ^
When called from ice_cfg_phy_fec() ldo is the uninitialized local
variable tlv. So initialize.
Fixes: ea78ce4dab ("ice: add link lenient and default override support")
Signed-off-by: Tom Rix <trix@redhat.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com> (A Contingent worker at Intel)
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Clang static analysis reports this issue
time64.h:69:50: warning: The left operand of '+'
is a garbage value
set_normalized_timespec64(&ts_delta, lhs.tv_sec + rhs.tv_sec,
~~~~~~~~~~ ^
In ice_ptp_adjtime_nonatomic(), the timespec64 variable 'now'
is set by ice_ptp_gettimex64(). This function can fail
with -EBUSY, so 'now' can have a gargbage value.
So check the return.
Fixes: 06c16d89d2 ("ice: register 1588 PTP clock device object for E810 devices")
Signed-off-by: Tom Rix <trix@redhat.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com> (A Contingent worker at Intel)
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Commit c503e63200 ("ice: Stop processing VF messages during teardown")
introduced a driver state flag, ICE_VF_DEINIT_IN_PROGRESS, which is
intended to prevent some issues with concurrently handling messages from
VFs while tearing down the VFs.
This change was motivated by crashes caused while tearing down and
bringing up VFs in rapid succession.
It turns out that the fix actually introduces issues with the VF driver
caused because the PF no longer responds to any messages sent by the VF
during its .remove routine. This results in the VF potentially removing
its DMA memory before the PF has shut down the device queues.
Additionally, the fix doesn't actually resolve concurrency issues within
the ice driver. It is possible for a VF to initiate a reset just prior
to the ice driver removing VFs. This can result in the remove task
concurrently operating while the VF is being reset. This results in
similar memory corruption and panics purportedly fixed by that commit.
Fix this concurrency at its root by protecting both the reset and
removal flows using the existing VF cfg_lock. This ensures that we
cannot remove the VF while any outstanding critical tasks such as a
virtchnl message or a reset are occurring.
This locking change also fixes the root cause originally fixed by commit
c503e63200 ("ice: Stop processing VF messages during teardown"), so we
can simply revert it.
Note that I kept these two changes together because simply reverting the
original commit alone would leave the driver vulnerable to worse race
conditions.
Fixes: c503e63200 ("ice: Stop processing VF messages during teardown")
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Accidentally filter flag for none encapsulated l4 port field is always
set. Even if user wants to add encapsulated l4 port field.
Remove this unnecessary flag setting.
Fixes: 9e300987d4 ("ice: VXLAN and Geneve TC support")
Signed-off-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com>
Tested-by: Sandeep Penigalapati <sandeep.penigalapati@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
In switchdev mode, slow-path rules need to match all protocols, in order
to correctly redirect unfiltered or missed packets to the uplink. To set
this up for the virtual function to uplink flow, the rule that redirects
packets to the control VSI must have the tunnel type set to
ICE_SW_TUN_AND_NON_TUN. As a result of that new tunnel type being set,
ice_get_compat_fv_bitmap will select ICE_PROF_ALL. At that point all
profiles would be selected for this rule, resulting in the desired
behavior. Without this change slow-path would not work with
tunnel protocols.
Fixes: 8b032a55c1 ("ice: low level support for tunnels")
Signed-off-by: Wojciech Drewek <wojciech.drewek@intel.com>
Tested-by: Sandeep Penigalapati <sandeep.penigalapati@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The status of support for RDMA is currently being tracked with two
separate status flags. This is unnecessary with the current state of
the driver.
Simplify status tracking down to a single flag.
Rename the helper function to denote the RDMA specific status and
universally use the helper function to test the status bit.
Signed-off-by: Dave Ertman <david.m.ertman@intel.com>
Tested-by: Leszek Kaliszczuk <leszek.kaliszczuk@intel.com>
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The COMMS package can enable the hardware parser to recognize IPSEC
frames with ESP header and SPI identifier. If this package is available
and configured for loading in /lib/firmware, then the driver will
succeed in enabling this protocol type for RSS.
This in turn allows the hardware to hash over the SPI and use it to pick
a consistent receive queue for the same secure flow. Without this all
traffic is steered to the same queue for multiple traffic threads from
the same IP address. For that reason this is marked as a fix, as the
driver supports the model, but it wasn't enabled.
If the package is not available, adding this type will fail, but the
failure is ignored on purpose as it has no negative affect.
Fixes: c90ed40cef ("ice: Enable writing hardware filtering tables")
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com> (A Contingent worker at Intel)
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
If a call to re-create the auxiliary device happens in a context that has
already taken the RTNL lock, then the call flow that recreates auxiliary
device can hang if there is another attempt to claim the RTNL lock by the
auxiliary driver.
To avoid this, any call to re-create auxiliary devices that comes from
an source that is holding the RTNL lock (e.g. netdev notifier when
interface exits a bond) should execute in a separate thread. To
accomplish this, add a flag to the PF that will be evaluated in the
service task and dealt with there.
Fixes: f9f5301e7e ("ice: Register auxiliary device to provide RDMA")
Signed-off-by: Dave Ertman <david.m.ertman@intel.com>
Reviewed-by: Jonathan Toppins <jtoppins@redhat.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Currently, the same handler is called for both a NETDEV_BONDING_INFO
LAG unlink notification as for a NETDEV_UNREGISTER call. This is
causing a problem though, since the netdev_notifier_info passed has
a different structure depending on which event is passed. The problem
manifests as a call trace from a BUG: KASAN stack-out-of-bounds error.
Fix this by creating a handler specific to NETDEV_UNREGISTER that only
is passed valid elements in the netdev_notifier_info struct for the
NETDEV_UNREGISTER event.
Also included is the removal of an unbalanced dev_put on the peer_netdev
and related braces.
Fixes: 6a8b357278 ("ice: Respond to a NETDEV_UNREGISTER event for LAG")
Signed-off-by: Dave Ertman <david.m.ertman@intel.com>
Acked-by: Jonathan Toppins <jtoppins@redhat.com>
Tested-by: Sunitha Mekala <sunithax.d.mekala@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The driver was avoiding offload for IPIP (at least) frames due to
parsing the inner header offsets incorrectly when trying to check
lengths.
This length check works for VXLAN frames but fails on IPIP frames
because skb_transport_offset points to the inner header in IPIP
frames, which meant the subtraction of transport_header from
inner_network_header returns a negative value (-20).
With the code before this patch, everything continued to work, but GSO
was being used to segment, causing throughputs of 1.5Gb/s per thread.
After this patch, throughput is more like 10Gb/s per thread for IPIP
traffic.
Fixes: e94d447866 ("ice: Implement filter sync, NDO operations and bump version")
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Reviewed-by: Paul Menzel <pmenzel@molgen.mpg.de>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Propagate the error code from ice_get_link_default_override() instead
of returning success.
Fixes: ea78ce4dab ("ice: add link lenient and default override support")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Tony Nguyen says:
====================
100GbE Intel Wired LAN Driver Updates 2022-02-09
This series contains updates to ice driver only.
Brett adds support for QinQ. This begins with code refactoring and
re-organization of VLAN configuration functions to allow for
introduction of VSI VLAN ops to enable setting and calling of
respective operations based on device support of single or double
VLANs. Implementations are added for outer VLAN support.
To support QinQ, the device must be set to double VLAN mode (DVM).
In order for this to occur, the DDP package and NVM must also support
DVM. Functions to determine compatibility and properly configure the
device are added as well as setting the proper bits to advertise and
utilize the proper offloads. Support for VIRTCHNL_VF_OFFLOAD_VLAN_V2
is also included to allow for VF to negotiate and utilize this
functionality.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Daniel Borkmann says:
====================
pull-request: bpf-next 2022-02-09
We've added 126 non-merge commits during the last 16 day(s) which contain
a total of 201 files changed, 4049 insertions(+), 2215 deletions(-).
The main changes are:
1) Add custom BPF allocator for JITs that pack multiple programs into a huge
page to reduce iTLB pressure, from Song Liu.
2) Add __user tagging support in vmlinux BTF and utilize it from BPF
verifier when generating loads, from Yonghong Song.
3) Add per-socket fast path check guarding from cgroup/BPF overhead when
used by only some sockets, from Pavel Begunkov.
4) Continued libbpf deprecation work of APIs/features and removal of their
usage from samples, selftests, libbpf & bpftool, from Andrii Nakryiko
and various others.
5) Improve BPF instruction set documentation by adding byte swap
instructions and cleaning up load/store section, from Christoph Hellwig.
6) Switch BPF preload infra to light skeleton and remove libbpf dependency
from it, from Alexei Starovoitov.
7) Fix architecture-agnostic macros in libbpf for accessing syscall
arguments from BPF progs for non-x86 architectures,
from Ilya Leoshkevich.
8) Rework port members in struct bpf_sk_lookup and struct bpf_sock to be
of 16-bit field with anonymous zero padding, from Jakub Sitnicki.
9) Add new bpf_copy_from_user_task() helper to read memory from a different
task than current. Add ability to create sleepable BPF iterator progs,
from Kenny Yu.
10) Implement XSK batching for ice's zero-copy driver used by AF_XDP and
utilize TX batching API from XSK buffer pool, from Maciej Fijalkowski.
11) Generate temporary netns names for BPF selftests to avoid naming
collisions, from Hangbin Liu.
12) Implement bpf_core_types_are_compat() with limited recursion for
in-kernel usage, from Matteo Croce.
13) Simplify pahole version detection and finally enable CONFIG_DEBUG_INFO_DWARF5
to be selected with CONFIG_DEBUG_INFO_BTF, from Nathan Chancellor.
14) Misc minor fixes to libbpf and selftests from various folks.
* https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next: (126 commits)
selftests/bpf: Cover 4-byte load from remote_port in bpf_sk_lookup
bpf: Make remote_port field in struct bpf_sk_lookup 16-bit wide
libbpf: Fix compilation warning due to mismatched printf format
selftests/bpf: Test BPF_KPROBE_SYSCALL macro
libbpf: Add BPF_KPROBE_SYSCALL macro
libbpf: Fix accessing the first syscall argument on s390
libbpf: Fix accessing the first syscall argument on arm64
libbpf: Allow overriding PT_REGS_PARM1{_CORE}_SYSCALL
selftests/bpf: Skip test_bpf_syscall_macro's syscall_arg1 on arm64 and s390
libbpf: Fix accessing syscall arguments on riscv
libbpf: Fix riscv register names
libbpf: Fix accessing syscall arguments on powerpc
selftests/bpf: Use PT_REGS_SYSCALL_REGS in bpf_syscall_macro
libbpf: Add PT_REGS_SYSCALL_REGS macro
selftests/bpf: Fix an endianness issue in bpf_syscall_macro test
bpf: Fix bpf_prog_pack build HPAGE_PMD_SIZE
bpf: Fix leftover header->pages in sparc and powerpc code.
libbpf: Fix signedness bug in btf_dump_array_data()
selftests/bpf: Do not export subtest as standalone test
bpf, x86_64: Fail gracefully on bpf_jit_binary_pack_finalize failures
...
====================
Link: https://lore.kernel.org/r/20220209210050.8425-1-daniel@iogearbox.net
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
VFs by default are able to see all tagged traffic regardless of trust
and VLAN filters. Based on legacy devices (i.e. ixgbe, i40e), customers
expect VFs to receive all VLAN tagged traffic with a matching
destination MAC.
Add an ethtool private flag 'vf-vlan-pruning' and set the default to
off so VFs will receive all VLAN traffic directed towards them. When
the flag is turned on, VF will only be able to receive untagged
traffic or traffic with VLAN tags it has created interfaces for.
Also, the flag cannot be changed while any VFs are allocated. This was
done to simplify the implementation. So, if this flag is needed, then
the PF admin must enable it. If the user tries to enable the flag while
VFs are active, then print an unsupported message with the
vf-vlan-pruning flag included. In case multiple flags were specified, this
makes it clear to the user which flag failed.
Signed-off-by: Brett Creeley <brett.creeley@intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Currently there is only support for 802.1Q port VLANs on SR-IOV VFs. Add
support to also allow 802.1ad port VLANs when double VLAN mode is
enabled.
Signed-off-by: Brett Creeley <brett.creeley@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
In order for the driver to support 802.1ad VLAN filtering and offloads,
it needs to advertise those VLAN features and also support modifying
those VLAN features, so make the necessary changes to
ice_set_netdev_features(). By default, enable CTAG insertion/stripping
and CTAG filtering for both Single and Double VLAN Modes (SVM/DVM).
Also, in DVM, enable STAG filtering by default. This is done by
setting the feature bits in netdev->features. Also, in DVM, support
toggling of STAG insertion/stripping, but don't enable them by
default. This is done by setting the feature bits in
netdev->hw_features.
Since 802.1ad VLAN filtering and offloads are only supported in DVM, make
sure they are not enabled by default and that they cannot be enabled
during runtime, when the device is in SVM.
Add an implementation for the ndo_fix_features() callback. This is
needed since the hardware cannot support multiple VLAN ethertypes for
VLAN insertion/stripping simultaneously and all supported VLAN filtering
must either be enabled or disabled together.
Disable inner VLAN stripping by default when DVM is enabled. If a VSI
supports stripping the inner VLAN in DVM, then it will have to configure
that during runtime. For example if a VF is configured in a port VLAN
while DVM is enabled it will be allowed to offload inner VLANs.
Signed-off-by: Brett Creeley <brett.creeley@intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
In order to support configuring the device in Double VLAN Mode (DVM),
the DDP and FW have to support DVM. If both support DVM, the PF that
downloads the package needs to update the default recipes, set the
VLAN mode, and update boost TCAM entries.
To support updating the default recipes in DVM, add support for
updating an existing switch recipe's lkup_idx and mask. This is done
by first calling the get recipe AQ (0x0292) with the desired recipe
ID. Then, if that is successful update one of the lookup indices
(lkup_idx) and its associated mask if the mask is valid otherwise
the already existing mask will be used.
The VLAN mode of the device has to be configured while the global
configuration lock is held while downloading the DDP, specifically after
the DDP has been downloaded. If supported, the device will default to
DVM.
Co-developed-by: Dan Nowlin <dan.nowlin@intel.com>
Signed-off-by: Dan Nowlin <dan.nowlin@intel.com>
Signed-off-by: Brett Creeley <brett.creeley@intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Add support for the VF driver to be able to request
VIRTCHNL_VF_OFFLOAD_VLAN_V2, negotiate its VLAN capabilities via
VIRTCHNL_OP_GET_OFFLOAD_VLAN_V2_CAPS, add/delete VLAN filters, and
enable/disable VLAN offloads.
VFs supporting VIRTCHNL_OFFLOAD_VLAN_V2 will be able to use the
following virtchnl opcodes:
VIRTCHNL_OP_GET_OFFLOAD_VLAN_V2_CAPS
VIRTCHNL_OP_ADD_VLAN_V2
VIRTCHNL_OP_DEL_VLAN_V2
VIRTCHNL_OP_ENABLE_VLAN_STRIPPING_V2
VIRTCHNL_OP_DISABLE_VLAN_STRIPPING_V2
VIRTCHNL_OP_ENABLE_VLAN_INSERTION_V2
VIRTCHNL_OP_DISABLE_VLAN_INSERTION_V2
Legacy VF drivers may expect the initial VLAN stripping settings to be
configured by the PF, so the PF initializes VLAN stripping based on the
VIRTCHNL_OP_GET_VF_RESOURCES opcode. However, with VLAN support via
VIRTCHNL_VF_OFFLOAD_VLAN_V2, this function is only expected to be used
for VFs that only support VIRTCHNL_VF_OFFLOAD_VLAN, which will only
be supported when a port VLAN is configured. Update the function
based on the new expectations. Also, change the message when the PF
can't enable/disable VLAN stripping to a dev_dbg() as this isn't fatal.
When a VF isn't in a port VLAN and it only supports
VIRTCHNL_VF_OFFLOAD_VLAN when Double VLAN Mode (DVM) is enabled, then
the PF needs to reject the VIRTCHNL_VF_OFFLOAD_VLAN capability and
configure the VF in software only VLAN mode. To do this add the new
function ice_vf_vsi_cfg_legacy_vlan_mode(), which updates the VF's
inner and outer ice_vsi_vlan_ops functions and sets up software only
VLAN mode.
Signed-off-by: Brett Creeley <brett.creeley@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Currently the driver only supports 802.1Q VLAN insertion and stripping.
However, once Double VLAN Mode (DVM) is fully supported, then both 802.1Q
and 802.1ad VLAN insertion and stripping will be supported. Unfortunately
the VSI context parameters only allow for one VLAN ethertype at a time
for VLAN offloads so only one or the other VLAN ethertype offload can be
supported at once.
To support this, multiple changes are needed.
Rx path changes:
[1] In DVM, the Rx queue context l2tagsel field needs to be cleared so
the outermost tag shows up in the l2tag2_2nd field of the Rx flex
descriptor. In Single VLAN Mode (SVM), the l2tagsel field should remain
1 to support SVM configurations.
[2] Modify the ice_test_staterr() function to take a __le16 instead of
the ice_32b_rx_flex_desc union pointer so this function can be used for
both rx_desc->wb.status_error0 and rx_desc->wb.status_error1.
[3] Add the new inline function ice_get_vlan_tag_from_rx_desc() that
checks if there is a VLAN tag in l2tag1 or l2tag2_2nd.
[4] In ice_receive_skb(), add a check to see if NETIF_F_HW_VLAN_STAG_RX
is enabled in netdev->features. If it is, then this is the VLAN
ethertype that needs to be added to the stripping VLAN tag. Since
ice_fix_features() prevents CTAG_RX and STAG_RX from being enabled
simultaneously, the VLAN ethertype will only ever be 802.1Q or 802.1ad.
Tx path changes:
[1] In DVM, the VLAN tag needs to be placed in the l2tag2 field of the Tx
context descriptor. The new define ICE_TX_FLAGS_HW_OUTER_SINGLE_VLAN was
added to the list of tx_flags to handle this case.
[2] When the stack requests the VLAN tag to be offloaded on Tx, the
driver needs to set either ICE_TX_FLAGS_HW_OUTER_SINGLE_VLAN or
ICE_TX_FLAGS_HW_VLAN, so the tag is inserted in l2tag2 or l2tag1
respectively. To determine which location to use, set a bit in the Tx
ring flags field during ring allocation that can be used to determine
which field to use in the Tx descriptor. In DVM, always use l2tag2,
and in SVM, always use l2tag1.
Signed-off-by: Brett Creeley <brett.creeley@intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Add a new outer_vlan_ops member to the ice_vsi structure as outer VLAN
ops are only available when the device is in Double VLAN Mode (DVM).
Depending on the VSI type, the requirements for what operations to
use/allow differ.
By default all VSI's have unsupported inner and outer VSI VLAN ops. This
implementation was chosen to prevent unexpected crashes due to null
pointer dereferences. Instead, if a VSI calls an unsupported op, it will
just return -EOPNOTSUPP.
Add implementations to support modifying outer VLAN fields for VSI
context. This includes the ability to modify VLAN stripping, insertion,
and the port VLAN based on the outer VLAN handling fields of the VSI
context.
These functions should only ever be used if DVM is enabled because that
means the firmware supports the outer VLAN fields in the VSI context. If
the device is in DVM, then always use the outer_vlan_ops, else use the
vlan_ops since the device is in Single VLAN Mode (SVM).
Also, move adding the untagged VLAN 0 filter from ice_vsi_setup() to
ice_vsi_vlan_setup() as the latter function is specific to the PF and
all other VSI types that need an untagged VLAN 0 filter already do this
in their specific flows. Without this change, Flow Director is failing
to initialize because it does not implement any VSI VLAN ops.
Signed-off-by: Brett Creeley <brett.creeley@intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Current operations act on inner VLAN fields. To support double VLAN, outer
VLAN operations and functions will be implemented. Add the "inner" naming
to existing VLAN operations to distinguish them from the upcoming outer
values and functions. Some spacing adjustments are made to align
values.
Note that the inner is not talking about a tunneled VLAN, but the second
VLAN in the packet. For SVM the driver uses inner or single VLAN
filtering and offloads and in Double VLAN Mode the driver uses the
inner filtering and offloads for SR-IOV VFs in port VLANs in order to
support offloading the guest VLAN while a port VLAN is configured.
Signed-off-by: Brett Creeley <brett.creeley@intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Currently the proto argument is unused. This is because the driver only
supports 802.1Q VLAN filtering. This policy is enforced via netdev
features that the driver sets up when configuring the netdev, so the
proto argument won't ever be anything other than 802.1Q. However, this
will allow for future iterations of the driver to seemlessly support
802.1ad filtering. Begin using the proto argument and extend the related
structures to support its use.
Signed-off-by: Brett Creeley <brett.creeley@intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The current vf->port_vlan_info variable is a packed u16 that contains
the port VLAN ID and QoS/prio value. This is fine, but changes are
incoming that allow for an 802.1ad port VLAN. Add flexibility by
changing the vf->port_vlan_info member to be an ice_vlan structure.
Signed-off-by: Brett Creeley <brett.creeley@intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Add a new struct for VLAN related information. Currently this holds
VLAN ID and priority values, but will be expanded to hold TPID value.
This reduces the changes necessary if any other values are added in
future. Remove the action argument from these calls as it's always
ICE_FWD_VSI.
Signed-off-by: Brett Creeley <brett.creeley@intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Incoming changes to support 802.1Q and/or 802.1ad VLAN filtering and
offloads require more flexibility when configuring VLANs. The VSI VLAN
interface will allow flexibility for configuring VLANs for all VSI
types. Add new files to separate the VSI VLAN ops and move functions to
make the code more organized.
Signed-off-by: Brett Creeley <brett.creeley@intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
There are multiple places where VLAN 0 is being added. Create a function
to be called in order to minimize changes as the implementation is expanded
to support double VLAN and avoid duplicated code.
Signed-off-by: Brett Creeley <brett.creeley@intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Add functions to configure Tx VLAN antispoof based on iproute
configuration and/or VLAN mode and VF driver support. This is needed
later so the driver can control when it can be configured. Also, add
functions that can be used to enable and disable MAC and VLAN
spoofcheck. Move spoofchk configuration during VSI setup into the
SR-IOV initialization path and into the post VSI rebuild flow for VF
VSIs.
Signed-off-by: Brett Creeley <brett.creeley@intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The ice driver provides QoS information to auxiliary drivers
through the exported function ice_get_qos_params. This function
doesn't currently support L3 DSCP QoS.
Add the necessary defines, structure elements and code to support
DSCP QoS through the IIDC functions.
Signed-off-by: Dave Ertman <david.m.ertman@intel.com>
Signed-off-by: Shiraz Saleem <shiraz.saleem@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
For now, if the XDP prog returns XDP_PASS on XSK, the metadata will
be lost as it doesn't get copied to the skb.
Copy it along with the frame headers. Account its size on skb
allocation, and when copying just treat it as a part of the frame
and do a pull after to "move" it to the "reserved" zone.
net_prefetch() xdp->data_meta and align the copy size to speed-up
memcpy() a little and better match ice_construct_skb().
Fixes: 2d4238f556 ("ice: Add support for AF_XDP")
Suggested-by: Jesper Dangaard Brouer <brouer@redhat.com>
Suggested-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Reviewed-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com>
Tested-by: Kiran Bhandare <kiranx.bhandare@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
{__,}napi_alloc_skb() allocates and reserves additional NET_SKB_PAD
+ NET_IP_ALIGN for any skb.
OTOH, ice_construct_skb_zc() currently allocates and reserves
additional `xdp->data - xdp->data_hard_start`, which is
XDP_PACKET_HEADROOM for XSK frames.
There's no need for that at all as the frame is post-XDP and will
go only to the networking stack core.
Pass the size of the actual data only to __napi_alloc_skb() and
don't reserve anything. This will give enough headroom for stack
processing.
Fixes: 2d4238f556 ("ice: Add support for AF_XDP")
Signed-off-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Reviewed-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com>
Tested-by: Kiran Bhandare <kiranx.bhandare@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
In "legacy-rx" mode represented by ice_construct_skb(), we can
still use XDP (and XDP metadata), but after XDP_PASS the metadata
will be lost as it doesn't get copied to the skb.
Copy it along with the frame headers. Account its size on skb
allocation, and when copying just treat it as a part of the frame
and do a pull after to "move" it to the "reserved" zone.
Point net_prefetch() to xdp->data_meta instead of data. This won't
change anything when the meta is not here, but will save some cache
misses otherwise.
Suggested-by: Jesper Dangaard Brouer <brouer@redhat.com>
Suggested-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Reviewed-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com>
Tested-by: Kiran Bhandare <kiranx.bhandare@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
As stated in [1], dma_set_mask() with a 64-bit mask never fails if
dev->dma_mask is non-NULL.
So, if it fails, the 32 bits case will also fail for the same reason.
Simplify code and remove some dead code accordingly.
[1]: https://lkml.org/lkml/2021/6/7/398
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
One of the things that commit 5574ff7b7b ("i40e: optimize AF_XDP Tx
completion path") introduced was the @xdp_tx_active field. Its usage
from i40e can be adjusted to ice driver and give us positive performance
results.
If the descriptor that @next_dd points to has been sent by HW (its DD
bit is set), then we are sure that at least quarter of the ring is ready
to be cleaned. If @xdp_tx_active is 0 which means that related xdp_ring
is not used for XDP_{TX, REDIRECT} workloads, then we know how many XSK
entries should placed to completion queue, IOW walking through the ring
can be skipped.
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Acked-by: Magnus Karlsson <magnus.karlsson@intel.com>
Link: https://lore.kernel.org/bpf/20220125160446.78976-9-maciej.fijalkowski@intel.com
Apply the logic that was done for regular XDP from commit 9610bd988d
("ice: optimize XDP_TX workloads") to the ZC side of the driver. On top
of that, introduce batching to Tx that is inspired by i40e's
implementation with adjustments to the cleaning logic - take into the
account NAPI budget in ice_clean_xdp_irq_zc().
Separating the stats structs onto separate cache lines seemed to improve
the performance.
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Link: https://lore.kernel.org/bpf/20220125160446.78976-8-maciej.fijalkowski@intel.com
Commit 9610bd988d ("ice: optimize XDP_TX workloads") introduced
@next_dd and @next_rs to ice_tx_ring struct. Currently, their state is
not restored in ice_clean_tx_ring(), which was not causing any troubles
as the XDP rings are gone after we're done with XDP prog on interface.
For upcoming usage of mentioned fields in AF_XDP, this might expose us
to a potential dead Tx side. Scenario would look like following (based
on xdpsock):
- two xdpsock instances are spawned in Tx mode
- one of them is killed
- XDP prog is kept on interface due to the other xdpsock still running
* this means that XDP rings stayed in place
- xdpsock is launched again on same queue id that was terminated on
- @next_dd and @next_rs setting is bogus, therefore transmit side is
broken
To protect us from the above, restore the initial @next_rs and @next_dd
values when cleaning the Tx ring.
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Acked-by: Magnus Karlsson <magnus.karlsson@intel.com>
Link: https://lore.kernel.org/bpf/20220125160446.78976-7-maciej.fijalkowski@intel.com
XDP_TX workloads use a concept of Tx threshold that indicates the
interval of setting RS bit on descriptors which in turn tells the HW to
generate an interrupt to signal the completion of Tx on HW side. It is
currently based on a constant value of 32 which might not work out well
for various sizes of ring combined with for example batch size that can
be set via SO_BUSY_POLL_BUDGET.
Internal tests based on AF_XDP showed that most convenient setup of
mentioned threshold is when it is equal to quarter of a ring length.
Make use of recently introduced ICE_RING_QUARTER macro and use this
value as a substitute for ICE_TX_THRESH.
Align also ethtool -G callback so that next_dd/next_rs fields are up to
date in terms of the ring size.
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Acked-by: Magnus Karlsson <magnus.karlsson@intel.com>
Link: https://lore.kernel.org/bpf/20220125160446.78976-5-maciej.fijalkowski@intel.com
Currently, if ice_clean_rx_irq_zc() processed the whole ring and
next_to_use != 0, then ice_alloc_rx_buf_zc() would not refill the whole
ring even if the XSK buffer pool would have enough free entries (either
from fill ring or the internal recycle mechanism) - it is because ring
wrap is not handled.
Improve the logic in ice_alloc_rx_buf_zc() to address the problem above.
Do not clamp the count of buffers that is passed to
xsk_buff_alloc_batch() in case when next_to_use + buffer count >=
rx_ring->count, but rather split it and have two calls to the mentioned
function - one for the part up until the wrap and one for the part after
the wrap.
Signed-off-by: Magnus Karlsson <magnus.karlsson@intel.com>
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Link: https://lore.kernel.org/bpf/20220125160446.78976-4-maciej.fijalkowski@intel.com
With the upcoming introduction of batching to XSK data path,
performance wise it will be the best to have the ring descriptor count
to be aligned to power of 2.
Check if ring sizes that user is going to attach the XSK socket fulfill
the condition above. For Tx side, although check is being done against
the Tx queue and in the end the socket will be attached to the XDP
queue, it is fine since XDP queues get the ring->count setting from Tx
queues.
Suggested-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Acked-by: Magnus Karlsson <magnus.karlsson@intel.com>
Link: https://lore.kernel.org/bpf/20220125160446.78976-3-maciej.fijalkowski@intel.com
Remove the likely before napi_complete_done as this is the unlikely case
when busy-poll is used. Removing this has a positive performance impact
for busy-poll and no negative impact to the regular case.
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Acked-by: Magnus Karlsson <magnus.karlsson@intel.com>
Link: https://lore.kernel.org/bpf/20220125160446.78976-2-maciej.fijalkowski@intel.com
kfree() and bitmap_free() are the same. But using the latter is more
consistent when freeing memory allocated with bitmap_zalloc().
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
When a bitmap is local to a function, it is safe to use the non-atomic
__[set|clear]_bit(). No concurrent accesses can occur.
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The 'possible_idx' bitmap is set just after it is zeroed, so we can save
the first step.
The 'free_idx' bitmap is used only at the end of the function as the
result of a bitmap xor operation. So there is no need to explicitly
zero it before.
So, slightly simply the code and remove 2 useless 'bitmap_zero()' call
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Tested-by: Sandeep Penigalapati <sandeep.penigalapati@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
In current switchdev implementation, every VF PR is assigned to
individual ring on switchdev ctrl VSI. For slow-path traffic, there
is a mapping VF->ring done in software based on src_vsi value (by
calling ice_eswitch_get_target_netdev function).
With this change, HW solution is introduced which is more
efficient. For each VF, src MAC (VF's MAC) filter will be created,
which forwards packets to the corresponding switchdev ctrl VSI queue
based on src MAC address.
This filter has to be removed and then replayed in case of
resetting one VF. Keep information about this rule in repr->mac_rule,
thanks to that we know which rule has to be removed and replayed
for a given VF.
In case of CORE/GLOBAL all rules are removed
automatically. We have to take care of readding them. This is done
by ice_replay_vsi_adv_rule.
When driver leaves switchdev mode, remove all advanced rules
from switchdev ctrl VSI. This is done by ice_rem_adv_rule_for_vsi.
Flag repr->rule_added is needed because in some cases reset
might be triggered before VF sends request to add MAC.
Co-developed-by: Grzegorz Nitka <grzegorz.nitka@intel.com>
Signed-off-by: Grzegorz Nitka <grzegorz.nitka@intel.com>
Signed-off-by: Wojciech Drewek <wojciech.drewek@intel.com>
Tested-by: Sandeep Penigalapati <sandeep.penigalapati@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
ice_replay_vsi_adv_rule will replay advanced rules for a given VSI.
Exit this function when list of rules for given recipe is empty.
Do not add rule when given vsi_handle does not match vsi_handle
from the rule info.
Use ICE_MAX_NUM_RECIPES instead of ICE_SW_LKUP_LAST in order to find
advanced rules as well.
Signed-off-by: Victor Raj <victor.raj@intel.com>
Signed-off-by: Wojciech Drewek <wojciech.drewek@intel.com>
Tested-by: Sandeep Penigalapati <sandeep.penigalapati@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Recent bpf-next merge brought in header changes which uncovered
includes missing in net-next which were not present in bpf-next.
Build problems happen only on less-popular arches like hppa,
sparc, alpha etc.
I could repro the build problem with ice but not the mlx5 problem
Abdul was reporting. mlx5 does look like it should include filter.h,
anyway.
Reported-by: Abdul Haleem <abdhalee@linux.vnet.ibm.com>
Fixes: e63a023489 ("Merge git://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next")
Link: https://lore.kernel.org/all/7c03768d-d948-c935-a7ab-b1f963ac7eed@linux.vnet.ibm.com/
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Alexei Starovoitov says:
====================
pull-request: bpf-next 2021-12-30
The following pull-request contains BPF updates for your *net-next* tree.
We've added 72 non-merge commits during the last 20 day(s) which contain
a total of 223 files changed, 3510 insertions(+), 1591 deletions(-).
The main changes are:
1) Automatic setrlimit in libbpf when bpf is memcg's in the kernel, from Andrii.
2) Beautify and de-verbose verifier logs, from Christy.
3) Composable verifier types, from Hao.
4) bpf_strncmp helper, from Hou.
5) bpf.h header dependency cleanup, from Jakub.
6) get_func_[arg|ret|arg_cnt] helpers, from Jiri.
7) Sleepable local storage, from KP.
8) Extend kfunc with PTR_TO_CTX, PTR_TO_MEM argument support, from Kumar.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Add support to enable flow-director filter when multiple TCs are
configured. Flow director filter can be configured using ethtool
(--config-ntuple option). When multiple TCs are configured, each
TC is mapped to an unique HW VSI. So VSI corresponding to queue
used in filter is identified and flow director context is updated
with correct VSI while configuring ntuple filter in HW.
Signed-off-by: Kiran Patil <kiran.patil@intel.com>
Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com>
Signed-off-by: Sudheer Mogilappagari <sudheer.mogilappagari@intel.com>
Tested-by: Bharathi Sreenivas <bharathi.sreenivas@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
sock.h is pretty heavily used (5k objects rebuilt on x86 after
it's touched). We can drop the include of filter.h from it and
add a forward declaration of struct sk_filter instead.
This decreases the number of rebuilt objects when bpf.h
is touched from ~5k to ~1k.
There's a lot of missing includes this was masking. Primarily
in networking tho, this time.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Marc Kleine-Budde <mkl@pengutronix.de>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
Acked-by: Nikolay Aleksandrov <nikolay@nvidia.com>
Acked-by: Stefano Garzarella <sgarzare@redhat.com>
Link: https://lore.kernel.org/bpf/20211229004913.513372-1-kuba@kernel.org
napi_build_skb() reuses per-cpu NAPI skbuff_head cache in order
to save some cycles on freeing/allocating skbuff_heads on every
new Rx or completed Tx.
ice driver runs Tx completion polling cycle right before the Rx
one and uses napi_consume_skb() to feed the cache with skbuff_heads
of completed entries, so it's never empty and always warm at that
moment. Switch to the napi_build_skb() to relax mm pressure on
heavy Rx.
Signed-off-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Reviewed-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Fix an odd indent where some code was left indented, and causes smatch
to warn:
ice_log_pkg_init() warn: inconsistent indenting
While here, for consistency, add a break after the default case.
This commit has a Fixes: but we caught this while it was only in net-next.
Fixes: 247dd97d71 ("ice: Refactor status flow for DDP load")
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Link: https://lore.kernel.org/r/20211221230538.2546315-1-jesse.brandeburg@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
E822 devices on supported platforms can generate a cross timestamp
between the platform ART and the device time. This process allows for
very precise measurement of the difference between the PTP hardware
clock and the platform time.
This is only supported if we know the TSC frequency relative to ART, so
we do not enable this unless the boot CPU has a known TSC frequency (as
required by convert_art_ns_to_tsc).
Because PCIe PTM support is not available on all platforms, introduce
CONFIG_ICE_HWTS and make it depend on X86 where we know the support
exists.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Once the E822 device has sent and received one packet, the hardware
computes the internal delay of the PHY using a process known as Vernier
calibration. This calibration calculates a more accurate offset for the
Tx and Rx timestamps. To make use of this offset, we need to exit the
bypass mode. This cannot be done until the PHY has completed offset
calibration, as indicated by the offset valid bits.
To handle this, introduce a kthread work item which will poll the offset
valid bits every few milliseconds seeing if it is safe to exit bypass
mode.
Once we have finished calibrating the offsets, we can program the total
Tx and Rx offset registers and turn off the bypass bit. This allows the
hardware to include the more precise vernier calibration offset, and
improves the timestamp precision.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The E822 device has a Clock Generation Unit (CGU) responsible for
determining the clock frequency that drives the timers.
Ensure this function is initialized when bringing up the PTP support, so
that the clock has a known frequency.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Implement support for the basic operations needed to enable the PTP
hardware clock on E822 devices.
This includes implementations for the various PHY access functions, as
well as the ability to start and stop the PHY timers. This is different
from the E810 device because the configuration depends on link speed, so
we cannot just start the PHYs immediately. We must wait until the link
is up to get proper values for the speed based initialization.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Convert the clk_freq value into the associated time_ref frequency value
for E822 devices. This simplifies determining the time reference value
for the clock.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
When we enable support for E822 devices, there are some additional
steps required to initialize the PTP hardware clock. To make this easier
to implement as device-specific behavior, refactor the register setups
in ice_ptp_init_owner to a new ice_ptp_init_phc function defined in
ice_ptp_hw.c
This function will have a common section, and an e810 specific
sub-implementation.
This will enable easily extending the functionality to cover the E822
specific setup required to initialize the hardware clock generation
unit. It also makes it clear which steps are E810 specific vs which ones
are necessary for all ice devices.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Paul Menzel <pmenzel@molgen.mpg.de>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The ice_ptp_hw.c file introduced a bunch of uses of "int status" instead
of the more traditional "int err" or "int ret". These are actually
traditional Linux error codes (as opposed to the recently removed
ice_status enumeration values).
We're about to add a bunch of new functions to ice_ptp_hw.c. It's
normally preferred in the ice driver to use "int ret" or "int err" when
dealing with error code values.
Instead of making the new functions use "int status" lets just fix all
of ice_ptp_hw.c to use "int err". This will match the new functions and
ensures a consistent style across at least the PTP related files.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The tstamp_config structure is being set inside of
ice_ptp_cfg_timestamp, which is the function used to set Tx and
Rx timestamping during initialization.
This function is also used in order to set the PHY port timestamping
status. However, it makes sense to always set the tstamp_config directly
whenever the ice_set_tx_tstamp or ice_set_rx_tstamp functions are
called.
Move assignment of tstamp_config into the related functions and out of
ice_ptp_cfg_timestamp.
Now that we assign the timestamp mode in the relevant functions, we no
longer modify the config value in ice_set_timestamp_mode. In turn, we
no longer want to copy that config value into the PF cached structure.
Instead, this is now the source of truth for actual configuration. On
success of ice_set_timestamp_mode, copy the real configured mode back to
report it out to userspace.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
A future change will add additional possible increment values for the
E822 device support. To handle this, we want to look up the increment
value to use instead of hard coding it to the nominal value for E810
devices. Introduce ice_base_incval as a function to get the best nominal
increment value to use.
For now, it just returns the E810 value, but will be refactored in the
future to look up the value based on the device type and configured
clock frequency.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The PF reset does not reset PHC and PHY clocks so it's unnecessary to
stop them and reinitialize after the reset.
Configuring timestamping changes the VSI fields so it needs to be
performed after VSIs are initialized, which was not done in case of a
reset.
Suggested-by: Patrick Talbert <ptalbert@redhat.com>
Signed-off-by: Karol Kolacinski <karol.kolacinski@intel.com>
Tested-by: Pasi Vaananen <pvaanane@redhat.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Currently cleaned_count is initialized to ICE_DESC_UNUSED(rx_ring) and
later on during the Rx processing it is incremented per each frame that
driver consumed. This can result in excessive buffers requested from xsk
pool based on that value.
To address this, just drop cleaned_count and pass
ICE_DESC_UNUSED(rx_ring) directly as a function argument to
ice_alloc_rx_bufs_zc(). Idea is to ask for buffers as many as consumed.
Let us also call ice_alloc_rx_bufs_zc unconditionally at the end of
ice_clean_rx_irq_zc. This has been changed in that way for corresponding
ice_clean_rx_irq, but not here.
Fixes: 2d4238f556 ("ice: Add support for AF_XDP")
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Tested-by: Kiran Bhandare <kiranx.bhandare@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Commit ac6f733a7b ("ice: allow empty Rx descriptors") stated that ice
HW can produce empty descriptors that are valid and they should be
processed.
Add this support to xsk ZC path to avoid potential processing problems.
Fixes: 2d4238f556 ("ice: Add support for AF_XDP")
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Tested-by: Kiran Bhandare <kiranx.bhandare@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The descriptor that ntu is pointing at when we exit
ice_alloc_rx_bufs_zc() should not have its corresponding DD bit cleared
as descriptor is not allocated in there and it is not valid for HW
usage.
The allocation routine at the entry will fill the descriptor that ntu
points to after it was set to ntu + nb_buffs on previous call.
Even the spec says:
"The tail pointer should be set to one descriptor beyond the last empty
descriptor in host descriptor ring."
Therefore, step away from clearing the status_error0 on ntu + nb_buffs
descriptor.
Fixes: db804cfc21 ("ice: Use the xsk batched rx allocation interface")
Reported-by: Elza Mathew <elza.mathew@intel.com>
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Tested-by: Kiran Bhandare <kiranx.bhandare@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The 'if (ntu == rx_ring->count)' block in ice_alloc_rx_buffers_zc()
was previously residing in the loop, but after introducing the
batched interface it is used only to wrap-around the NTU descriptor,
thus no more need to assign 'xdp'.
Fixes: db804cfc21 ("ice: Use the xsk batched rx allocation interface")
Signed-off-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Acked-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Tested-by: Kiran Bhandare <kiranx.bhandare@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Currently, the zero-copy data path is reusing the memory region that was
initially allocated for an array of struct ice_rx_buf for its own
purposes. This is error prone as it is based on the ice_rx_buf struct
always being the same size or bigger than what the zero-copy path needs.
There can also be old values present in that array giving rise to errors
when the zero-copy path uses it.
Fix this by freeing the ice_rx_buf region and allocating a new array for
the zero-copy path that has the right length and is initialized to zero.
Fixes: 57f7f8b6bc ("ice: Use xdp_buf instead of rx_buf for xsk zero-copy")
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Tested-by: Kiran Bhandare <kiranx.bhandare@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Currently we only NULL the xdp_buff pointer in the internal SW ring but
we never give it back to the xsk buffer pool. This means that buffers
can be leaked out of the buff pool and never be used again.
Add missing xsk_buff_free() call to the routine that is supposed to
clean the entries that are left in the ring so that these buffers in the
umem can be used by other sockets.
Also, only go through the space that is actually left to be cleaned
instead of a whole ring.
Fixes: 2d4238f556 ("ice: Add support for AF_XDP")
Signed-off-by: Magnus Karlsson <magnus.karlsson@intel.com>
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Tested-by: Kiran Bhandare <kiranx.bhandare@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
David S. Miller
Alexander Lobakin
Maciej Fijalkowski
Anatolii Gerasymenko
Ivan Vecera
Magnus Karlsson
Jakub Kicinski
Jacob Keller
Yang Yingliang
Wojciech Drewek
Sudheer Mogilappagari
Marcin Szycik
Michal Swiatkowski
Amritha Nambiar
Jonathan Toppins
Martyna Szapar-Mudlaw
Jedrzej Jagielski
Christophe JAILLET
Dave Ertman
Karol Kolacinski
Tom Rix
Jesse Brandeburg
Dan Carpenter
Brett Creeley
Victor Raj
Kiran Patil