Commit bd7c00605e ("net: move aRFS rmap management and CPU affinity
to core") allows the drivers to delegate the IRQ affinity to the NAPI
instance. However, the driver needs to use a persistent NAPI config
and explicitly set/unset the NAPI<->IRQ association.
Convert to the new IRQ affinity API.
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Signed-off-by: Ahmed Zaki <ahmed.zaki@intel.com>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
i40e, ice, and iAVF all use 'hena' as a shorthand for the "hash enable"
configuration. This comes originally from the X710 datasheet 'xxQF_HENA'
registers. In the context of the registers the meaning is fairly clear.
However, on its own, hena is a weird name that can be more difficult to
understand. This is especially true in ice. The E810 hardware doesn't even
have registers with HENA in the name.
Replace the shorthand 'hena' with 'hashcfg'. This makes it clear the
variables deal with the Hash configuration, not just a single boolean
on/off for all hashing.
Do not update the register names. These come directly from the datasheet
for X710 and X722, and it is more important that the names can be searched.
Suggested-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Tested-by: Rinitha S <sx.rinitha@intel.com> (A Contingent worker at Intel)
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Get rid of the crit lock.
That frees us from the error prone logic of try_locks.
Thanks to netdev_lock() by Jakub it is now easy, and in most cases we were
protected by it already - replace crit lock by netdev lock when it was not
the case.
Lockdep reports that we should cancel the work under crit_lock [splat1],
and that was the scheme we have mostly followed since [1] by Slawomir.
But when that is done we still got into deadlocks [splat2]. So instead
we should look at the bigger problem, namely "weird locking/scheduling"
of the iavf. The first step to fix that is to remove the crit lock.
I will followup with a -next series that simplifies scheduling/tasks.
Cancel the work without netdev lock (weird unlock+lock scheme),
to fix the [splat2] (which would be totally ugly if we would kept
the crit lock).
Extend protected part of iavf_watchdog_task() to include scheduling
more work.
Note that the removed comment in iavf_reset_task() was misplaced,
it belonged to inside of the removed if condition, so it's gone now.
[splat1] - w/o this patch - The deadlock during VF removal:
WARNING: possible circular locking dependency detected
sh/3825 is trying to acquire lock:
((work_completion)(&(&adapter->watchdog_task)->work)){+.+.}-{0:0}, at: start_flush_work+0x1a1/0x470
but task is already holding lock:
(&adapter->crit_lock){+.+.}-{4:4}, at: iavf_remove+0xd1/0x690 [iavf]
which lock already depends on the new lock.
[splat2] - when cancelling work under crit lock, w/o this series,
see [2] for the band aid attempt
WARNING: possible circular locking dependency detected
sh/3550 is trying to acquire lock:
((wq_completion)iavf){+.+.}-{0:0}, at: touch_wq_lockdep_map+0x26/0x90
but task is already holding lock:
(&dev->lock){+.+.}-{4:4}, at: iavf_remove+0xa6/0x6e0 [iavf]
which lock already depends on the new lock.
[1] fc2e6b3b13 ("iavf: Rework mutexes for better synchronisation")
[2] https://github.com/pkitszel/linux/commit/52dddbfc2bb60294083f5711a158a
Fixes: d1639a1731 ("iavf: fix a deadlock caused by rtnl and driver's lock circular dependencies")
Signed-off-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The union iavf_32byte_rx_desc consists of two unnamed structs defined
inside. One of them represents legacy 32 byte descriptor and second the
16 byte descriptor (extended to 32 byte). Each of them consists of
bunch of unions, structs and __le fields that represent specific fields
in descriptor.
This commit changes the representation of iavf_32byte_rx_desc union
to store four __le64 fields (qw0, qw1, qw2, qw3) that represent
quad-words. Those quad-words will be then accessed by calling
leXY_get_bits macros in upcoming commits.
Suggested-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Signed-off-by: Mateusz Polchlopek <mateusz.polchlopek@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Add a new extended capabilities negotiation to exchange information from
the PF about what PTP capabilities are supported by this VF. This
requires sending a VIRTCHNL_OP_1588_PTP_GET_CAPS message, and waiting
for the response from the PF. Handle this early on during the VF
initialization.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Co-developed-by: Mateusz Polchlopek <mateusz.polchlopek@intel.com>
Signed-off-by: Mateusz Polchlopek <mateusz.polchlopek@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Enable support for VIRTCHNL_VF_OFFLOAD_RX_FLEX_DESC, to enable the VF
driver the ability to determine what Rx descriptor formats are
available. This requires sending an additional message during
initialization and reset, the VIRTCHNL_OP_GET_SUPPORTED_RXDIDS. This
operation requests the supported Rx descriptor IDs available from the
PF.
This is treated the same way that VLAN V2 capabilities are handled. Add
a new set of extended capability flags, used to process send and receipt
of the VIRTCHNL_OP_GET_SUPPORTED_RXDIDS message.
This ensures we finish negotiating for the supported descriptor formats
prior to beginning configuration of receive queues.
This change stores the supported format bitmap into the iavf_adapter
structure. Additionally, if VIRTCHNL_VF_OFFLOAD_RX_FLEX_DESC is enabled
by the PF, we need to make sure that the Rx queue configuration
specifies the format.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Co-developed-by: Mateusz Polchlopek <mateusz.polchlopek@intel.com>
Signed-off-by: Mateusz Polchlopek <mateusz.polchlopek@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
During driver initialization VF determines QOS capability is allowed
by PF and receives QOS parameters. After which quanta size for queues
is configured which is not configurable and is set to 1KB currently.
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: Sudheer Mogilappagari <sudheer.mogilappagari@intel.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Link: https://patch.msgid.link/72cbeb9c88d40e557053c57d7531c96bed490576.1728460186.git.pabeni@redhat.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Implement net_shaper_ops support for IAVF. This enables configuration
of rate limiting on per queue basis. Customer intends to enforce
bandwidth limit on Tx traffic steered to the queue by configuring
rate limits on the queue.
To set rate limiting for a queue, update shaper object of given queues
in driver and send VIRTCHNL_OP_CONFIG_QUEUE_BW to PF to update HW
configuration.
Deleting shaper configured for queue is nothing but configuring shaper
with bw_max 0. The PF restores the default rate limiting config
when bw_max is zero.
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: Sudheer Mogilappagari <sudheer.mogilappagari@intel.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Link: https://patch.msgid.link/5a882cb51998c4c2c3d21fed521498eba1c8f079.1728460186.git.pabeni@redhat.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
There is no caller and implementation in tree.
Signed-off-by: Yue Haibing <yuehaibing@huawei.com>
Reviewed-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Add support for offloading cls U32 filters. Only "skbedit queue_mapping"
and "drop" actions are supported. Also, only "ip" and "802_3" tc
protocols are allowed. The PF must advertise the VIRTCHNL_VF_OFFLOAD_TC_U32
capability flag.
Since the filters will be enabled via the FD stage at the PF, a new type
of FDIR filters is added and the existing list and state machine are used.
The new filters can be used to configure flow directors based on raw
(binary) pattern in the rx packet.
Examples:
0. # tc qdisc add dev enp175s0v0 ingress
1. Redirect UDP from src IP 192.168.2.1 to queue 12:
# tc filter add dev <dev> protocol ip ingress u32 \
match u32 0x45000000 0xff000000 at 0 \
match u32 0x00110000 0x00ff0000 at 8 \
match u32 0xC0A80201 0xffffffff at 12 \
match u32 0x00000000 0x00000000 at 24 \
action skbedit queue_mapping 12 skip_sw
2. Drop all ICMP:
# tc filter add dev <dev> protocol ip ingress u32 \
match u32 0x45000000 0xff000000 at 0 \
match u32 0x00010000 0x00ff0000 at 8 \
match u32 0x00000000 0x00000000 at 24 \
action drop skip_sw
3. Redirect ICMP traffic from MAC 3c:fd:fe:a5:47:e0 to queue 7
(note proto: 802_3):
# tc filter add dev <dev> protocol 802_3 ingress u32 \
match u32 0x00003CFD 0x0000ffff at 4 \
match u32 0xFEA547E0 0xffffffff at 8 \
match u32 0x08004500 0xffffff00 at 12 \
match u32 0x00000001 0x000000ff at 20 \
match u32 0x0000 0x0000 at 40 \
action skbedit queue_mapping 7 skip_sw
Notes on matches:
1 - All intermediate fields that are needed to parse the correct PTYPE
must be provided (in e.g. 3: Ethernet Type 0x0800 in MAC, IP version
and IP length: 0x45 and protocol: 0x01 (ICMP)).
2 - The last match must provide an offset that guarantees all required
headers are accounted for, even if the last header is not matched.
For example, in #2, the last match is 4 bytes at offset 24 starting
from IP header, so the total is 14 (MAC) + 24 + 4 = 42, which is the
sum of MAC+IP+ICMP headers.
Reviewed-by: Sridhar Samudrala <sridhar.samudrala@intel.com>
Reviewed-by: Marcin Szycik <marcin.szycik@linux.intel.com>
Signed-off-by: Ahmed Zaki <ahmed.zaki@intel.com>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
In preparation for a second type of FDIR filters that can be added by
tc-u32, move the add/del of the FDIR logic to be entirely contained in
iavf_fdir.c.
The iavf_find_fdir_fltr_by_loc() is renamed to iavf_find_fdir_fltr()
to be more agnostic to the filter ID parameter (for now @loc, which is
relevant only to current FDIR filters added via ethtool).
The FDIR filter deletion is moved from iavf_del_fdir_ethtool() in
ethtool.c to iavf_fdir_del_fltr(). While at it, fix a minor bug where
the "fltr" is accessed out of the fdir_fltr_lock spinlock protection.
Reviewed-by: Sridhar Samudrala <sridhar.samudrala@intel.com>
Reviewed-by: Marcin Szycik <marcin.szycik@linux.intel.com>
Signed-off-by: Ahmed Zaki <ahmed.zaki@intel.com>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Ever since build_skb() became stable, the old way with allocating an skb
for storing the headers separately, which will be then copied manually,
was slower, less flexible, and thus obsolete.
* It had higher pressure on MM since it actually allocates new pages,
which then get split and refcount-biased (NAPI page cache);
* It implies memcpy() of packet headers (40+ bytes per each frame);
* the actual header length was calculated via eth_get_headlen(), which
invokes Flow Dissector and thus wastes a bunch of CPU cycles;
* XDP makes it even more weird since it requires headroom for long and
also tailroom for some time (since mbuf landed). Take a look at the
ice driver, which is built around work-arounds to make XDP work with
it.
Even on some quite low-end hardware (not a common case for 100G NICs) it
was performing worse.
The only advantage "legacy-rx" had is that it didn't require any
reserved headroom and tailroom. But iavf didn't use this, as it always
splits pages into two halves of 2k, while that save would only be useful
when striding. And again, XDP effectively removes that sole pro.
There's a train of features to land in IAVF soon: Page Pool, XDP, XSk,
multi-buffer etc. Each new would require adding more and more Danse
Macabre for absolutely no reason, besides making hotpath less and less
effective.
Remove the "feature" with all the related code. This includes at least
one very hot branch (typically hit on each new frame), which was either
always-true or always-false at least for a complete NAPI bulk of 64
frames, the whole private flags cruft, and so on. Some stats:
Function: add/remove: 0/4 grow/shrink: 0/7 up/down: 0/-721 (-721)
RO Data: add/remove: 0/1 grow/shrink: 0/0 up/down: 0/-40 (-40)
Reviewed-by: Alexander Duyck <alexanderduyck@fb.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Allow the user to set the symmetric Toeplitz hash function via:
# ethtool -X eth0 hfunc toeplitz symmetric-xor
The driver will reject any new RSS configuration if a field other than
(IP src/dst and L4 src/dst ports) is requested for hashing.
The symmetric RSS will not be supported on PFs not advertising the ADV RSS
Offload flag (ADV_RSS_SUPPORT()), for example the E700 series (i40e).
Reviewed-by: Madhu Chittim <madhu.chittim@intel.com>
Signed-off-by: Ahmed Zaki <ahmed.zaki@intel.com>
Link: https://lore.kernel.org/r/20231213003321.605376-9-ahmed.zaki@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
New states introduced:
IAVF_FDIR_FLTR_DIS_REQUEST
IAVF_FDIR_FLTR_DIS_PENDING
IAVF_FDIR_FLTR_INACTIVE
Current FDIR state machines (SM) are not adequate to handle a few
scenarios in the link DOWN/UP event, reset event and ntuple-feature.
For example, when VF link goes DOWN and comes back UP administratively,
the expectation is that previously installed filters should also be
restored. But with current SM, filters are not restored.
So with new SM, during link DOWN filters are marked as INACTIVE in
the iavf list but removed from PF. After link UP, SM will transition
from INACTIVE to ADD_REQUEST to restore the filter.
Similarly, with VF reset, filters will be removed from the PF, but
marked as INACTIVE in the iavf list. Filters will be restored after
reset completion.
Steps to reproduce:
-------------------
1. Create a VF. Here VF is enp8s0.
2. Assign IP addresses to VF and link partner and ping continuously
from remote. Here remote IP is 1.1.1.1.
3. Check default RX Queue of traffic.
ethtool -S enp8s0 | grep -E "rx-[[:digit:]]+\.packets"
4. Add filter - change default RX Queue (to 15 here)
ethtool -U ens8s0 flow-type ip4 src-ip 1.1.1.1 action 15 loc 5
5. Ensure filter gets added and traffic is received on RX queue 15 now.
Link event testing:
-------------------
6. Bring VF link down and up. If traffic flows to configured queue 15,
test is success, otherwise it is a failure.
Reset event testing:
--------------------
7. Reset the VF. If traffic flows to configured queue 15, test is success,
otherwise it is a failure.
Fixes: 0dbfbabb84 ("iavf: Add framework to enable ethtool ntuple filters")
Signed-off-by: Piotr Gardocki <piotrx.gardocki@intel.com>
Reviewed-by: Larysa Zaremba <larysa.zaremba@intel.com>
Signed-off-by: Ranganatha Rao <ranganatha.rao@intel.com>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The iavf client interface was added in 2017 by commit ed0e894de7
("i40evf: add client interface"), but there have never been any in-tree
callers.
It's not useful for future development either. The Intel out-of-tree
iavf and irdma drivers instead use an auxiliary bus, which is a better
solution.
Remove the iavf client interface code. Also gone are the client_task
work and the client_lock mutex.
Signed-off-by: Michal Schmidt <mschmidt@redhat.com>
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Link: https://lore.kernel.org/r/20231027175941.1340255-9-jacob.e.keller@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The information whether a netdev has been registered is already present
in the netdev itself. There's no need for a driver flag with the same
meaning.
Signed-off-by: Michal Schmidt <mschmidt@redhat.com>
Reviewed-by: Wojciech Drewek <wojciech.drewek@intel.com>
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Link: https://lore.kernel.org/r/20231027175941.1340255-6-jacob.e.keller@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Add helper for set iavf aq request AVF_FLAG_AQ_* and immediately
schedule watchdog_task. Helper will be used in cases where it is
necessary to run aq requests asap
Signed-off-by: Petr Oros <poros@redhat.com>
Co-developed-by: Michal Schmidt <mschmidt@redhat.com>
Signed-off-by: Michal Schmidt <mschmidt@redhat.com>
Co-developed-by: Ivan Vecera <ivecera@redhat.com>
Signed-off-by: Ivan Vecera <ivecera@redhat.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Previously CRC stripping was always enabled for VF.
Now it is possible to turn off CRC stripping via ethtool:
#ethtool -K <interface> rx-fcs on
To turn off CRC stripping, first VLAN stripping must be disabled:
#ethtool -K <interface> rx-vlan-offload off
if any VLAN interfaces exists, otherwise VLAN stripping will be turned
off by the driver.
In iavf_configure_queues add check if CRC stripping is enabled for
VF, if it's enabled then set crc_disabled to false on every VF's
queue. In iavf_set_features add check if CRC stripping setting was
changed then schedule reset.
Signed-off-by: Norbert Zulinski <norbertx.zulinski@intel.com>
Reviewed-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: Ahmed Zaki <ahmed.zaki@intel.com>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Currently when configuring promiscuous mode on the AVF we detect a
change in the netdev->flags. We use IFF_PROMISC and IFF_ALLMULTI to
determine whether or not we need to request/release promiscuous mode
and/or multicast promiscuous mode. The problem is that the AQ calls for
setting/clearing promiscuous/multicast mode are treated separately. This
leads to a case where we can trigger two promiscuous mode AQ calls in
a row with the incorrect state. To fix this make a few changes.
Use IAVF_FLAG_AQ_CONFIGURE_PROMISC_MODE instead of the previous
IAVF_FLAG_AQ_[REQUEST|RELEASE]_[PROMISC|ALLMULTI] flags.
In iavf_set_rx_mode() detect if there is a change in the
netdev->flags in comparison with adapter->flags and set the
IAVF_FLAG_AQ_CONFIGURE_PROMISC_MODE aq_required bit. Then in
iavf_process_aq_command() only check for IAVF_FLAG_CONFIGURE_PROMISC_MODE
and call iavf_set_promiscuous() if it's set.
In iavf_set_promiscuous() check again to see which (if any) promiscuous
mode bits have changed when comparing the netdev->flags with the
adapter->flags. Use this to set the flags which get sent to the PF
driver.
Add a spinlock that is used for updating current_netdev_promisc_flags
and only allows one promiscuous mode AQ at a time.
[1] Fixes the fact that we will only have one AQ call in the aq_required
queue at any one time.
[2] Streamlines the change in promiscuous mode to only set one AQ
required bit.
[3] This allows us to keep track of the current state of the flags and
also makes it so we can take the most recent netdev->flags promiscuous
mode state.
[4] This fixes the problem where a change in the netdev->flags can cause
IAVF_FLAG_AQ_CONFIGURE_PROMISC_MODE to be set in iavf_set_rx_mode(),
but cleared in iavf_set_promiscuous() before the change is ever made via
AQ call.
Fixes: 47d3483988 ("i40evf: Add driver support for promiscuous mode")
Signed-off-by: Brett Creeley <brett.creeley@intel.com>
Signed-off-by: Ahmed Zaki <ahmed.zaki@intel.com>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
There are five virtchnl structures, which are allocated and checked in
the code as `nents + 1`, meaning that they always have memory for one
excessive element regardless of their actual number. This comes from
that their sizeof() includes space for 1 element and then they get
allocated via struct_size() or its open-coded equivalents, passing
the actual number of elements.
Expand virtchnl_struct_size() to handle such structures and replace
those 1-elem arrays with proper flex ones. Also fix several places
which open-code %IAVF_VIRTCHNL_VF_RESOURCE_SIZE. Finally, let the
virtchnl_ether_addr_list size be computed automatically when there's
no enough space for the whole list, otherwise we have to open-code
reverse struct_size() logics.
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Reviewed-by: Kees Cook <keescook@chromium.org>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The reset task is currently scheduled from the watchdog or adminq tasks.
First, all direct calls to schedule the reset task are replaced with the
iavf_schedule_reset(), which is modified to accept the flag showing the
type of reset.
To prevent the reset task from starting once iavf_remove() starts, we need
to check the __IAVF_IN_REMOVE_TASK bit before we schedule it. This is now
easily added to iavf_schedule_reset().
Finally, remove the check for IAVF_FLAG_RESET_NEEDED in the watchdog task.
It is redundant since all callers who set the flag immediately schedules
the reset task.
Fixes: 3ccd54ef44 ("iavf: Fix init state closure on remove")
Fixes: 14756b2ae2 ("iavf: Fix __IAVF_RESETTING state usage")
Signed-off-by: Ahmed Zaki <ahmed.zaki@intel.com>
Signed-off-by: Mateusz Palczewski <mateusz.palczewski@intel.com>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
A driver's lock (crit_lock) is used to serialize all the driver's tasks.
Lockdep, however, shows a circular dependency between rtnl and
crit_lock. This happens when an ndo that already holds the rtnl requests
the driver to reset, since the reset task (in some paths) tries to grab
rtnl to either change real number of queues of update netdev features.
[566.241851] ======================================================
[566.241893] WARNING: possible circular locking dependency detected
[566.241936] 6.2.14-100.fc36.x86_64+debug #1 Tainted: G OE
[566.241984] ------------------------------------------------------
[566.242025] repro.sh/2604 is trying to acquire lock:
[566.242061] ffff9280fc5ceee8 (&adapter->crit_lock){+.+.}-{3:3}, at: iavf_close+0x3c/0x240 [iavf]
[566.242167]
but task is already holding lock:
[566.242209] ffffffff9976d350 (rtnl_mutex){+.+.}-{3:3}, at: iavf_remove+0x6b5/0x730 [iavf]
[566.242300]
which lock already depends on the new lock.
[566.242353]
the existing dependency chain (in reverse order) is:
[566.242401]
-> #1 (rtnl_mutex){+.+.}-{3:3}:
[566.242451] __mutex_lock+0xc1/0xbb0
[566.242489] iavf_init_interrupt_scheme+0x179/0x440 [iavf]
[566.242560] iavf_watchdog_task+0x80b/0x1400 [iavf]
[566.242627] process_one_work+0x2b3/0x560
[566.242663] worker_thread+0x4f/0x3a0
[566.242696] kthread+0xf2/0x120
[566.242730] ret_from_fork+0x29/0x50
[566.242763]
-> #0 (&adapter->crit_lock){+.+.}-{3:3}:
[566.242815] __lock_acquire+0x15ff/0x22b0
[566.242869] lock_acquire+0xd2/0x2c0
[566.242901] __mutex_lock+0xc1/0xbb0
[566.242934] iavf_close+0x3c/0x240 [iavf]
[566.242997] __dev_close_many+0xac/0x120
[566.243036] dev_close_many+0x8b/0x140
[566.243071] unregister_netdevice_many_notify+0x165/0x7c0
[566.243116] unregister_netdevice_queue+0xd3/0x110
[566.243157] iavf_remove+0x6c1/0x730 [iavf]
[566.243217] pci_device_remove+0x33/0xa0
[566.243257] device_release_driver_internal+0x1bc/0x240
[566.243299] pci_stop_bus_device+0x6c/0x90
[566.243338] pci_stop_and_remove_bus_device+0xe/0x20
[566.243380] pci_iov_remove_virtfn+0xd1/0x130
[566.243417] sriov_disable+0x34/0xe0
[566.243448] ice_free_vfs+0x2da/0x330 [ice]
[566.244383] ice_sriov_configure+0x88/0xad0 [ice]
[566.245353] sriov_numvfs_store+0xde/0x1d0
[566.246156] kernfs_fop_write_iter+0x15e/0x210
[566.246921] vfs_write+0x288/0x530
[566.247671] ksys_write+0x74/0xf0
[566.248408] do_syscall_64+0x58/0x80
[566.249145] entry_SYSCALL_64_after_hwframe+0x72/0xdc
[566.249886]
other info that might help us debug this:
[566.252014] Possible unsafe locking scenario:
[566.253432] CPU0 CPU1
[566.254118] ---- ----
[566.254800] lock(rtnl_mutex);
[566.255514] lock(&adapter->crit_lock);
[566.256233] lock(rtnl_mutex);
[566.256897] lock(&adapter->crit_lock);
[566.257388]
*** DEADLOCK ***
The deadlock can be triggered by a script that is continuously resetting
the VF adapter while doing other operations requiring RTNL, e.g:
while :; do
ip link set $VF up
ethtool --set-channels $VF combined 2
ip link set $VF down
ip link set $VF up
ethtool --set-channels $VF combined 4
ip link set $VF down
done
Any operation that triggers a reset can substitute "ethtool --set-channles"
As a fix, add a new task "finish_config" that do all the work which
needs rtnl lock. With the exception of iavf_remove(), all work that
require rtnl should be called from this task.
As for iavf_remove(), at the point where we need to call
unregister_netdevice() (and grab rtnl_lock), we make sure the finish_config
task is not running (cancel_work_sync()) to safely grab rtnl. Subsequent
finish_config work cannot restart after that since the task is guarded
by the __IAVF_IN_REMOVE_TASK bit in iavf_schedule_finish_config().
Fixes: 5ac49f3c27 ("iavf: use mutexes for locking of critical sections")
Signed-off-by: Ahmed Zaki <ahmed.zaki@intel.com>
Signed-off-by: Mateusz Palczewski <mateusz.palczewski@intel.com>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
There was a fail when trying to add the interface to bonding
right after changing the MTU on the interface. It was caused
by bonding interface unable to open the interface due to
interface being in __RESETTING state because of MTU change.
Add new reset_waitqueue to indicate that reset has finished.
Add waiting for reset to finish in callbacks which trigger hw reset:
iavf_set_priv_flags(), iavf_change_mtu() and iavf_set_ringparam().
We use a 5000ms timeout period because on Hyper-V based systems,
this operation takes around 3000-4000ms. In normal circumstances,
it doesn't take more than 500ms to complete.
Add a function iavf_wait_for_reset() to reuse waiting for reset code and
use it also in iavf_set_channels(), which already waits for reset.
We don't use error handling in iavf_set_channels() as this could
cause the device to be in incorrect state if the reset was scheduled
but hit timeout or the waitng function was interrupted by a signal.
Fixes: 4e5e6b5d9d ("iavf: Fix return of set the new channel count")
Signed-off-by: Marcin Szycik <marcin.szycik@linux.intel.com>
Co-developed-by: Dawid Wesierski <dawidx.wesierski@intel.com>
Signed-off-by: Dawid Wesierski <dawidx.wesierski@intel.com>
Signed-off-by: Sylwester Dziedziuch <sylwesterx.dziedziuch@intel.com>
Signed-off-by: Kamil Maziarz <kamil.maziarz@intel.com>
Signed-off-by: Mateusz Palczewski <mateusz.palczewski@intel.com>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Make all possible functions static.
Move iavf_force_wb() up to avoid forward declaration.
Suggested-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Reviewed-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Enable more than 32 IRQs by removing the u32 bit mask in
iavf_irq_enable_queues(). There is no need for the mask as there are no
callers that select individual IRQs through the bitmask. Also, if the PF
allocates more than 32 IRQs, this mask will prevent us from using all of
them.
Modify the comment in iavf_register.h to show that the maximum number
allowed for the IRQ index is 63 as per the iAVF standard 1.0 [1].
link: [1] https://www.intel.com/content/dam/www/public/us/en/documents/product-specifications/ethernet-adaptive-virtual-function-hardware-spec.pdf
Fixes: 5eae00c57f ("i40evf: main driver core")
Signed-off-by: Ahmed Zaki <ahmed.zaki@intel.com>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Reviewed-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Link: https://lore.kernel.org/r/20230608200226.451861-1-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The VLAN filters info is currently being held in a list and 2 bitmaps
(active_cvlans and active_svlans). We are experiencing some racing where
data is not in sync in the list and bitmaps. For example, the VLAN is
initially added to the list but only when the PF replies, it is added to
the bitmap. If a user adds many V2 VLANS before the PF responds:
while [ $((i++)) ]
ip l add l eth0 name eth0.$i type vlan id $i
we might end up with more VLAN list entries than the designated limit.
Also, The "ip link show" will show more links added than the PF limit.
On the other and, the bitmaps are only used to check the number of VLAN
filters and to re-enable the filters when the interface goes from DOWN to
UP.
This patch gets rid of the bitmaps and uses the list only. To do that,
the states of the VLAN filter are modified:
1 - IAVF_VLAN_REMOVE: the entry needs to be totally removed after informing
the PF. This is the "ip link del eth0.$i" path.
2 - IAVF_VLAN_DISABLE: (new) the netdev went down. The filter needs to be
removed from the PF and then marked INACTIVE.
3 - IAVF_VLAN_INACTIVE: (new) no PF filter exists, but the user did not
delete the VLAN.
Fixes: 48ccc43ecf ("iavf: Add support VIRTCHNL_VF_OFFLOAD_VLAN_V2 during netdev config")
Signed-off-by: Ahmed Zaki <ahmed.zaki@intel.com>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The VLAN filter states are currently being saved as individual bits.
This is error prone as multiple bits might be mistakenly set.
Fix by replacing the bits with a single state enum. Also, add an
"ACTIVE" state for filters that are accepted by the PF.
Signed-off-by: Ahmed Zaki <ahmed.zaki@intel.com>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Since mqprio is a scheduler and not a classifier, move its offload
structure to pkt_sched.h, where struct tc_taprio_qopt_offload also lies.
Also update some header inclusions in drivers that access this
structure, to the best of my abilities.
Cc: Igor Russkikh <irusskikh@marvell.com>
Cc: Yisen Zhuang <yisen.zhuang@huawei.com>
Cc: Salil Mehta <salil.mehta@huawei.com>
Cc: Jesse Brandeburg <jesse.brandeburg@intel.com>
Cc: Tony Nguyen <anthony.l.nguyen@intel.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Saeed Mahameed <saeedm@nvidia.com>
Cc: Leon Romanovsky <leon@kernel.org>
Cc: Horatiu Vultur <horatiu.vultur@microchip.com>
Cc: Lars Povlsen <lars.povlsen@microchip.com>
Cc: Steen Hegelund <Steen.Hegelund@microchip.com>
Cc: Daniel Machon <daniel.machon@microchip.com>
Cc: UNGLinuxDriver@microchip.com
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Since the latest Intel hardware does both IWARP and ROCE, rename the
term IWARP in the virtchnl header to be RDMA. Do this for both upper and
lower case instances. Many of the non-virtchnl.h changes were done with
regular expression replacements using perl like:
perl -p -i -e 's/_IWARP/_RDMA/' <files>
perl -p -i -e 's/_iwarp/_rdma/' <files>
and I had to pick up a few instances manually.
The virtchnl.h header has some comments and clarity added around when to
use certain defines.
note: had to fix a checkpatch warning for a long line by wrapping one of
the lines I changed.
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Tested-by: Jakub Andrysiak <jakub.andrysiak@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
We are seeing an issue where setting the MAC address on iavf fails with
EAGAIN after the 2.5s timeout expires in iavf_set_mac().
There is the following deadlock scenario:
iavf_set_mac(), holding rtnl_lock, waits on:
iavf_watchdog_task (within iavf_wq) to send a message to the PF,
and
iavf_adminq_task (within iavf_wq) to receive a response from the PF.
In this adapter state (>=__IAVF_DOWN), these tasks do not need to take
rtnl_lock, but iavf_wq is a global single-threaded workqueue, so they
may get stuck waiting for another adapter's iavf_watchdog_task to run
iavf_init_config_adapter(), which does take rtnl_lock.
The deadlock resolves itself by the timeout in iavf_set_mac(),
which results in EAGAIN returned to userspace.
Let's break the deadlock loop by changing iavf_wq into a per-adapter
workqueue, so that one adapter's tasks are not blocked by another's.
Fixes: 35a2443d09 ("iavf: Add waiting for response from PF in set mac")
Co-developed-by: Ivan Vecera <ivecera@redhat.com>
Signed-off-by: Ivan Vecera <ivecera@redhat.com>
Signed-off-by: Michal Schmidt <mschmidt@redhat.com>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
IAVF_FLAG_INITIAL_MAC_SET prevents waiting on iavf_is_mac_set_handled()
the first time the MAC is set. This breaks gratuitous ARP because the
MAC address has not been updated yet when the gARP packet is sent out.
Current behaviour:
$ echo 1 > /sys/class/net/ens4f0/device/sriov_numvfs
iavf 0000:88:02.0: MAC address: ee:04:19:14:ec:ea
$ ip addr add 192.168.1.1/24 dev ens4f0v0
$ ip link set dev ens4f0v0 up
$ echo 1 > /proc/sys/net/ipv4/conf/ens4f0v0/arp_notify
$ ip link set ens4f0v0 addr 00:11:22:33:44:55
07:23:41.676611 ee:04:19:14:ec:ea > ff:ff:ff:ff:ff:ff, ethertype ARP (0x0806), length 42: Request who-has 192.168.1.1 tell 192.168.1.1, length 28
With IAVF_FLAG_INITIAL_MAC_SET removed:
$ echo 1 > /sys/class/net/ens4f0/device/sriov_numvfs
iavf 0000:88:02.0: MAC address: 3e:8a:16:a2:37:6d
$ ip addr add 192.168.1.1/24 dev ens4f0v0
$ ip link set dev ens4f0v0 up
$ echo 1 > /proc/sys/net/ipv4/conf/ens4f0v0/arp_notify
$ ip link set ens4f0v0 addr 00:11:22:33:44:55
07:28:01.836608 00:11:22:33:44:55 > ff:ff:ff:ff:ff:ff, ethertype ARP (0x0806), length 42: Request who-has 192.168.1.1 tell 192.168.1.1, length 28
Fixes: 35a2443d09 ("iavf: Add waiting for response from PF in set mac")
Signed-off-by: Stefan Assmann <sassmann@kpanic.de>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Fix max_rate option in TC, check for proper quanta boundaries.
Check for minimum value provided and if it fits expected 50Mbps
quanta.
Without this patch, iavf could send settings for max_rate limiting
that would be accepted from by PF even the max_rate option is less
than expected 50Mbps quanta. It results in no rate limiting
on traffic as rate limiting will be floored to 0.
Example:
tc qdisc add dev $vf root mqprio num_tc 3 map 0 2 1 queues \
2@0 2@2 2@4 hw 1 mode channel shaper bw_rlimit \
max_rate 50Mbps 500Mbps 500Mbps
Should limit TC0 to circa 50 Mbps
tc qdisc add dev $vf root mqprio num_tc 3 map 0 2 1 queues \
2@0 2@2 2@4 hw 1 mode channel shaper bw_rlimit \
max_rate 0Mbps 100Kbit 500Mbps
Should return error
Fixes: d5b33d0244 ("i40evf: add ndo_setup_tc callback to i40evf")
Signed-off-by: Przemyslaw Patynowski <przemyslawx.patynowski@intel.com>
Signed-off-by: Jun Zhang <xuejun.zhang@intel.com>
Tested-by: Bharathi Sreenivas <bharathi.sreenivas@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Fix debug prints, by adding missing state prints.
Extend iavf_state_str by strings for __IAVF_INIT_EXTENDED_CAPS and
__IAVF_INIT_CONFIG_ADAPTER.
Without this patch, when enabling debug prints for iavf.h, user will
see:
iavf 0000:06:0e.0: state transition from:__IAVF_INIT_GET_RESOURCES to:__IAVF_UNKNOWN_STATE
iavf 0000:06:0e.0: state transition from:__IAVF_UNKNOWN_STATE to:__IAVF_UNKNOWN_STATE
Fixes: 605ca7c5c6 ("iavf: Fix kernel BUG in free_msi_irqs")
Signed-off-by: Przemyslaw Patynowski <przemyslawx.patynowski@intel.com>
Signed-off-by: Jun Zhang <xuejun.zhang@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Remove from supported_coalesce_params ETHTOOL_COALESCE_MAX_FRAMES
and ETHTOOL_COALESCE_MAX_FRAMES_IRQ. As tx-frames-irq allowed
user to change budget for iavf_clean_tx_irq, remove work_limit
and use define for budget.
Without this patch there would be possibility to change rx/tx-frames
and rx/tx-frames-irq, which for rx/tx-frames did nothing, while for
rx/tx-frames-irq it changed rx/tx-frames and only changed budget
for cleaning NAPI poll.
Fixes: fbb7ddfef2 ("i40evf: core ethtool functionality")
Signed-off-by: Przemyslaw Patynowski <przemyslawx.patynowski@intel.com>
Signed-off-by: Jun Zhang <xuejun.zhang@intel.com>
Tested-by: Marek Szlosek <marek.szlosek@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Fix VLAN addition, so that PF driver does not reject whole VLAN batch.
Add VLAN reject handling, so rejected VLANs, won't litter VLAN filter
list. Fix handling of active_(c/s)vlans, so it will be possible to
re-add VLAN filters for user.
Without this patch, after changing trust to off, with VLAN filters
saturated, no VLAN is added, due to PF rejecting addition.
Fixes: 92fc508598 ("iavf: Restrict maximum VLAN filters for VIRTCHNL_VF_OFFLOAD_VLAN_V2")
Signed-off-by: Przemyslaw Patynowski <przemyslawx.patynowski@intel.com>
Signed-off-by: Jedrzej Jagielski <jedrzej.jagielski@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Make iavf_set_mac synchronous by waiting for a response
from a PF. Without this iavf_set_mac is always returning
success even though set_mac can be rejected by a PF.
This ensures that when set_mac exits netdev MAC is updated.
This is needed for sending ARPs with correct MAC after
changing VF's MAC. This is also needed by bonding module.
Signed-off-by: Sylwester Dziedziuch <sylwesterx.dziedziuch@intel.com>
Signed-off-by: Mateusz Palczewski <mateusz.palczewski@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
In some cases overloaded flag IAVF_FLAG_REINIT_ITR_NEEDED
which should indicate that interrupts need to be completely
reinitialized during reset leads to RTNL deadlocks using ethtool -C
while a reset is in progress.
To fix, it was added a new flag IAVF_FLAG_REINIT_MSIX_NEEDED
used to trigger MSI-X reinit.
New combined setting is fixed adopt after VF reset.
This has been implemented by call reinit interrupt scheme
during VF reset.
Without this fix new combined setting has never been adopted.
Fixes: 209f2f9c71 ("iavf: Add support for VIRTCHNL_VF_OFFLOAD_VLAN_V2 negotiation")
Signed-off-by: Grzegorz Szczurek <grzegorzx.szczurek@intel.com>
Signed-off-by: Mitch Williams <mitch.a.williams@intel.com>
Signed-off-by: Michal Maloszewski <michal.maloszewski@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Several functions in the iAVF core files take status values of the enum
iavf_status and convert them into integer values. This leads to
confusion as functions return both Linux errno values and status codes
intermixed. Reporting status codes as if they were "errno" values can
lead to confusion when reviewing error logs. Additionally, it can lead
to unexpected behavior if a return value is not interpreted properly.
Fix this by introducing iavf_status_to_errno, a switch that explicitly
converts from the status codes into an appropriate error value. Also
introduce a virtchnl_status_to_errno function for the one case where we
were returning both virtchnl status codes and iavf_status codes in the
same function.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Signed-off-by: Mateusz Palczewski <mateusz.palczewski@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
In order to handle the capability exchange necessary for
VIRTCHNL_VF_OFFLOAD_VLAN_V2, the driver must send
a VIRTCHNL_OP_GET_OFFLOAD_VLAN_V2_CAPS message. This must occur prior to
__IAVF_CONFIG_ADAPTER, and the driver must wait for the response from
the PF.
To handle this, the __IAVF_INIT_GET_OFFLOAD_VLAN_V2_CAPS state was
introduced. This state is intended to process the response from the VLAN
V2 caps message. This works ok, but is difficult to extend to adding
more extended capability exchange.
Existing (and future) AVF features are relying more and more on these
sort of extended ops for processing additional capabilities. Just like
VLAN V2, this exchange must happen prior to __IAVF_CONFIG_ADPATER.
Since we only send one outstanding AQ message at a time during init, it
is not clear where to place this state. Adding more capability specific
states becomes a mess. Instead of having the "previous" state send
a message and then transition into a capability-specific state,
introduce __IAVF_EXTENDED_CAPS state. This state will use a list of
extended_caps that determines what messages to send and receive. As long
as there are extended_caps bits still set, the driver will remain in
this state performing one send or one receive per state machine loop.
Refactor the VLAN V2 negotiation to use this new state, and remove the
capability-specific state. This makes it significantly easier to add
a new similar capability exchange going forward.
Extended capabilities are processed by having an associated SEND and
RECV extended capability bit. During __IAVF_EXTENDED_CAPS, the
driver checks these bits in order by feature, first the send bit for
a feature, then the recv bit for a feature. Each send flag will call
a function that sends the necessary response, while each receive flag
will wait for the response from the PF. If a given feature can't be
negotiated with the PF, the associated flags will be cleared in
order to skip processing of that feature.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Signed-off-by: Mateusz Palczewski <mateusz.palczewski@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
