The second tagged commit started sometimes (very rarely, but possible)
throwing WARNs from
net/core/page_pool.c:page_pool_disable_direct_recycling().
Turned out idpf frees interrupt vectors with embedded NAPIs *before*
freeing the queues making page_pools' NAPI pointers lead to freed
memory before these pools are destroyed by libeth.
It's not clear whether there are other accesses to the freed vectors
when destroying the queues, but anyway, we usually free queue/interrupt
vectors only when the queues are destroyed and the NAPIs are guaranteed
to not be referenced anywhere.
Invert the allocation and freeing logic making queue/interrupt vectors
be allocated first and freed last. Vectors don't require queues to be
present, so this is safe. Additionally, this change allows to remove
that useless queue->q_vector pointer cleanup, as vectors are still
valid when freeing the queues (+ both are freed within one function,
so it's not clear why nullify the pointers at all).
Fixes: 1c325aac10 ("idpf: configure resources for TX queues")
Fixes: 90912f9f4f ("idpf: convert header split mode to libeth + napi_build_skb()")
Reported-by: Michal Kubiak <michal.kubiak@intel.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Tested-by: Krishneil Singh <krishneil.k.singh@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Link: https://patch.msgid.link/20240806220923.3359860-4-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The initialization of vport interrupt consists of two functions:
1) idpf_vport_intr_init() where a generic configuration is done
2) idpf_vport_intr_req_irq() where the irq for each q_vector is
requested.
The first function used to create a base name for each interrupt using
"kasprintf()" call. Unfortunately, although that call allocated memory
for a text buffer, that memory was never released.
Fix this by removing creating the interrupt base name in 1).
Instead, always create a full interrupt name in the function 2), because
there is no need to create a base name separately, considering that the
function 2) is never called out of idpf_vport_intr_init() context.
Fixes: d4d5587182 ("idpf: initialize interrupts and enable vport")
Cc: stable@vger.kernel.org # 6.7
Signed-off-by: Michal Kubiak <michal.kubiak@intel.com>
Reviewed-by: Pavan Kumar Linga <pavan.kumar.linga@intel.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Tested-by: Krishneil Singh <krishneil.k.singh@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Link: https://patch.msgid.link/20240806220923.3359860-3-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The second tagged commit introduced a UAF, as it removed restoring
q_vector->vport pointers after reinitializating the structures.
This is due to that all queue allocation functions are performed here
with the new temporary vport structure and those functions rewrite
the backpointers to the vport. Then, this new struct is freed and
the pointers start leading to nowhere.
But generally speaking, the current logic is very fragile. It claims
to be more reliable when the system is low on memory, but in fact, it
consumes two times more memory as at the moment of running this
function, there are two vports allocated with their queues and vectors.
Moreover, it claims to prevent the driver from running into "bad state",
but in fact, any error during the rebuild leaves the old vport in the
partially allocated state.
Finally, if the interface is down when the function is called, it always
allocates a new queue set, but when the user decides to enable the
interface later on, vport_open() allocates them once again, IOW there's
a clear memory leak here.
Just don't allocate a new queue set when performing a reset, that solves
crashes and memory leaks. Readd the old queue number and reopen the
interface on rollback - that solves limbo states when the device is left
disabled and/or without HW queues enabled.
Fixes: 02cbfba1ad ("idpf: add ethtool callbacks")
Fixes: e4891e4687 ("idpf: split &idpf_queue into 4 strictly-typed queue structures")
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Tested-by: Krishneil Singh <krishneil.k.singh@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Link: https://patch.msgid.link/20240806220923.3359860-2-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
idpf uses Page Pool for data buffers with hardcoded buffer lengths of
4k for "classic" buffers and 2k for "short" ones. This is not flexible
and does not ensure optimal memory usage. Why would you need 4k buffers
when the MTU is 1500?
Use libeth for the data buffers and don't hardcode any buffer sizes. Let
them be calculated from the MTU for "classics" and then divide the
truesize by 2 for "short" ones. The memory usage is now greatly reduced
and 2 buffer queues starts make sense: on frames <= 1024, you'll recycle
(and resync) a page only after 4 HW writes rather than two.
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Currently, idpf uses the following model for the header buffers:
* buffers are allocated via dma_alloc_coherent();
* when receiving, napi_alloc_skb() is called and then the header is
copied to the newly allocated linear part.
This is far from optimal as DMA coherent zone is slow on many systems
and memcpy() neutralizes the idea and benefits of the header split. Not
speaking of that XDP can't be run on DMA coherent buffers, but at the
same time the idea of allocating an skb to run XDP program is ill.
Instead, use libeth to create page_pools for the header buffers, allocate
them dynamically and then build an skb via napi_build_skb() around them
with no memory copy. With one exception...
When you enable header split, you expect you'll always have a separate
header buffer, so that you could reserve headroom and tailroom only
there and then use full buffers for the data. For example, this is how
TCP zerocopy works -- you have to have the payload aligned to PAGE_SIZE.
The current hardware running idpf does *not* guarantee that you'll
always have headers placed separately. For example, on my setup, even
ICMP packets are written as one piece to the data buffers. You can't
build a valid skb around a data buffer in this case.
To not complicate things and not lose TCP zerocopy etc., when such thing
happens, use the empty header buffer and pull either full frame (if it's
short) or the Ethernet header there and build an skb around it. GRO
layer will pull more from the data buffer later. This W/A will hopefully
be removed one day.
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Page Pool Ethtool stats are deprecated since the Netlink Page Pool
interface introduction.
idpf receives big changes in Rx buffer management, including &page_pool
layout, so keeping these deprecated stats does only harm, not speaking
of that CONFIG_IDPF selects CONFIG_PAGE_POOL_STATS unconditionally,
while the latter is often turned off for better performance.
Remove all the references to PP stats from the Ethtool code. The stats
are still available in their full via the generic Netlink interface.
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
idpf's in-kernel parsed ptype structure is almost identical to the one
used in the previous Intel drivers, which means it can be converted to
use libeth's definitions and even helpers. The only difference is that
it doesn't use a constant table (libie), rather than one obtained from
the device.
Remove the driver counterpart and use libeth's helpers for hashes and
checksums. This slightly optimizes skb fields processing due to faster
checks. Also don't define big static array of ptypes in &idpf_vport --
allocate them dynamically. The pointer to it is anyway cached in
&idpf_rx_queue.
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Currently, all HW supporting idpf supports the singleq model, but none
of it advertises it by default, as splitq is supported and preferred
for multiple reasons. Still, this almost dead code often times adds
hotpath branches and redundant cacheline accesses.
While it can't currently be removed, add CONFIG_IDPF_SINGLEQ and build
the singleq code only when it's enabled manually. This corresponds to
-10 Kb of object code size and a good bunch of hotpath checks.
idpf_is_queue_model_split() works as a gate and compiles out to `true`
when the config option is disabled.
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>
It makes no sense to have a second &net_device_ops struct (800 bytes of
rodata) with only one difference in .ndo_start_xmit, which can easily
be just one `if`. This `if` is a drop in the ocean and you won't see
any difference.
Define unified idpf_xmit_start(). The preparation for sending is the
same, just call either idpf_tx_splitq_frame() or idpf_tx_singleq_frame()
depending on the active model to actually map and send the skb.
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Now that the queue and queue vector structures are separated and laid
out optimally, group the fields as read-mostly, read-write, and cold
cachelines and add size assertions to make sure new features won't push
something out of its place and provoke perf regression.
Despite looking innocent, this gives up to 2% of perf bump on Rx.
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
With CONFIG_MAXSMP, sizeof(cpumask_t) is 1 Kb. The queue vector
structure has them embedded, which means 1 additional Kb of not
really hotpath data.
We have cpumask_var_t, which is either an embedded cpumask or a pointer
for allocating it dynamically when it's big. Use it instead of plain
cpumasks and put &idpf_q_vector on a good diet.
Also remove redundant pointer to the interrupt name from the structure.
request_irq() saves it and free_irq() returns it on deinit, so that you
can free the memory.
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>
Currently, sizeof(struct idpf_queue) is 32 Kb.
This is due to the 12-bit hashtable declaration at the end of the queue.
This HT is needed only for Tx queues when the flow scheduling mode is
enabled. But &idpf_queue is unified for all of the queue types,
provoking excessive memory usage.
The unified structure in general makes the code less effective via
suboptimal fields placement. You can't avoid that unless you make unions
each 2 fields. Even then, different field alignment etc., doesn't allow
you to optimize things to the limit.
Split &idpf_queue into 4 structures corresponding to the queue types:
RQ (Rx queue), SQ (Tx queue), FQ (buffer queue), and CQ (completion
queue). Place only needed fields there and shortcuts handy for hotpath.
Allocate the abovementioned hashtable dynamically and only when needed,
keeping &idpf_tx_queue relatively short (192 bytes, same as Rx). This HT
is used only for OOO completions, which aren't really hotpath anyway.
Note that this change must be done atomically, otherwise it's really
easy to get lost and miss something.
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
In C, we have structures and unions.
Casting `void *` via macros is not only error-prone, but also looks
confusing and awful in general.
In preparation for splitting the queue structs, replace it with a
union and direct array dereferences.
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Reviewed-by: Mina Almasry <almasrymina@google.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Currently, idpf enables NAPI and interrupts prior to allocating Rx
buffers.
This may lead to frame loss (there are no buffers to place incoming
frames) and even crashes on quick ifup-ifdown. Interrupts must be
enabled only after all the resources are here and available.
Split interrupt init into two phases: initialization and enabling,
and perform the second only after the queues are fully initialized.
Note that we can't just move interrupt initialization down the init
process, as the queues must have correct a ::q_vector pointer set
and NAPI already added in order to allocate buffers correctly.
Also, during the deinit process, disable HW interrupts first and
only then disable NAPI. Otherwise, there can be a HW event leading
to napi_schedule(), but the NAPI will already be unavailable.
Fixes: d4d5587182 ("idpf: initialize interrupts and enable vport")
Reported-by: Michal Kubiak <michal.kubiak@intel.com>
Reviewed-by: Wojciech Drewek <wojciech.drewek@intel.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Tested-by: Krishneil Singh <krishneil.k.singh@intel.com>
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Link: https://lore.kernel.org/r/20240523-net-2024-05-23-intel-net-fixes-v1-1-17a923e0bb5f@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Unlike ice, idpf does not check, if user has requested at least 1 combined
channel. Instead, it relies on a check in the core code. Unfortunately, the
check does not trigger for us because of the hacky .set_channels()
interpretation logic that is not consistent with the core code.
This naturally leads to user being able to trigger a crash with an invalid
input. This is how:
1. ethtool -l <IFNAME> -> combined: 40
2. ethtool -L <IFNAME> rx 0 tx 0
combined number is not specified, so command becomes {rx_count = 0,
tx_count = 0, combined_count = 40}.
3. ethnl_set_channels checks, if there is at least 1 RX and 1 TX channel,
comparing (combined_count + rx_count) and (combined_count + tx_count)
to zero. Obviously, (40 + 0) is greater than zero, so the core code
deems the input OK.
4. idpf interprets `rx 0 tx 0` as 0 channels and tries to proceed with such
configuration.
The issue has to be solved fundamentally, as current logic is also known to
cause AF_XDP problems in ice [0].
Interpret the command in a way that is more consistent with ethtool
manual [1] (--show-channels and --set-channels) and new ice logic.
Considering that in the idpf driver only the difference between RX and TX
queues forms dedicated channels, change the correct way to set number of
channels to:
ethtool -L <IFNAME> combined 10 /* For symmetric queues */
ethtool -L <IFNAME> combined 8 tx 2 rx 0 /* For asymmetric queues */
[0] https://lore.kernel.org/netdev/20240418095857.2827-1-larysa.zaremba@intel.com/
[1] https://man7.org/linux/man-pages/man8/ethtool.8.html
Fixes: 02cbfba1ad ("idpf: add ethtool callbacks")
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Reviewed-by: Igor Bagnucki <igor.bagnucki@intel.com>
Signed-off-by: Larysa Zaremba <larysa.zaremba@intel.com>
Tested-by: Krishneil Singh <krishneil.k.singh@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Disabling tcp-data-split on idpf silently fails:
# ethtool -G $NETDEV tcp-data-split off
# ethtool -g $NETDEV | grep 'TCP data split'
TCP data split: on
But it works if you also change 'tx' or 'rx':
# ethtool -G $NETDEV tcp-data-split off tx 256
# ethtool -g $NETDEV | grep 'TCP data split'
TCP data split: off
The bug is in idpf_set_ringparam, where it takes a shortcut out if the
TX and RX sizes are not changing. Fix it by checking also if the
tcp-data-split setting remains unchanged. Only then can the soft reset
be skipped.
Fixes: 9b1aa3ef23 ("idpf: add get/set for Ethtool's header split ringparam")
Reported-by: Xu Du <xudu@redhat.com>
Closes: https://issues.redhat.com/browse/RHEL-36182
Signed-off-by: Michal Schmidt <mschmidt@redhat.com>
Reviewed-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Link: https://lore.kernel.org/r/20240515092414.158079-1-mschmidt@redhat.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Simon reported that ndo_change_mtu() methods were never
updated to use WRITE_ONCE(dev->mtu, new_mtu) as hinted
in commit 501a90c945 ("inet: protect against too small
mtu values.")
We read dev->mtu without holding RTNL in many places,
with READ_ONCE() annotations.
It is time to take care of ndo_change_mtu() methods
to use corresponding WRITE_ONCE()
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: Simon Horman <horms@kernel.org>
Closes: https://lore.kernel.org/netdev/20240505144608.GB67882@kernel.org/
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Sabrina Dubroca <sd@queasysnail.net>
Reviewed-by: Simon Horman <horms@kernel.org>
Acked-by: Shannon Nelson <shannon.nelson@amd.com>
Link: https://lore.kernel.org/r/20240506102812.3025432-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
In the very rare case where a packet type is unknown to the driver,
idpf_rx_process_skb_fields would return early without calling
eth_type_trans to set the skb protocol / the network layer handler.
This is especially problematic if tcpdump is running when such a
packet is received, i.e. it would cause a kernel panic.
Instead, call eth_type_trans for every single packet, even when
the packet type is unknown.
Fixes: 3a8845af66 ("idpf: add RX splitq napi poll support")
Reported-by: Balazs Nemeth <bnemeth@redhat.com>
Signed-off-by: Joshua Hay <joshua.a.hay@intel.com>
Reviewed-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Tested-by: Salvatore Daniele <sdaniele@redhat.com>
Signed-off-by: Pavan Kumar Linga <pavan.kumar.linga@intel.com>
Tested-by: Krishneil Singh <krishneil.k.singh@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Both virtchnl2.h and its consumer idpf_virtchnl.c are very error-prone.
There are 10 structures with flexible arrays at the end, but 9 of them
has flex member counter in Little Endian.
Make the code a bit more robust by applying __counted_by_le() to those
9. LE platforms is the main target for this driver, so they would
receive additional protection.
While we're here, add __counted_by() to virtchnl2_ptype::proto_id, as
its counter is `u8` regardless of the Endianness.
Compile test on x86_64 (LE) didn't reveal any new issues after applying
the attributes.
Acked-by: Kees Cook <keescook@chromium.org>
Acked-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Link: https://lore.kernel.org/r/20240327142241.1745989-4-aleksander.lobakin@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
To ease maintaining of virtchnl2.h, which already is messy enough,
make it self-contained by adding missing if_ether.h include due to
%ETH_ALEN usage.
At the same time, virtchnl2_lan_desc.h is not used anywhere in the
file, so move this include to idpf_txrx.h to speed up C preprocessing.
Acked-by: Kees Cook <keescook@chromium.org>
Acked-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Link: https://lore.kernel.org/r/20240327142241.1745989-3-aleksander.lobakin@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
__napi_alloc_skb() is napi_alloc_skb() with the added flexibility
of choosing gfp_mask. This is a NAPI function, so GFP_ATOMIC is
implied. The only practical choice the caller has is whether to
set __GFP_NOWARN. But that's a false choice, too, allocation failures
in atomic context will happen, and printing warnings in logs,
effectively for a packet drop, is both too much and very likely
non-actionable.
This leads me to a conclusion that most uses of napi_alloc_skb()
are simply misguided, and should use __GFP_NOWARN in the first
place. We also have a "standard" way of reporting allocation
failures via the queue stat API (qstats::rx-alloc-fail).
The direct motivation for this patch is that one of the drivers
used at Meta calls napi_alloc_skb() (so prior to this patch without
__GFP_NOWARN), and the resulting OOM warning is the top networking
warning in our fleet.
Reviewed-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/20240327040213.3153864-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Fix softirq's not being handled during napi_schedule() call when
receiving marker packets for queue disable by disabling local bottom
half.
The issue can be seen on ifdown:
NOHZ tick-stop error: Non-RCU local softirq work is pending, handler #08!!!
Using ftrace to catch the failing scenario:
ifconfig [003] d.... 22739.830624: softirq_raise: vec=3 [action=NET_RX]
<idle>-0 [003] ..s.. 22739.831357: softirq_entry: vec=3 [action=NET_RX]
No interrupt and CPU is idle.
After the patch when disabling local BH before calling napi_schedule:
ifconfig [003] d.... 22993.928336: softirq_raise: vec=3 [action=NET_RX]
ifconfig [003] ..s1. 22993.928337: softirq_entry: vec=3 [action=NET_RX]
Fixes: c2d548cad1 ("idpf: add TX splitq napi poll support")
Reviewed-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Signed-off-by: Alan Brady <alan.brady@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Tested-by: Krishneil Singh <krishneil.k.singh@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
This error message is at best not really helpful and at worst
misleading. If we're here in idpf_intr_rel we're likely trying to do
remove or reset. If we're in reset, this message will fail because we
lose the virtchnl on reset and HW is going to clean up those resources
regardless in that case. If we're in remove and we get an error here,
we're going to reset the device at the end of remove anyway so not a big
deal. Just remove this message it's not useful.
Tested-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Alan Brady <alan.brady@intel.com>
Tested-by: Krishneil Singh <krishneil.k.singh@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
While we're here improving virtchnl we can include two minor fixes for
the lower level ctrlq flow.
This adds a memory barrier to idpf_post_rx_buffs before we update tail
on the controlq. We should make sure our writes have had a chance to
finish before we tell HW it can touch them.
This also removes some defensive programming in idpf_ctrlq_recv. The
caller should not be using a num_q_msg value of zero or more than the
ring size and it's their responsibility to call functions sanely.
Tested-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Alan Brady <alan.brady@intel.com>
Tested-by: Krishneil Singh <krishneil.k.singh@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
In idpf_remove we need to tear down the virtchnl core with
idpf_vc_core_deinit so we can free up resources and leave things in a
good state. However, in the case where we failed to establish VC
communications we may not have ever actually successfully initialized
the virtchnl core.
This fixes it by setting a bit once we successfully init the virtchnl
core. Then, in deinit, we'll check for it before going on further,
otherwise we just return. Also clear the bit at the end of deinit so we
know it's gone now.
Tested-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Alan Brady <alan.brady@intel.com>
Tested-by: Krishneil Singh <krishneil.k.singh@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
We can now remove a bunch of gross code we don't need anymore like the
vc state bits and vc_buf_lock since everything is using transaction API
now.
Tested-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Reviewed-by: Igor Bagnucki <igor.bagnucki@intel.com>
Signed-off-by: Alan Brady <alan.brady@intel.com>
Tested-by: Krishneil Singh <krishneil.k.singh@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Now that all the messages are using the transaction API, we can rework
idpf_recv_mb_msg quite a lot to simplify it. Due to this, we remove
idpf_find_vport as no longer used and alter idpf_recv_event_msg
slightly.
Tested-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Alan Brady <alan.brady@intel.com>
Tested-by: Krishneil Singh <krishneil.k.singh@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
There are situations where the driver needs to add a MAC filter but
we're explicitly not allowed to sleep so we can wait for a virtchnl
message to complete.
This adds an async_handler for asynchronously sent messages for MAC
filters so that we can better handle if there's an error of some kind.
If success we don't need to do anything else, but if we failed to
program the new filter we really should remove it from our list of MAC
filters. If we don't remove bad filters, what I expect to happen is
after a reset of some kind we try to program the MAC filter again and it
fails again. This is clearly wrong and I would expect to be confusing
for the user.
It could also be the failure is for a delete MAC filter message but
those filters get deleted regardless. Not much we can do about a delete
failure.
Tested-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Alan Brady <alan.brady@intel.com>
Tested-by: Krishneil Singh <krishneil.k.singh@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
This takes care of RSS/SRIOV/MAC and other misc virtchnl messages. This
again is mostly mechanical.
In absence of an async_handler for MAC filters, this will simply
generically report any errors from idpf_vc_xn_forward_async. This
maintains the existing behavior. Follow up patch will add an async
handler for MAC filters to remove bad filters from our list.
While we're here we can also make the code much nicer by converting some
variables to auto-variables where appropriate. This makes it cleaner and
less prone to memory leaking.
There's still a bit more cleanup we can do here to remove stuff that's
not being used anymore now; follow-up patches will take care of loose
ends.
Tested-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Alan Brady <alan.brady@intel.com>
Tested-by: Krishneil Singh <krishneil.k.singh@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
This reworks queue specific virtchnl messages to use the added
transaction API. It is fairly mechanical and generally makes the
functions using it more simple. Functions using transaction API no
longer need to take the vc_buf_lock since it's not using it anymore.
After filling out an idpf_vc_xn_params struct, idpf_vc_xn_exec takes
care of the send and recv handling.
This also converts those functions where appropriate to use
auto-variables instead of manually calling kfree. This greatly
simplifies the memory alloc paths and makes them less prone memory
leaks.
Tested-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Reviewed-by: Igor Bagnucki <igor.bagnucki@intel.com>
Signed-off-by: Alan Brady <alan.brady@intel.com>
Tested-by: Krishneil Singh <krishneil.k.singh@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
This reworks the way vport related virtchnl messages work to take
advantage of the added transaction API. It is fairly mechanical as, to
use the transaction API, the function just needs to fill out an
appropriate idpf_vc_xn_params struct to pass to idpf_vc_xn_exec which
will take care of the actual send and recv.
Tested-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Reviewed-by: Igor Bagnucki <igor.bagnucki@intel.com>
Co-developed-by: Joshua Hay <joshua.a.hay@intel.com>
Signed-off-by: Joshua Hay <joshua.a.hay@intel.com>
Signed-off-by: Alan Brady <alan.brady@intel.com>
Tested-by: Krishneil Singh <krishneil.k.singh@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
This starts refactoring how virtchnl messages are handled by adding a
transaction manager (idpf_vc_xn_manager).
There are two primary motivations here which are to enable handling of
multiple messages at once and to make it more robust in general. As it
is right now, the driver may only have one pending message at a time and
there's no guarantee that the response we receive was actually intended
for the message we sent prior.
This works by utilizing a "cookie" field of the message descriptor. It
is arbitrary what data we put in the cookie and the response is required
to have the same cookie the original message was sent with. Then using a
"transaction" abstraction that uses the completion API to pair responses
to the message it belongs to.
The cookie works such that the first half is the index to the
transaction in our array, and the second half is a "salt" that gets
incremented every message. This enables quick lookups into the array and
also ensuring we have the correct message. The salt is necessary because
after, for example, a message times out and we deem the response was
lost for some reason, we could theoretically reuse the same index but
using a different salt ensures that when we do actually get a response
it's not the old message that timed out previously finally coming in.
Since the number of transactions allocated is U8_MAX and the salt is 8
bits, we can never have a conflict because we can't roll over the salt
without using more transactions than we have available.
This starts by only converting the VIRTCHNL2_OP_VERSION message to use
this new transaction API. Follow up patches will convert all virtchnl
messages to use the API.
Tested-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Reviewed-by: Igor Bagnucki <igor.bagnucki@intel.com>
Co-developed-by: Joshua Hay <joshua.a.hay@intel.com>
Signed-off-by: Joshua Hay <joshua.a.hay@intel.com>
Signed-off-by: Alan Brady <alan.brady@intel.com>
Tested-by: Krishneil Singh <krishneil.k.singh@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
idpf.h is quite heavy. We can reduce the burden a fair bit by
introducing an idpf_virtchnl.h file. This mostly just moves function
declarations but there are many of them. This also makes an attempt to
group those declarations in a way that makes some sense instead of
mishmashed.
Suggested-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Alan Brady <alan.brady@intel.com>
Tested-by: Krishneil Singh <krishneil.k.singh@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
In the arm random config file, kconfig option 'CONFIG_AEABI' is
disabled which results in adding the compiler flag '-mabi=apcs-gnu'.
This causes the compiler to add padding in virtchnl2_ptype
structure to align it to 8 bytes, resulting in the following
size check failure:
include/linux/build_bug.h:78:41: error: static assertion failed: "(6) == sizeof(struct virtchnl2_ptype)"
78 | #define __static_assert(expr, msg, ...) _Static_assert(expr, msg)
| ^~~~~~~~~~~~~~
include/linux/build_bug.h:77:34: note: in expansion of macro '__static_assert'
77 | #define static_assert(expr, ...) __static_assert(expr, ##__VA_ARGS__, #expr)
| ^~~~~~~~~~~~~~~
drivers/net/ethernet/intel/idpf/virtchnl2.h:26:9: note: in expansion of macro 'static_assert'
26 | static_assert((n) == sizeof(struct X))
| ^~~~~~~~~~~~~
drivers/net/ethernet/intel/idpf/virtchnl2.h:982:1: note: in expansion of macro 'VIRTCHNL2_CHECK_STRUCT_LEN'
982 | VIRTCHNL2_CHECK_STRUCT_LEN(6, virtchnl2_ptype);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~
Avoid the compiler padding by using "__packed" structure
attribute for the virtchnl2_ptype struct. Also align the
structure by using "__aligned(2)" for better code optimization.
Fixes: 0d7502a9b4 ("virtchnl: add virtchnl version 2 ops")
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202312220250.ufEm8doQ-lkp@intel.com
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Reviewed-by: Paul Menzel <pmenzel@molgen.mpg.de>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Pavan Kumar Linga <pavan.kumar.linga@intel.com>
Tested-by: Krishneil Singh <krishneil.k.singh@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Link: https://lore.kernel.org/r/20240131222241.2087516-1-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
idpf registers multiple netdevs (virtual ports) for one PCI function,
but it does not provide a way for userspace to distinguish them with
sysfs attributes. Per Documentation/ABI/testing/sysfs-class-net, it is
a bug not to set dev_port for independent ports on the same PCI bus,
device and function.
Without dev_port set, systemd-udevd's default naming policy attempts
to assign the same name ("ens2f0") to all four idpf netdevs on my test
system and obviously fails, leaving three of them with the initial
eth<N> name.
With this patch, systemd-udevd is able to assign unique names to the
netdevs (e.g. "ens2f0", "ens2f0d1", "ens2f0d2", "ens2f0d3").
The Intel-provided out-of-tree idpf driver already sets dev_port. In
this patch I chose to do it in the same place in the idpf_cfg_netdev
function.
Fixes: 0fe45467a1 ("idpf: add create vport and netdev configuration")
Signed-off-by: Michal Schmidt <mschmidt@redhat.com>
Reviewed-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Size of the virtchnl2_rss_key struct should be 7 bytes but the
compiler introduces a padding byte for the structure alignment.
This results in idpf sending an additional byte of memory to the device
control plane than the expected buffer size. As the control plane
enforces virtchnl message size checks to validate the message,
set RSS key message fails resulting in the driver load failure.
Remove implicit compiler padding by using "__packed" structure
attribute for the virtchnl2_rss_key struct.
Also there is no need to use __DECLARE_FLEX_ARRAY macro for the
'key_flex' struct field. So drop it.
Fixes: 0d7502a9b4 ("virtchnl: add virtchnl version 2 ops")
Reviewed-by: Larysa Zaremba <larysa.zaremba@intel.com>
Signed-off-by: Pavan Kumar Linga <pavan.kumar.linga@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Tested-by: Scott Register <scott.register@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
idpf_ring::skb serves only for keeping an incomplete frame between
several NAPI Rx polling cycles, as one cycle may end up before
processing the end of packet descriptor. The pointer is taken from
the ring onto the stack before entering the loop and gets written
there after the loop exits. When inside the loop, only the onstack
pointer is used.
For some reason, the logics is broken in the singleq mode, where the
pointer is taken from the ring each iteration. This means that if a
frame got fragmented into several descriptors, each fragment will have
its own skb, but only the last one will be passed up the stack
(containing garbage), leaving the rest leaked.
Then, on ifdown, rxq::skb is being freed only in the splitq mode, while
it can point to a valid skb in singleq as well. This can lead to a yet
another skb leak.
Just don't touch the ring skb field inside the polling loop, letting
the onstack skb pointer work as expected: build a new skb if it's the
first frame descriptor and attach a frag otherwise. On ifdown, free
rxq::skb unconditionally if the pointer is non-NULL.
Fixes: a5ab9ee0df ("idpf: add singleq start_xmit and napi poll")
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Reviewed-by: Michal Kubiak <michal.kubiak@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Tested-by: Scott Register <scott.register@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Most of idpf correctly uses FIELD_GET and FIELD_PREP, but a couple spots
were missed so fix those.
Automated conversion with coccinelle script and manually fixed up,
including audits for opportunities to convert to {get,encode,replace}
bits functions.
Add conversions to le16_get/encode/replace_bits where appropriate. And
in one place fix up a cast from a u16 to a u16.
@prep2@
constant shift,mask;
type T;
expression a;
@@
-(((T)(a) << shift) & mask)
+FIELD_PREP(mask, a)
@prep@
constant shift,mask;
type T;
expression a;
@@
-((T)((a) << shift) & mask)
+FIELD_PREP(mask, a)
@get@
constant shift,mask;
type T;
expression a;
@@
-((T)((a) & mask) >> shift)
+FIELD_GET(mask, a)
and applied via:
spatch --sp-file field_prep.cocci --in-place --dir \
drivers/net/ethernet/intel/
CC: Alexander Lobakin <aleksander.lobakin@intel.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The get/set_rxfh ethtool ops currently takes the rxfh (RSS) parameters
as direct function arguments. This will force us to change the API (and
all drivers' functions) every time some new parameters are added.
This is part 1/2 of the fix, as suggested in [1]:
- First simplify the code by always providing a pointer to all params
(indir, key and func); the fact that some of them may be NULL seems
like a weird historic thing or a premature optimization.
It will simplify the drivers if all pointers are always present.
- Then make the functions take a dev pointer, and a pointer to a
single struct wrapping all arguments. The set_* should also take
an extack.
Link: https://lore.kernel.org/netdev/20231121152906.2dd5f487@kernel.org/ [1]
Suggested-by: Jakub Kicinski <kuba@kernel.org>
Suggested-by: Jacob Keller <jacob.e.keller@intel.com>
Signed-off-by: Ahmed Zaki <ahmed.zaki@intel.com>
Link: https://lore.kernel.org/r/20231213003321.605376-2-ahmed.zaki@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
idpf supports the header split feature and that feature is always
enabled by default.
However, for flexibility reasons and to simplify some scenarios, it
would be useful to have the support for switching the header split
off (and on) from the userspace.
Address that need by adding the user config parameter, the functions
for disabling (or enabling) the header split feature, and calls to
them from the Ethtool ringparam callbacks.
It still is enabled by default if supported by the hardware.
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Signed-off-by: Michal Kubiak <michal.kubiak@intel.com>
Co-developed-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Link: https://lore.kernel.org/r/20231212142752.935000-3-aleksander.lobakin@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
This patch converts some basic cases of ethtool_sprintf() to
ethtool_puts().
The conversions are used in cases where ethtool_sprintf() was being used
with just two arguments:
| ethtool_sprintf(&data, buffer[i].name);
or when it's used with format string: "%s"
| ethtool_sprintf(&data, "%s", buffer[i].name);
which both now become:
| ethtool_puts(&data, buffer[i].name);
Signed-off-by: Justin Stitt <justinstitt@google.com>
Reviewed-by: Wei Fang <wei.fang@nxp.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Louis Peens <louis.peens@corigine.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
PP_FLAG_PAGE_FRAG is not really needed after pp_frag_count
handling is unified and page_pool_alloc_frag() is supported
in 32-bit arch with 64-bit DMA, so remove it.
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
CC: Lorenzo Bianconi <lorenzo@kernel.org>
CC: Alexander Duyck <alexander.duyck@gmail.com>
CC: Liang Chen <liangchen.linux@gmail.com>
CC: Alexander Lobakin <aleksander.lobakin@intel.com>
Link: https://lore.kernel.org/r/20231020095952.11055-3-linyunsheng@huawei.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
In idpf_vc_core_init, the mailbox work is queued
on a mailbox workqueue but it is not cancelled on error.
This results in a call trace when idpf_mbx_task tries
to access the freed mailbox queue pointer. Fix it by
cancelling the mailbox work in the error path.
Fixes: 4930fbf419 ("idpf: add core init and interrupt request")
Reviewed-by: Wojciech Drewek <wojciech.drewek@intel.com>
Signed-off-by: Pavan Kumar Linga <pavan.kumar.linga@intel.com>
Tested-by: Krishneil Singh <krishneil.k.singh@intel.com>
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Link: https://lore.kernel.org/r/20231023202655.173369-3-jacob.e.keller@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The HW must be programmed differently for queue-based scheduling mode.
To program the completion queue context correctly, the control plane
must know the scheduling mode not only for the Tx queue, but also for
the completion queue.
Unfortunately, currently the driver sets the scheduling mode only for
the Tx queues.
Propagate the scheduling mode data for the completion queue as
well when sending the queue configuration messages.
Fixes: 1c325aac10 ("idpf: configure resources for TX queues")
Reviewed-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Michal Kubiak <michal.kubiak@intel.com>
Reviewed-by: Alan Brady <alan.brady@intel.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Tested-by: Krishneil Singh <krishneil.k.singh@intel.com>
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Link: https://lore.kernel.org/r/20231023202655.173369-2-jacob.e.keller@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Add support for SRIOV: send the requested number of VFs
to the device Control Plane, via the virtchnl message
and then enable the VFs using 'pci_enable_sriov'.
Add other ndo ops supported by the driver such as features_check,
set_rx_mode, validate_addr, set_mac_address, change_mtu, get_stats64,
set_features, and tx_timeout. Initialize the statistics task which
requests the queue related statistics to the CP. Add loopback
and promiscuous mode support and the respective virtchnl messages.
Finally, add documentation and build support for the driver.
Signed-off-by: Joshua Hay <joshua.a.hay@intel.com>
Co-developed-by: Alan Brady <alan.brady@intel.com>
Signed-off-by: Alan Brady <alan.brady@intel.com>
Co-developed-by: Madhu Chittim <madhu.chittim@intel.com>
Signed-off-by: Madhu Chittim <madhu.chittim@intel.com>
Co-developed-by: Phani Burra <phani.r.burra@intel.com>
Signed-off-by: Phani Burra <phani.r.burra@intel.com>
Reviewed-by: Sridhar Samudrala <sridhar.samudrala@intel.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Co-developed-by: Pavan Kumar Linga <pavan.kumar.linga@intel.com>
Signed-off-by: Pavan Kumar Linga <pavan.kumar.linga@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Initialize all the ethtool ops that are supported by the driver and
add the necessary support for the ethtool callbacks. Also add
asynchronous link notification virtchnl support where the device
Control Plane sends the link status and link speed as an
asynchronous event message. Driver report the link speed on
ethtool .idpf_get_link_ksettings query.
Introduce soft reset function which is used by some of the ethtool
callbacks such as .set_channels, .set_ringparam etc. to change the
existing queue configuration. It deletes the existing queues by sending
delete queues virtchnl message to the CP and calls the 'vport_stop' flow
which disables the queues, vport etc. New set of queues are requested to
the CP and reconfigure the queue context by calling the 'vport_open'
flow. Soft reset flow also adjusts the number of vectors associated to a
vport if .set_channels is called.
Signed-off-by: Alan Brady <alan.brady@intel.com>
Co-developed-by: Alice Michael <alice.michael@intel.com>
Signed-off-by: Alice Michael <alice.michael@intel.com>
Co-developed-by: Joshua Hay <joshua.a.hay@intel.com>
Signed-off-by: Joshua Hay <joshua.a.hay@intel.com>
Co-developed-by: Phani Burra <phani.r.burra@intel.com>
Signed-off-by: Phani Burra <phani.r.burra@intel.com>
Reviewed-by: Sridhar Samudrala <sridhar.samudrala@intel.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Co-developed-by: Pavan Kumar Linga <pavan.kumar.linga@intel.com>
Signed-off-by: Pavan Kumar Linga <pavan.kumar.linga@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
