Bind RX/TX queues and MSI-X vectors to matching CPUs based on the RSS
bucket entries.
Introduce sysctls for the following RSS functionality:
- rss.indir_table: indirection table mapping
- rss.indir_table_size: indirection table size
- rss.key: RSS hash key (if Toeplitz used)
Said sysctls are only available when compiled without `option RSS`, as
kernel-side RSS support currently doesn't offer RSS reconfiguration.
Migrate the hash algorithm from CRC32 to Toeplitz and change the initial
hash value to 0x0 in order to match the standard Toeplitz implementation.
Provide helpers for hash key inversion required for HW operations.
Obtained from: Semihalf
MFC after: 2 weeks
Sponsored by: Amazon, Inc.
Provide the following sysctl statistics in order to stay aligned with
the Linux driver:
* rx_ring.csum_good
* tx_ring.unmask_interrupt_num
Also rename the 'bad_csum' statistic name to 'csum_bad' for alignment.
Obtained from: Semihalf
MFC after: 2 weeks
Sponsored by: Amazon, Inc.
In order to use `ena_global_lock` in sysctl context, it must be kept
outside the driver instance's software context, as sysctls can be called
before attach and after detach, leading to lock use before sx_init and
after sx_destroy otherwise.
Solve this issue by turning `ena_global_lock` into a file scope
variable, shared between all instances of the driver and associated
sysctl context, and in turn initialized/destroyed in dedicated
SYSINIT/SYSUNINIT functions.
As a side effect, this change also fixes existing race in the reset
routine, when simultaneously accessing sysctl exposed properties.
Obtained from: Semihalf
MFC after: 2 weeks
Sponsored by: Amazon, Inc.
If LLQ is being used, `ena_tx_ctx.meta_valid` must stay enabled. This
fixes netmap support on latest generation ENA HW and aligns it with the
core driver behavior.
As netmap doesn't support any csum offloads, the
`adapter->disable_meta_caching` value can be simply passed to the HW.
Obtained from: Semihalf
MFC after: 2 weeks
Sponsored by: Amazon, Inc.
Delegate RSS related functionality into separate .c/.h files in
preparation for the full RSS support.
While at it, reorder functions and remove prototypes for ones with
internal linkage.
Obtained from: Semihalf
MFC after: 2 weeks
Sponsored by: Amazon, Inc.
ENA silently assumed that ena_up, ena_down and ena_start_xmit routines
should be called within locked context. Driver's logic heavily assumes
on concurrent access to those routines, so for safety and better
documentation about this assumption, the locking assertions were added
to the above functions.
The assertion was added only for the main steps (skipping the helper
functions) which can be called from multiple places including the kernel
and the driver itself.
Obtained from: Semihalf
MFC after: 2 weeks
Sponsored by: Amazon, Inc.
Stay aligned with the Linux driver by adding the following logs:
* inform the user about retrying queue creation
* warn on non-empty ena_tx_buffer.mbuf prior to ena_tx_map_mbuf
Obtained from: Semihalf
MFC after: 2 weeks
Sponsored by: Amazon, Inc.
Check for ENA_FLAG_TRIGGER_RESET inside a locked context in order to
avoid potential race conditions with ena_destroy_device. This aligns the
reset task logic with the Linux driver.
Obtained from: Semihalf
MFC after: 2 weeks
Sponsored by: Amazon, Inc.
All ena_com_prepare_tx errors other than ENA_COM_NO_MEM are fatal and
require device reset.
Obtained from: Semihalf
MFC after: 2 weeks
Sponsored by: Amazon, Inc.
In case of Low-latency Queue, one small enough descriptor can be pushed
directly to the ENA hw, thus saving one fragment. Check for this
condition before performing collapse.
Obtained from: Semihalf
MFC after: 2 weeks
Sponsored by: Amazon, Inc.
Some of the changes in this release:
* Large LLQ headers,
* Bug/stability fixes,
* Change of the README/Documentation.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
MFC after: 2 weeks
Sponsored by: Amazon, Inc.
Update the driver in order not to break its compilation
and make use of the new ENA logging system
Migrate platform code to the new logging system provided by ena_com
layer.
Make ENA_INFO the new default log level.
Remove all explicit use of `device_printf`, all new logs requiring one
of the log macros to be used.
IO queue related attributes are registered statically at driver attach
with the rest of the ENA specific sysctl nodes. However, the number of
queues can be changed at runtime via the `ena_sysctl_io_queues_nb`
request, leading to a potential exposure of attributes for non-existing
queues.
Introduce a new `ena_sysctl_update_queue_node_nb` function, which
updates the sysctl nodes after the number of queues is altered.
This happens by either registering or unregistering node specific oids,
based on a delta between the previous and current queue count.
NOTE: All unregistered oids must be registered again before the driver
detach, e.g. by another call to this function.
Submitted by: Artur Rojek <ar@semihalf.com>
Obtained from: Semihalf
MFC after: 2 weeks
Sponsored by: Amazon, Inc.
Calling free on a NULL pointer is valid, as appropriate check is already
done internally:
/* free(NULL, ...) does nothing */
if (addr == NULL)
return;
Submitted by: Artur Rojek <ar@semihalf.com>
Obtained from: Semihalf
MFC after: 2 weeks
Sponsored by: Amazon, Inc.
Default LLQ (Low-latency queue) maximum header size is 96 bytes and can
be too small for some types of packets - like IPv6 packets with multiple
extension. This can be fixed, by using large LLQ headers.
If the device supports larger LLQ headers, the user can activate this
feature by setting sysctl tunable 'hw.ena.force_large_llq_header' to '1'
in the /boot/loader.conf file.
In case the device isn't supporting this feature, the default value (96B)
will be used.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
MFC after: 2 weeks
Sponsored by: Amazon, Inc.
According to man style(9), only C-style comments should be used.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
MFC after: 2 weeks
Sponsored by: Amazon, Inc.
In the new ENA-based instances like c6gn, the vector table moved to a
new PCIe bar - BAR1. Previously it was always located on the BAR0, so
the resources were already allocated together with the registers.
As the FreeBSD isn't doing any resource allocation behind the scenes,
the driver is responsible to allocate them explicitly, before other
parts of the OS (like the PCI code allocating MSIx) will be able to
access them.
To determine dynamically BAR on which the MSIx vector table is present
the pci_msix_table_bar() is being used and the new BAR is allocated if
needed.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc
MFC after: 3 days
Some of the PCI ID were described as ENA with LLQ support - it's not
fully accurate and because of that, their names were changed.
Instead of LLQ, use RSERV0 for the description of those devices.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc
MFC after: 1 week
Differential revision: https://reviews.freebsd.org/D27119
The new HAL allows the driver to read extra ENI stats. Exact meaning of
each of them can be found in base/ena_defs/ena_admin_defs.h file and
structure ena_admin_eni_stats.
Those stats are being updated inside of the timer service, which is
executed every second.
ENI metrics are turned off by default. They can be enabled, using the
sysctl node: dev.ena.X.eni_metrics.update_delay
0 value in this node means that the update is turned off. Other values
determine how many seconds must pass, before ENI metrics will be
updated.
They can be acquired, using sysctl:
sysctl dev.ena.X.eni_metrics
Where X stands for the interface number.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc
MFC after: 1 week
Differential revision: https://reviews.freebsd.org/D27118
Refering to guide: https://wiki.freebsd.org/SPDX the SPDX tag should not
replace the standard license text, however it should be added over the
standard license text to make the automation easier.
Because of that, the old license was kept, but the SPDX tag was added
on top of every ENA driver file.
Submited by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc
MFC after: 1 week
Differential revision: https://reviews.freebsd.org/D27117
For the first descriptor in a chain the data may start at an offset.
It is optional feature of some devices, so the driver must ack that
it supports it.
The data pointer of the mbuf is simply shifted by the given value.
Submitted by: Maciej Bielski <mba@semihalf.com>
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc
MFC after: 1 week
Differential revision: https://reviews.freebsd.org/D27116
* Use the new API of ena_trace_*
* Fix typo syndrom --> syndrome
* Remove validation of the Rx req ID (already performed in the ena-com)
* Remove usage of deprecated ENA_ASSERT macro
Submitted by: Ido Segev <idose@amazon.com>
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc
MFC after: 1 week
Differential revision: https://reviews.freebsd.org/D27115
The latest generation hardware requires IO CQ (completion queue)
descriptors memory to be aligned to a 4K. It needs that feature for
the best performance.
Allocating unaligned descriptors will have a big performance impact as
the packet processing in a HW won't be optimized properly. For that
purpose adjust ena_dma_alloc() to support it.
It's a critical fix, especially for the arm64 EC2 instances.
Submitted by: Ido Segev <idose@amazon.com>
Obtained from: Amazon, Inc
MFC after: 1 week
Differential revision: https://reviews.freebsd.org/D27114
Networking is broken if the driver configures its (virtual) hardware to
use a hash algorithm (or a key) different from the one that the network
stack (software RSS) uses. This can be seen with connections initiated
from the host. The PCB will be placed into the hash table based on the
hash value calculated by the software. The hardware-calculated hash
value in reponse packets will be different, so the PCB won't be found.
Tested with a kernel compiled with 'options RSS' on an instance with ena
driver.
Reviewed by: mw, adrian
MFC after: 2 weeks
Sponsored by: Panzura
Differential Revision: https://reviews.freebsd.org/D24733
Driver version upgrade is connected with support for the new device
fetures, like Tx drops reporting or disabling meta caching.
Moreover, the driver configuration from the sysctl was reworked to
provide safer and better flow for configuring:
* number of IO queues (new feature),
* drbr size on Tx,
* Rx queue size.
Moreover, a lot of minor bug fixes and improvements were added.
Copyright date in the license of the modified files in this release was
updated to 2020.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
There is no guarantee from bus_dmamap_load_mbuf_sg() for matching
mbuf chain segments to dma physical segments.
This patch ensure correctly mapping to LLQ header and DMA segments.
Submitted by: Ido Segev <idose@amazon.com>
Obtained from: Amazon, Inc.
There is ena_free_all_io_rings_resources() called twice on device
detach:
ena_detach():
ena_destroy_device():
/* First call */
ena_free_all_io_rings_resources()
/* Second call */
ena_free_all_io_rings_resources()
The double-free causes panic() on kldunload, for example.
As the ena_destroy_device() is also called by ena_reset_task() it is
better to stay unchanged. Thus, remove the "Second call" of the function.
Submitted by: Maciej Bielski <mba@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
Determined by a flag passed from the device. No metadata is set within
ena_tx_csum when caching is disabled.
Submitted by: Maciej Bielski <mba@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
If requested size of IO queues is not supported try to decrease it until
finding the highest value that can be satisfied.
Submitted by: Maciej Bielski <mba@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
By default, in ena_attach() the driver attempts to acquire
ena_adapter::max_num_io_queues MSI-X vectors for the purpose of IO
queues, however this is not guaranteed. The number of vectors acquired
depends also on system resources availability.
Regardless of that, enable the number of effectively used IO queues to
be further limited through the sysctl node.
Example: Assumming that there are 8 IO queues configured by default, the
command
$ sysctl dev.ena.0.io_queues_nb=4
will reduce the number of available IO queues to 4. Similarly, the value
can be also increased up to maximum supported value. A value higher than
maximum supported number of IO queues is ignored. Zero is ignored too.
Submitted by: Maciej Bielski <mba@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
Make the ena_adapter::num_io_queues a number of effectively used IO
queues. While the ena_adapter::max_num_io_queues is an upper-bound
specified by the HW, the ena_adapter::num_io_queues may be lower than
that, depending on runtime system resources availability.
On reset, there are called ena_destroy_device() and then
ena_restore_device(). The latter calls, in turn, ena_enable_msix(),
which will attempt to re-acquire ena_adapter::max_num_io_queues of
MSIX vectors again.
Thus, the value of ena_adapter::num_io_queues may be different before
and after reset. For this reason, free the IO rings structures (drbr,
counters) in ena_destroy_device() and allocate again in
ena_restore_device().
Submitted by: Maciej Bielski <mba@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
This method has been aligned with the way how the Rx queue size is being
updated - so it's now done synchronously instead of resetting the
device.
Moreover, the input parameter is now being validated if it's a power of
2. Without this, it can cause kernel panic.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
This patch reworks how the Rx queue size is being reconfigured and how
the information from the device is being processed.
Reconfiguration of the queues and reset of the device in order to make
the changes alive isn't the best approach. It can be done synchronously
and it will let to pass information if the reconfiguration was
successful to the user. It now is done in the ena_update_queue_size()
function.
To avoid reallocation of the ring buffer, statistic counters and the
reinitialization of the mutexes when only new size has to be assigned,
the io queues initialization function has been split into 2 stages:
basic, which is just copying appropriate fields and the advanced, which
allocates and inits more advanced structures for the IO rings.
Moreover, now the max allowed Rx and Tx ring size is being kept
statically in the adapter and the size of the variables holding those
values has been changed to uint32_t everywhere.
Information about IO queues size is now being logged in the up routine
instead of the attach.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
Recent changes to the epoch requires driver to notify that they knows
epoch in order to prevent input packet function to enter epoch each
time the packet is received.
ENA is using NET_TASK for handling Rx, so it's entering epoch
automatically whenever this task is being executed.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
If the conditional check for ENA_FLAG_DEV_UP is negated, the body of the
function can have smaller indentation and it makes the code cleaner.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
As functions which are declared in the header files are intended to be
the interface and are going to be used by other files, it's better to
include argument names in the definition, so the caller won't have to
check the .c file in order to check their meaning and order.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
Currently, the driver had 2 global locks - one was sx lock used for
up/down synchronization and the second one was mutex, which was used
for link configuration and timer service callout.
It is better to have single lock for that. We cannot use mutex, as it
can sleep and cause witness errors in up/down configuration, so sx lock
seems to be the only choice.
Callout cannot use sx lock, but the timer service is MP safe, so we just
need to avoid race between ena_down() and ena_detach(). It can be
avoided by acquiring sx lock.
Simple macros were added that are encapsulating implementation of the
lock and makes the code cleaner.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
As the reset triggering is no longer a simple macro that was just
setting appropriate flag, the new function for triggering reset was
added. It improves code readability a lot, as we are avoiding additional
indentation.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
The function ena_enable_msix_and_set_admin_interrupts takes two
arguments while the second is not used and so can be spared. This is a
static function, only ena.c is affected.
Submitted by: Maciej Bielski <mba@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
Tx drops statistics are fetched from HW every ena_keepalive_wd() call
and are observable using one of the commands:
* sysctl dev.ena.0.hw_stats.tx_drops
* netstat -I ena0 -d
Submitted by: Maciej Bielski <mba@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
* Removed adaptive interrupt moderation (not suported on FreeBSD).
* Use ena_com_free_q_entries instead of ena_com_free_desc.
* Don't use ENA_MEM_FREE outside of the ena_com.
* Don't use barriers before calling doorbells as it's already done in
the HAL.
* Add function that generates random RSS key, common for all driver's
interfaces.
* Change admin stats sysctls to U64.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
Sometimes, especially when there is not much memory in the system left,
allocating mbuf jumbo clusters (like 9KB or 16KB) can take a lot of time
and it is not guaranteed that it'll succeed. In that situation, the
fallback will work, but if the refill needs to take a place for a lot of
descriptors at once, the time spent in m_getjcl looking for memory can
cause system unresponsiveness due to high priority of the Rx task. This
can also lead to driver reset, because Tx cleanup routine is being
blocked and timer service could detect that Tx packets aren't cleaned
up. The reset routine can further create another unresponsiveness - Rx
rings are being refilled there, so m_getjcl will again burn the CPU.
This was causing NVMe driver timeouts and resets, because network driver
is having higher priority.
Instead of 16KB jumbo clusters for the Rx buffers, 9KB clusters are
enough - ENA MTU is being set to 9K anyway, so it's very unlikely that
more space than 9KB will be needed.
However, 9KB jumbo clusters can still cause issues, so by default the
page size mbuf cluster will be used for the Rx descriptors. This can have a
small (~2%) impact on the throughput of the device, so to restore
original behavior, one must change sysctl "hw.ena.enable_9k_mbufs" to
"1" in "/boot/loader.conf" file.
As a part of this patch (important fix), the version of the driver
was updated to v2.1.2.
Submitted by: cperciva
Reviewed by: Michal Krawczyk <mk@semihalf.com>
Reviewed by: Ido Segev <idose@amazon.com>
Reviewed by: Guy Tzalik <gtzalik@amazon.com>
MFC after: 3 days
PR: 225791, 234838, 235856, 236989, 243531
Differential Revision: https://reviews.freebsd.org/D24546
r357614 added CTLFLAG_NEEDGIANT to make it easier to find nodes that are
still not MPSAFE (or already are but aren’t properly marked).
Use it in preparation for a general review of all nodes.
This is non-functional change that adds annotations to SYSCTL_NODE and
SYSCTL_PROC nodes using one of the soon-to-be-required flags.
Mark all obvious cases as MPSAFE. All entries that haven't been marked
as MPSAFE before are by default marked as NEEDGIANT
Approved by: kib (mentor, blanket)
Commented by: kib, gallatin, melifaro
Differential Revision: https://reviews.freebsd.org/D23718
Driver working in LLQ mode in some cases can send only few last segments
of the mbuf using DMA engine, and the rest of them are sent to the
device using direct PCI transaction. To map the only necessary data, two DMA
maps were used. That solution was very rough and was causing a bug - if
both maps were used (head_map and seg_map), there was a race in between
two flows on two queues and the device was receiving corrupted
data which could be further received on the other host if the Tx cksum
offload was enabled.
As it's ok to map whole mbuf and then send to the device only needed
segments, the design was simplified to use only single DMA map.
The driver version was updated to v2.1.1 as it's important bug fix.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
MFC after: 2 weeks
Sponsored by: Amazon, Inc.
In this release the netmap support was introduced.
Moreover, it is also now possible to use the LLQ mode of the driver on
the arm64 AWS instances (A1 type).
Differential Revision: https://reviews.freebsd.org/D21938
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
In NETMAP mode not all queues need to be allocated to NETMAP. Some of
them could be left to the kernel. Configuration is managed by the flags
nr_mode and nr_pending_mode provided per each NETMAP kring.
ENA driver checks those flags and perform proper rings initialization.
Differential Revision: https://reviews.freebsd.org/D21937
Submitted by: Rafal Kozik <rk@semihalf.com>
Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
Two new tables are added to ena_tx_buffer structure:
* netmap_map_seg stores DMA mapping structures,
* netmap_buf_idx stores buff indexes taken from the slots.
When Tx resources are being set, the new mapping structures are created
and netmap Tx rings are being reset.
When Tx resources are being released, used netmap bufs are unmapped from
DMA and then mapping structures are destroyed.
When Tx interrupt occurrs, ena_netmap_tx_irq is called.
ena_netmap_txsync callback signalizes that there are new packets which
should be transmitted.
First, it fills ena_netmap_ctx. Then it performs two actions:
* ena_netmap_tx_frames moves packets from netmap ring to NIC,
* ena_netmap_tx_cleanup restores buffers from NIC and gives them back
to the userspace app.
0 is returned in case of Tx error that could be handled by the driver.
ena_netmap_tx_frames checks if there are packets ready for transmission.
Then, for each of them, ena_netmap_tx_frame is called. If error occurs,
transmitting is stopped, but if the error was cause due to HW ring being
full, information about that is not propagated to the userspace app.
When all packets are ready, doorbell is written to NIC and netmap ring
state is updated.
Parsing of one packet is done by the ena_netmap_tx_frame function.
First, it checks if number of slots does not exceed NIC limit. Invalid
packets are being dropped and the error is propagated to the upper
layer. As each netmap buffer has equal size, which is typically greater
then 2KiB, there shouldn't be any packets which contain too many slots.
Then, the ena_com_tx_ctx structure is being filled. As netmap does not
support any hardware offloads, ena_com_tx_meta structure is set to zero.
After that, ena_netmap_map_slots maps all memory slots for DMA.
If the device works in the LLQ mode, the push header is being determined
by checking if the header fits within the first socket.
If so, the portion of data is being copied directly from the slot.
In other case, the data is copied to the intermediate buffer.
First slots are treated the same as as the others, because DMA mapping
has no impact on LLQ mode. Index of each netmap buffer is taken from
slot and stored in netmap_buf_idx array. In case of mapping error,
memory is unmapped and packets are put back to the netmap ring.
ena_netmap_tx_cleanup performs out of order cleanup of sent buffers.
First, req_id is taken and is validated. As validate_tx_req_id from
ena.c is specific to kernels mbuf, another implementation is provided.
Each req_id is cleaned up by ena_netmap_tx_clean_one function. Buffers
are being unmaped from DMA and put back to netmap ring. In the end,
state of netmap and NIC rings are being updated.
Differential Revision: https://reviews.freebsd.org/D21936
Submitted by: Rafal Kozik <rk@semihalf.com>
Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
Most of code used for Rx ring initialization could be reused in NETMAP.
Reset of NETMAP ring and new alloc method was added. Driver decides if
use kernels mbufs or NETMAPs slots based on IFCAP_NETMAP flag. It
allows to reuse ena_refill_rx_bufs, which provides proper handling of
Rx out of order completion.
ena_netmap_alloc_rx_slot takes exactly the same arguments as
ena_alloc_rx_mbuf, but instead of allocating one mbuf it takes one slot
from NETMAP ring. Based on queue id proper netmap_ring is found. As
NETMAP provides the "partial opening" feature not all of the rings are
avaiable. Not used points to invalid ring. If there is available slot,
it is taken from the ring. Its buffer is mapped to DMA and its index is
stored in ena_rx_buffer field in ena_rx_buffer structure. Then ena_buf
is filled with addresses and ring state is updated.
Cleanup is handled by ena_netmap_free_rx_slot. It unmaps DMA and returns
buffer to ring. As we could not return more bufs than we have taken and
we should not override occupied slots, buf_index should be 0. It is
being checked by assertion.
ena_netmap_rxsync callback puts received packets back to NETMAP ring and
passes them to user space by updating ring pointers. First it fills
ena_netmap_ctx.
Then it performs two actions:
* ena_netmap_rx_frames moves received frames from NIC to NETMAP ring,
* ena_netmap_rx_cleanup fills NIC ring with slots released by userspace
app.
In case of Rx error that could be handled by NIC driver (for example by
performing reset) rx sync should return 0.
ena_netmap_rx_frames first checks if NETMAP ring is in consistent
state and then in the loop receives new frames. When all available
frames are taken nr_hwtail is updated.
Receiving one frame is handled by ena_netmap_rx_frame. If no error
occurrs, each Descriptor is loaded by ena_netmap_rx_load_desc function.
If packets take more than one segments NS_MOREFRAG flag must be set in
all, but not last slot. In case of wrong req_id packet is removed from
NETMAP ring. If packet is successful received counters are updated.
Refiling of NIC ring is performed by ena_netmap_rx_cleanup function.
It calculates number of available slots and call ena_refill_rx_bufs with
proper number.
Differential Revision: https://reviews.freebsd.org/D21935
Submitted by: Rafal Kozik <rk@semihalf.com>
Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
Mock implementation of NETMAP routines is located in ena_netmap.c/.h
files. All code is protected under the DEV_NETMAP macro. Makefile was
updated with files and flag.
As ENA driver provide own implementations of (un)likely it must be
undefined before including NETMAP headers.
ena_netmap_attach function is called on the end of NIC attach. It fills
structure with NIC configuration and callbacks. Then provides it to
netmap_attach. Similarly netmap_detach is called during ena_detach.
Three callbacks are used.
nm_register is implemented by ena_netmap_reg. It is called when user
space application open or close NIC in NETMAP mode. Current action is
recognized based on onoff parameter: true means on and false off. As
NICs rings need to be reconfigured ena_down and ena_up are reused.
When user space application wants to receive new packets from NIC
nm_rxsync is called, and when there are new packets ready for Tx
nm_txsync is called.
Differential Revision: https://reviews.freebsd.org/D21934
Submitted by: Rafal Kozik <rk@semihalf.com>
Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
Move Rx/Tx routines to separate file.
Some functions:
* ena_restore_device,
* ena_destroy_device,
* ena_up,
* ena_down,
* ena_refill_rx_bufs
could be reused in upcoming netmap code in the driver. To make it
possible, they were moved to ena.h header.
Differential Revision: https://reviews.freebsd.org/D21933
Submitted by: Rafal Kozik <rk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
When the ENA_FLAG_DEVICE_RUNNING flag is disabled, the AENQ handlers
aren't executed. To fix that, the watchdog timestamp should be updated
just before enabling the watchdog.
Timer service was always being enabled, even if the device wasn't up
before the reset. That shouldn't happen, as the timer service is being
executed only for working interface.
Differential Revision: https://reviews.freebsd.org/D21932
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
As the pmamp_change_attr() is public on arm64 since r351131, it can be
used on the arm64 to map memory range as with the write combined
attribute.
It requires the driver to use generic VM_MEMATTR_WRITE_COMBINING flag
instead of the x86 specific PAT_WRITE_COMBINING.
Differential Revision: https://reviews.freebsd.org/D21931
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
ENAv2 introduces many new features, bug fixes and improvements.
Main new features are LLQ (Low Latency Queues) and independent queues
reconfiguration using sysctl commands.
The year in copyright notice was updated to 2019.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
For easier debugging, the reset is being triggered and the reset reason is
being set only in case it is done for the first time. Such approach will
ensure that the first reset reason is not going to be overwritten and
will make it easier for debugging.
Also, add a reset trigger upon invalid Tx requested ID.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
If the call to ena_up() in ena_restore_device() fails, next usage of
`ifconfig up` will cause NULL pointer dereference.
This patch adds additional checks to prevent that.
Submitted by: Rafal Kozik <rk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
Some messages were missing new line character and traces were not having
unified behavior. To fix that, each trace and printout should add new
line character at the end of each string - that should improve
readability.
Submitted by: Rafal Kozik <rk@semihalf.com>
Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
If the headers of the packets are split into multiple segments of the
mbuf chain, the previous version of ena_tx_csum which was assuming,
that all segments will lay in the first mbuf, will eventually fail to
map the headers properties to meta descriptor.
That will cause Tx checksum offload to do not work and was leading to
memory corruption. It could even cause the crash of the system.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
For alignment with Linux driver and better handling ena_detach(), the
reset is now calling ena_device_restore() and ena_device_destroy().
The ena_device_destroy() is also being called on ena_detach(), so the
code will be more readable.
The watchdog is now being activated after reset only, if it was active
before.
There were added additional checks to ensure, that there is no race with
the link state change AENQ handler.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
As the ENA can have multiple states turned on/off, it is more convenient
to store them in single bitfield instead of multiple boolean variables.
The bitset FreeBSD API was used for the bitfield implementation, as it
provides flexible structure together with API which also supports atomic
bitfield operations.
For better readability basic macros from API were wrapped into custom
ENA_FLAG_* macros, which are filling up common parameters for all calls.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
Before the patch, error handling was not releasing all resources and
was not issuing device reset if the reset task failed.
That could cause memory leak and fault of the device.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
The host info bdf field is the abbreviation for the bus, device,
function of the PCI on which the device is being attached to.
Now the driver is filling information about that using FreeBSD RID
resource.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
The new ENA HAL is introducing API, which can determine on Tx path if
the doorbell is needed.
That way, it can tell the driver, that it should call an doorbell.
The old threshold value wasn't removed, as not all HW is supporting this
feature - so it was reworked to also work with the new API.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
The Rx ring size can be as high as 8k. Because of that we want to limit
the cleanup threshold by maximum value of 256.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
LLQ (Low Latency Queue) is the feature, that allows pushing header
directly to the device through PCI before even DMA is triggered.
It reduces latency, because device can start preparing packet before
payload is sent through DMA.
To speed up sending data through PCI, the Write Combining is enabled,
which allows hardware to buffer data before sending them on the PCI - it
allows to reduce number of PCI IO operations.
ENAv2 is using special descriptor for the negotiation of the LLQ.
Currently, only the default configuration is supported.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
Handle IO interrupts using filter routine. That way, the main cleanup
task could be moved to the separate thread using taskqueue.
The deferred Rx cleanup task was removed, and now the cleanup task is
begin called instead. That way, the Rx lock could be removed.
In addition, Queue management (wake up and stop TX ring) was added, so
the TX cleanup task can be performed mostly lockless.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
The driver now supports per adapter tuning of buffer ring size and HW Rx
ring size.
It can be achieved using sysctl node dev.ena.X.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
If the requested ID was out of range, the tx_info structure was NULL and
the function was trying to access the field of the NULL object.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
The if_detach was causing crash if the MSI-x configuration in the attach
failed. To prevent this issue, the ifnet is being configured at the end
of the attach function.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
In rare case, when the ifconfig is called just before kldunload, it is
possible, that ena_up routine will be called after queue locks are
released.
To prevent that, ifp is detached before the last ena_down is called and
further, the ifp is freed at the end of the function.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
The ENA driver needs at least 2 MSI-x - one for admin queue, and one for
IO queues pair. If there were not enough resources to allocate more than
one MSI-x, the device should not be attached.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
bus_alloc_resource_any() is not returning error value in case of an
error.
If the function call fails, the error value was not passed to the
ena_up() function.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
To prevent errors from assigning values from the DMA structure in case
of an error, zero the vaddr and paddr values upon failure.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
The DMA in FreeBSD requires explicit synchronization. ENA driver was
only doing PREREAD and PREWRITE synchronizations. Missing
bus_dmamap_sync() calls were added.
It is also required to synchronize DMA engine before unloading DMA map.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
If the first MSI-x won't be executed, then the timer service will detect
that and trigger device reset.
The checking for missing Tx completion was reworked, so it will also
check for missing interrupts. Checking number of missing Tx completions
can be performed after loop, instead of checking it every iteration.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
The new ena_com allows the number of CPUs to be passed to the device in
the host info structure as a hint.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
Information about Tx error should be only displayed, if packet
preparation failed due to error other than out of memory.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
Whenever the driver will receive too many descriptors from the device,
it should trigger the device reset, as it is indicating that the device
is in invalid state.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
Receive Side Scaling is optional feature that could be enabled in kernel
configuration by defining flag RSS.
Kernel uses hash to store and find protocol control block which is
stored in hash tables.
Kernel and NIC hash functions must be consistent. Otherwise case lookup
fails.
To achieve this kernel provides API to set proper hash key to NIC.
As it is not possible to change key for virtual ENA NIC, this driver
cannot support RSS function.
ENA is designed to work in virtual environments so supporting hardware
version of this card is unnecessary.
Submitted by: Rafal Kozik <rk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
Notification AENQ handler is responsible for handling requests from ENA
device. Missing Tx threshold, Tx timeout and keep alive timeout can be
set using hints from the aenq descriptor which can be delivered in the
ENA admin notification.
The queue suspending and resuming tasks are not supported by the
driver.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
ENA_ADMIN_FATAL_ERROR and ENA_ADMIN_WARNING aenq groups were indicated
as supported, so the unimplemented_aenq_handler() will print out error
message, whenever an error will occur within the ENA admin context.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
As the ENA is working only in virtualized environment, the active media
is not specified. Instead, the active link type is set as unknown.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
Recent HAL change preparing to support ENAv2 required minor driver
modifications.
The ena_com_sq_empty_space() is not available in this ena-com, so it had
to be replaced with ena_com_free_desc().
Moreover, the ena_com_admin_init() is no longer using 3rd argument
indicating if the spin lock should be initialized, so it was removed.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
The resource is already being activated in the bus_alloc_resource(),
because the flag RF_ACTIVE is being passed.
Double activation on arm64 is causing kernel panic.
Version of the driver was upgraded to 0.8.4.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Reported-by: Greg V <greg@unrelenting.technology>
Tested-by: cperciva, Greg V <greg@unrelenting.technology>
Obtained from: Semihalf
MFC after: 2 weeks
Sponsored by: Amazon, Inc.
Differential revision: https://reviews.freebsd.org/D19655
In out of order mode Rx buffer are accesses by req_id.
Accessing and validating mbuf using ntc is causing false error.
Increase driver revision after latest RX OOO completion fixes.
Submitted by: Rafal Kozik <rk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
MFC after: 1 week
Requested ID should be validated when the packet is received and not
when the driver is repopulating the mbufs.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
MFC after: 1 week
In FreeBSD, this is normal situation that the Tx ring is being full. In
hat case, the packet is put back into drbr and the next attempt to send
it is taken after the cleanup.
Too much logs like this can cause system instability and even cause the
device reset (because keep alive or cleanup could be missed).
To fix that, the log level of this message is changed to debug.
Upon this change upgrade the driver version to v0.8.2.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
Remove unused and easy to misuse PNP macro parameter
Inspired by r338025, just remove the element size parameter to the
MODULE_PNP_INFO macro entirely. The 'table' parameter is now required to
have correct pointer (or array) type. Since all invocations of the macro
already had this property and the emitted PNP data continues to include the
element size, there is no functional change.
Mostly done with the coccinelle 'spatch' tool:
$ cat modpnpsize0.cocci
@normaltables@
identifier b,c;
expression a,d,e;
declarer MODULE_PNP_INFO;
@@
MODULE_PNP_INFO(a,b,c,d,
-sizeof(d[0]),
e);
@singletons@
identifier b,c,d;
expression a;
declarer MODULE_PNP_INFO;
@@
MODULE_PNP_INFO(a,b,c,&d,
-sizeof(d),
1);
$ rg -l MODULE_PNP_INFO -- sys | \
xargs spatch --in-place --sp-file modpnpsize0.cocci
(Note that coccinelle invokes diff(1) via a PATH search and expects diff to
tolerate the -B flag, which BSD diff does not. So I had to link gdiff into
PATH as diff to use spatch.)
Tinderbox'd (-DMAKE_JUST_KERNELS).
Approved by: re (glen)
I was not aware Warner was making or planning to make forward progress in
this area and have since been informed of that.
It's easy to apply/reapply when churn dies down.
Inspired by r338025, just remove the element size parameter to the
MODULE_PNP_INFO macro entirely. The 'table' parameter is now required to
have correct pointer (or array) type. Since all invocations of the macro
already had this property and the emitted PNP data continues to include the
element size, there is no functional change.
Mostly done with the coccinelle 'spatch' tool:
$ cat modpnpsize0.cocci
@normaltables@
identifier b,c;
expression a,d,e;
declarer MODULE_PNP_INFO;
@@
MODULE_PNP_INFO(a,b,c,d,
-sizeof(d[0]),
e);
@singletons@
identifier b,c,d;
expression a;
declarer MODULE_PNP_INFO;
@@
MODULE_PNP_INFO(a,b,c,&d,
-sizeof(d),
1);
$ rg -l MODULE_PNP_INFO -- sys | \
xargs spatch --in-place --sp-file modpnpsize0.cocci
(Note that coccinelle invokes diff(1) via a PATH search and expects diff to
tolerate the -B flag, which BSD diff does not. So I had to link gdiff into
PATH as diff to use spatch.)
Tinderbox'd (-DMAKE_JUST_KERNELS).
Make unsigned values uint16_t for pnp table. They are properly
uint16_t befause they are 16-bit PCI IDs. The PNP_INFO language has no
type for bare unsigned.
Reviewed by: imp, chuck
Submitted by: Lakhan Shiva Kamireddy <lakhanshiva@gmail.com>
Sponsored by: Google, Inc. (GSoC 2018)
Pull Request: https://github.com/bsdimp/freebsd/pull/5
Header length is optional hint for the ENA device. Because It is not
guaranteed that every packet header will be in the first mbuf
segment, it is better to skip passing any information. If the header
length will be indicating invalid value (different than 0), then the
packet will be dropped.
This kind situation can appear, when the UDP packet will be fragmented
by the stack in the ip_fragment() function.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Reported by: Krishna Yenduri <kyenduri@brkt.com>
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
On AWS, a network interface can get reinitialized every 30 minutes due
to the MTU being (re)set when a new DHCP lease is obtained. This can
cause packet drop, along with annoying syslog messages.
Skip setting the MTU in the ena driver if the new MTU is the same as the
old MTU. Note this fix is already in the netfront driver.
Testing: Verified ena up/down messages do not appear every 30 min in
/var/log/messages with the fix in place.
Submitted by: Krishna Yenduri <kyenduri@brkt.com>
Reviewed by: Michal Krawczyk <mk@semihalf.com>
With the current state of the AENQ handlers in the ENA driver, only
implemented handlers should be indicated.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Reviewed by: byenduri_gmail.com
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
Differential Revision: https://reviews.freebsd.org/D12872
Using only 1 descriptor on RX could be an issue, if system would be low
on resources and could not provide driver with large chunks of
contiguous memory.
Submitted by: Michal Krawczyk <mk@semihalf.com>
Reviewed by: byenduri_gmail.com
Obtained from: Semihalf
Sponsored by: Amazon, Inc.
Differential Revision: https://reviews.freebsd.org/D12871
