This patch adds telemetry as a dependecy to all applications. Without these
changes, the --telemetry flag will not be recognised and applications will
fail to run if they want to enable telemetry.
Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
Signed-off-by: Kevin Laatz <kevin.laatz@intel.com>
Signed-off-by: Radu Nicolau <radu.nicolau@intel.com>
Acked-by: Harry van Haaren <harry.van.haaren@intel.com>
This patch adds functionality to enable/disable the selftest.
This functionality will be extended in future to make the
enabling/disabling more dynamic and remove this 'hardcoded' approach. We
are temporarily using this approach due to the design changes (vdev vs eal)
made to the library.
Signed-off-by: Ciara Power <ciara.power@intel.com>
Signed-off-by: Brian Archbold <brian.archbold@intel.com>
Signed-off-by: Kevin Laatz <kevin.laatz@intel.com>
Acked-by: Harry van Haaren <harry.van.haaren@intel.com>
This patch adds functionality to create a JSON message in
order to send it to a client socket.
When stats are requested by a client, they are retrieved from
the metrics library and encoded in JSON format.
Signed-off-by: Ciara Power <ciara.power@intel.com>
Signed-off-by: Brian Archbold <brian.archbold@intel.com>
Signed-off-by: Kevin Laatz <kevin.laatz@intel.com>
Acked-by: Harry van Haaren <harry.van.haaren@intel.com>
This patch adds functionality to update the statistics in
the metrics library with values from the ethdev stats.
Values need to be updated before they are encoded into a JSON
message and sent to the client that requested them. The JSON encoding
will be added in a subsequent patch.
Signed-off-by: Ciara Power <ciara.power@intel.com>
Signed-off-by: Brian Archbold <brian.archbold@intel.com>
Signed-off-by: Kevin Laatz <kevin.laatz@intel.com>
Acked-by: Harry van Haaren <harry.van.haaren@intel.com>
This patch adds the parser file. This is used to parse any
messages that are received on any of the client sockets.
Currently, the unregister functionality works using the parser.
Functionality relating to getting statistic values for certain ports
will be added in a subsequent patch, however the parsing involved
for that command is added in this patch.
Some of the parser code included is in preparation for future
functionality, that is not implemented yet in this patchset.
Signed-off-by: Ciara Power <ciara.power@intel.com>
Signed-off-by: Brian Archbold <brian.archbold@intel.com>
Signed-off-by: Kevin Laatz <kevin.laatz@intel.com>
Acked-by: Harry van Haaren <harry.van.haaren@intel.com>
This patch introduces clients to the telemetry API.
When a client makes a connection through the initial telemetry
socket, they can send a message through the socket to be
parsed. Register messages are expected through this socket, to
enable clients to register and have a client socket setup for
future communications.
A TAILQ is used to store all clients information. Using this, the
client sockets are polled for messages, which will later be parsed
and dealt with accordingly.
Functionality that make use of the client sockets were introduced
in this patch also, such as writing to client sockets, and sending
error responses.
Signed-off-by: Ciara Power <ciara.power@intel.com>
Signed-off-by: Brian Archbold <brian.archbold@intel.com>
Signed-off-by: Kevin Laatz <kevin.laatz@intel.com>
Acked-by: Harry van Haaren <harry.van.haaren@intel.com>
This patch adds the telemetry UNIX socket. It is used to
allow connections from external clients.
On the initial connection from a client, ethdev stats are
registered in the metrics library, to allow for their retrieval
at a later stage.
Signed-off-by: Ciara Power <ciara.power@intel.com>
Signed-off-by: Brian Archbold <brian.archbold@intel.com>
Signed-off-by: Kevin Laatz <kevin.laatz@intel.com>
Acked-by: Harry van Haaren <harry.van.haaren@intel.com>
This patch adds the infrastructure and initial code for the telemetry
library.
The telemetry init is registered with eal_init(). We can then check to see
if --telemetry was passed as an eal option. If --telemetry was parsed, then
we call telemetry init at the end of eal init.
Control threads are used to get CPU cycles for telemetry, which are
configured in this patch also.
Signed-off-by: Ciara Power <ciara.power@intel.com>
Signed-off-by: Brian Archbold <brian.archbold@intel.com>
Signed-off-by: Kevin Laatz <kevin.laatz@intel.com>
Signed-off-by: Radu Nicolau <radu.nicolau@intel.com>
Acked-by: Harry van Haaren <harry.van.haaren@intel.com>
This patch makes the eal_get_runtime_dir() API public so it can be used
from outside EAL.
Signed-off-by: Kevin Laatz <kevin.laatz@intel.com>
Acked-by: Harry van Haaren <harry.van.haaren@intel.com>
This commit adds infrastructure to EAL that allows an application to
register it's init function with EAL. This allows libraries to be
initialized at the end of EAL init.
This infrastructure allows libraries that depend on EAL to be initialized
as part of EAL init, removing circular dependency issues.
Signed-off-by: Kevin Laatz <kevin.laatz@intel.com>
Acked-by: Harry van Haaren <harry.van.haaren@intel.com>
Acked-by: Gaetan Rivet <gaetan.rivet@6wind.com>
The offload name functions are useful, but since they are
marked experimental they can not be used by upstream projects.
For example, VPP duplicates the same table in its code.
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Add a new rte_delay_us_sleep() function that uses nanosleep().
This function can be used by applications to not implement
their own nanosleep() based callback and by internal DPDK
code if CPU non-blocking delay needed.
Signed-off-by: Ilya Maximets <i.maximets@samsung.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
The iterator was matching all representors if it was not specified
in the devargs string. It was a wrong default behaviour.
If there is no representor parameter in the devargs, the iterator
should not match any representor port.
The implementation of the default behaviour would be simpler
if a "no match" handler is added to rte_kvargs_process().
As it requires an API breakage, it will be reworked later.
Fixes: a7d3c6271d55 ("ethdev: support representor id as iterator filter")
Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
If a port is being created and rollbacked because of an error,
the event RTE_ETH_EVENT_DESTROY should not be sent.
It makes no sense to receive a destroy event for a port which
was not yet announced via RTE_ETH_EVENT_NEW.
Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Add the Mpls header structure in librte_net. It will be used by next
patch that adds the support of Mpls L2 layer in the software packet
type parser.
Signed-off-by: Olivier Matz <olivier.matz@6wind.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Currently, postcopy_ufd is initialized to 0 implicitly, so fd 0
could be closed unexpectedly by vhost_backend_cleanup(). Fix this
issue by initializing postcopy_ufd to -1 explicitly.
Fixes: 9eefef3b5970 ("vhost: introduce postcopy advise message")
Signed-off-by: Tiwei Bie <tiwei.bie@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
In both split and packed dequeue paths, flush_shadow_used_ring
and vhost_ring_call variants gets called even if not packets
have been dequeued, and so no descriptors updates happened.
It has an impact on CPU pipeline, as memory barriers are used
in these functions.
This patch don't call these functions if no descriptors have
been dequeued. The performance gain with split ring when
dequeue zero-copy is disabled should be null, but should be
noticeable with packed ring or dequeue zero-copy enabled.
Fixes: ae999ce49dcb ("vhost: add Tx support for packed ring")
Fixes: 915cf9404225 ("vhost: use shadow used ring in dequeue path")
Cc: stable@dpdk.org
Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Reviewed-by: Jens Freimann <jfreimann@redhat.com>
Tested-by: Jens Freimann <jfreimann@redhat.com>
Reviewed-by: Tiwei Bie <tiwei.bie@intel.com>
The MAC addresses of a port can be matched with devargs.
As the conflict between rte_ether.h and netinet/ether.h is not resolved,
the MAC parsing is done with a rte_cmdline function.
As a result, cmdline library becomes a dependency of ethdev.
Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Acked-by: Andrew Rybchenko <arybchenko@solarflare.com>
The representor id is added in rte_eth_dev_data in order to be able
to match a port with its representor id in devargs.
Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Reviewed-by: Andrew Rybchenko <arybchenko@solarflare.com>
The functions for representor devargs parsing were static
in the file rte_ethdev.c.
In order to reuse them in the file rte_class_eth.c,
they are moved to the files ethdev_private.c/.h.
A log is fixed by adding a missing line feed.
Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Acked-by: Andrew Rybchenko <arybchenko@solarflare.com>
If a value contains a comma, rte_kvargs_tokenize() will split here.
In order to support list syntax [a,b] as value, an extra parsing of
the square brackets is added.
Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
These hotplug functions were deprecated and have some new replacements.
As announced earlier, the oldest ones are now removed.
Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Reviewed-by: Andrew Rybchenko <arybchenko@solarflare.com>
The hotplug attach/detach features are implemented in EAL layer.
There is a new ethdev iterator to retrieve ports from ethdev layer.
As announced earlier, the (buggy) ethdev functions are now removed.
Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Reviewed-by: Andrew Rybchenko <arybchenko@solarflare.com>
If no rte_device is given in the iterator,
eth_dev_match() is looking at all ports without any restriction,
except the ethdev kvargs filter.
It allows to iterate with a devargs filter referencing only
some ethdev parameters. The format (from the new devargs syntax) is:
class=eth,paramY=Y
Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Reviewed-by: Andrew Rybchenko <arybchenko@solarflare.com>
The iterator will return the ethdev port ids matching a devargs string.
It is recommended to use the macro RTE_ETH_FOREACH_MATCHING_DEV()
for usage convenience.
The class string is prefixed with '+' in order to skip the validation
of the parameter keys. It is tolerated for the compatibility with
the old (current) syntax where all parameters (bus, class and driver)
are mixed in the same string without any delimiter.
Thanks to this compatibility prefix, the driver parameters will be
skipped during the ethdev parsing, and not considered invalid.
A macro is introduced in rte_common.h to workaround a const field.
This hack is needed to free const strings in the iterator.
It is preferred to keep the const for these fields, because it gives
a hint that they are not changed at each iteration.
Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Reviewed-by: Andrew Rybchenko <arybchenko@solarflare.com>
We should return the length of the buffers described by
the current descriptor chain after filling the buffer
vector. So we need to zero the *len first.
Fixes: 2f3225a7d69b ("vhost: add vector filling support for packed ring")
Cc: stable@dpdk.org
Signed-off-by: Tiwei Bie <tiwei.bie@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Currently it's not possible to build DPDK as shared library with
cryptodev disabled since vhost is trying to link with rte_crypto,
but rte_crypto and rte_hash are only needed when you build vhost_crypto
and so only when cryptodev is enabled.
This patch fix this by linking rte_vhost with rte_crypto and rte_hash
only when cryptodev is enabled.
Fixes: b4ca81298613 ("vhost/crypto: fix build without cryptodev")
Fixes: 939066d96563 ("vhost/crypto: add public function implementation")
Cc: stable@dpdk.org
Signed-off-by: Timothy Redaelli <tredaelli@redhat.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Currenlty the encap/decap actions only support encapsulation
of VXLAN and NVGRE L2 packets (L2 encapsulation is where
the inner packet has a valid Ethernet header, while L3 encapsulation
is where the inner packet doesn't have the Ethernet header).
In addtion the parameter to to the encap action is a list of rte items,
this results in 2 extra translation, between the application to the
actioni and from the action to the NIC. This results in negative impact
on the insertion performance.
Looking forward there are going to be a need to support many more tunnel
encapsulations. For example MPLSoGRE, MPLSoUDP.
Adding the new encapsulation will result in duplication of code.
For example the code for handling NVGRE and VXLAN are exactly the same,
and each new tunnel will have the same exact structure.
This patch introduce a raw encapsulation that can support L2 tunnel types
and L3 tunnel types. In addtion the new
encapsulations commands are using raw buffer inorder to save the
converstion time, both for the application and the PMD.
In order to encapsulate L3 tunnel type there is a need to use both
actions in the same rule: The decap to remove the L2 of the original
packet, and then encap command to encapsulate the packet with the
tunnel.
For decap L3 there is also a need to use both commands in the same flow
first the decap command to remove the outer tunnel header and then encap
to add the L2 header.
Signed-off-by: Ori Kam <orika@mellanox.com>
Acked-by: Andrew Rybchenko <arybchenko@solarflare.com>
As described in [1], a new rte_flow item is added to support metadata
to use as flow rule match pattern.
The metadata is an opaque item, fully controlled by the application.
The use of metadata is relevant for egress rules only.
It can be set in the flow rule using the RTE_FLOW_ITEM_META.
An additional member 'tx_metadata' is added in union with existing member
'hash' of struct 'rte_mbuf', located to avoid conflicts with existing
fields. This additional member is used to carry the metadata item.
Application should set the packet metadata in the mbuf dedicated field,
and set the PKT_TX_METADATA flag in the mbuf->ol_flags.
The NIC will use the packet metadata as match criteria for relevant
flow rules.
This patch introduces metadata item type for rte_flow RTE_FLOW_ITEM_META,
along with corresponding struct rte_flow_item_meta and ol_flag
PKT_TX_METADATA.
[1] "[RFC,v2] ethdev: support metadata as flow rule criteria"
Signed-off-by: Dekel Peled <dekelp@mellanox.com>
Acked-by: Andrew Rybchenko <arybchenko@solarflare.com>
After closing a port, it cannot be restarted.
So there is no reason to not free all associated resources.
The last step was done with rte_eth_dev_detach() which is deprecated.
Instead of blindly removing the associated rte_device, the driver should
check if no more port (ethdev, cryptodev, etc) is open for the device.
The last ethdev freeing which were done by rte_eth_dev_detach(),
are now done at the end of rte_eth_dev_close() if the driver supports
the flag RTE_ETH_DEV_CLOSE_REMOVE.
There will be a transition period for PMDs to enable this new flag
and migrate to the new behaviour.
When enabling RTE_ETH_DEV_CLOSE_REMOVE, the PMD must free all its
private resources for the port, in its dev_close function.
It is advised to call the dev_close function in the remove function
in order to support removing a device without closing its ports.
Some drivers does not allocate MAC addresses dynamically or separately.
In those cases, the pointer is set to NULL, in order to avoid wrongly
freeing them in rte_eth_dev_release_port().
A closed port will have the state RTE_ETH_DEV_UNUSED which is
considered as invalid by rte_eth_dev_is_valid_port().
So validity is not checked anymore for closed ports in testpmd.
Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Reviewed-by: Andrew Rybchenko <arybchenko@solarflare.com>
After previous changes, the function rte_eth_dev_release_port()
can be used for primary or secondary process as well.
The only difference with rte_eth_dev_release_port_secondary()
is the shared lock used in rte_eth_dev_release_port().
The function rte_eth_dev_release_port_secondary() was recently
added in 18.11 cycle.
Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Reviewed-by: Andrew Rybchenko <arybchenko@solarflare.com>
This is a clean-up of common ethdev data freeing.
All data freeing are moved to rte_eth_dev_release_port()
and done only in case of primary process.
It is probably fixing some memory leaks for PMDs which were
not freeing all data.
Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Acked-by: Andrew Rybchenko <arybchenko@solarflare.com>
Some doxygen comments were wrongly associated to the next field
because of syntax /** instead of /**<
Some other cleanups (like alignment) are done.
Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Reviewed-by: Andrew Rybchenko <arybchenko@solarflare.com>
When no-huge mode is enabled, we always overwrite the socket ID to be
SOCKET_ID_ANY in rte_malloc, because there is no NUMA awareness in no-huge
mode. However, with external memory support, a socket ID may have other
meaning, and we cannot overwrite the socket ID in those cases.
Fixes: 65ff37b105f7 ("malloc: add function to check if socket is external")
Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
There is a rte_pause in hash table reset function.
Since the loop is not a polling loop on shared
data structure, the rte_pause is not needed.
Fixes: b26473ff8f4a ("hash: add reset function")
Cc: stable@dpdk.org
Signed-off-by: Yipeng Wang <yipeng1.wang@intel.com>
Acked-by: Bruce Richardson <bruce.richardson@intel.com>
Add a new API function to KNI, rte_kni_update_link() to allow DPDK
applications to update the link status for KNI network interfaces in
the linux kernel.
Signed-off-by: Dan Gora <dg@adax.com>
Acked-by: Ferruh Yigit <ferruh.yigit@intel.com>
Syncing the values by adding c11 atomic memory barriers to make sure
the values being synced before updating fifo_write and fifo_read.
Signed-off-by: Phil Yang <phil.yang@arm.com>
Reviewed-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
Reviewed-by: Gavin Hu <gavin.hu@arm.com>
Reviewed-by: Ola Liljedahl <ola.liljedahl@arm.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Adding memory barrier to make sure the values being synced
before updating fifo_write in kni_fifo_put and fifo_read in
kni_fifo_get.
Fixes: 3fc5ca2f6352 ("kni: initial import")
Cc: stable@dpdk.org
Signed-off-by: Phil Yang <phil.yang@arm.com>
Reviewed-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
Reviewed-by: Gavin Hu <gavin.hu@arm.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
With existing code in kni_fifo_put, rx_q values are not being updated
before updating fifo_write. While reading rx_q in kni_net_rx_normal,
This is causing the sync issue on other core. The same situation happens
in kni_fifo_get as well.
So syncing the values by adding memory barriers to make sure the values
being synced before updating fifo_write and fifo_read.
Fixes: 3fc5ca2f6352 ("kni: initial import")
Cc: stable@dpdk.org
Signed-off-by: Phil Yang <phil.yang@arm.com>
Reviewed-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
Reviewed-by: Gavin Hu <gavin.hu@arm.com>
Reviewed-by: Ola Liljedahl <ola.liljedahl@arm.com>
Reviewed-by: Jerin Jacob <jerin.jacob@caviumnetworks.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Keep only single config option RTE_USE_C11_MEM_MODEL for C11 memory
model, so all modules can leverage C11 atomic extension by enable this
option.
Signed-off-by: Phil Yang <phil.yang@arm.com>
Reviewed-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
Reviewed-by: Gavin Hu <gavin.hu@arm.com>
Acked-by: Jerin Jacob <jerin.jacob@caviumnetworks.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Add support for rte_pause() implementation for ppc64.
Signed-off-by: Jerin Jacob <jerin.jacob@caviumnetworks.com>
Acked-by: Chao Zhu <chaozhu@linux.vnet.ibm.com>
Add lock-free read-write concurrency. This is achieved by the
following changes.
1) Add memory ordering to avoid race conditions. The only race
condition that can occur is - using the key store element
before the key write is completed. Hence, while inserting the element
the release memory order is used. Any other race condition is caught
by the key comparison. Memory orderings are added only where needed.
For ex: reads in the writer's context do not need memory ordering
as there is a single writer.
key_idx in the bucket entry and pdata in the key store element are
used for synchronisation. key_idx is used to release an inserted
entry in the bucket to the reader. Use of pdata for synchronisation
is required due to updation of an existing entry where-in only
the pdata is updated without updating key_idx.
2) Reader-writer concurrency issue, caused by moving the keys
to their alternative locations during key insert, is solved
by introducing a global counter(tbl_chng_cnt) indicating a
change in table.
3) Add the flag to enable reader-writer concurrency during
run time.
Signed-off-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
Reviewed-by: Gavin Hu <gavin.hu@arm.com>
Reviewed-by: Ola Liljedahl <ola.liljedahl@arm.com>
Reviewed-by: Steve Capper <steve.capper@arm.com>
Reviewed-by: Yipeng Wang <yipeng1.wang@intel.com>
Acked-by: Bruce Richardson <bruce.richardson@intel.com>
Fix the key store array element alignment such that every array
element is aligned on KEY_ALIGNMENT boundary. This is required to
make 'pdata' in 'struct rte_hash_key' align on its natural boundary
for atomic load/store.
Fixes: 473d1bebce43 ("hash: allow to store data in hash table")
Cc: stable@dpdk.org
Signed-off-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
Reviewed-by: Gavin Hu <gavin.hu@arm.com>
Reviewed-by: Ola Liljedahl <ola.liljedahl@arm.com>
Reviewed-by: Steve Capper <steve.capper@arm.com>
Reviewed-by: Yipeng Wang <yipeng1.wang@intel.com>
Acked-by: Bruce Richardson <bruce.richardson@intel.com>
rte_hash_lookup_xxx APIs return the index of slot in
the key store. Application(reader) can use that index to reference
other data structures in its scope. Because of this, the
index should not be freed till the application completes
using the index.
RTE_HASH_EXTRA_FLAGS_NO_FREE_ON_DEL is introduced to support this.
When this flag is enabled rte_hash_del_xxx APIs do not free the
key-store index/internal memory associated with the deleted
entry. The new API rte_hash_free_key_with_position should be called
to free the key-store index/internal memory after calling
rte_hash_del_xxx APIs.
Suggested-by: Yipeng Wang <yipeng1.wang@intel.com>
Signed-off-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
Reviewed-by: Gavin Hu <gavin.hu@arm.com>
Reviewed-by: Yipeng Wang <yipeng1.wang@intel.com>
Acked-by: Bruce Richardson <bruce.richardson@intel.com>
RW concurrency is required with single writer and multiple reader
usecase as well. Hence, multi-writer should not be enabled by default when
RW concurrency is enabled.
Fixes: f2e3001b53ec ("hash: support read/write concurrency")
Cc: stable@dpdk.org
Signed-off-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
Reviewed-by: Gavin Hu <gavin.hu@arm.com>
Reviewed-by: Yipeng Wang <yipeng1.wang@intel.com>
Acked-by: Bruce Richardson <bruce.richardson@intel.com>
This patch does a couple of things:
* Adds a new message type for removing policies (PKT_POLICY_REMOVE)
Used when we want to remove a previously created policy.
* Adds a core_type bool to the channel packet struct to specify whether
the type of core we want to control is virtual or physical.
Signed-off-by: David Hunt <david.hunt@intel.com>
Acked-by: Anatoly Burakov <anatoly.burakov@intel.com>
1. Abstract
For packet processing workloads such as DPDK polling is continuous.
This means CPU cores always show 100% busy independent of how much work
those cores are doing. It is critical to accurately determine how busy
a core is hugely important for the following reasons:
* No indication of overload conditions.
* User does not know how much real load is on a system, resulting
in wasted energy as no power management is utilized.
Compared to the original l3fwd-power design, instead of going to sleep
after detecting an empty poll, the new mechanism just lowers the core
frequency. As a result, the application does not stop polling the device,
which leads to improved handling of bursts of traffic.
When the system become busy, the empty poll mechanism can also increase the
core frequency (including turbo) to do best effort for intensive traffic.
This gives us more flexible and balanced traffic awareness over the
standard l3fwd-power application.
2. Proposed solution
The proposed solution focuses on how many times empty polls are executed.
The less the number of empty polls, means current core is busy with
processing workload, therefore, the higher frequency is needed. The high
empty poll number indicates the current core not doing any real work
therefore, we can lower the frequency to safe power.
In the current implementation, each core has 1 empty-poll counter which
assume 1 core is dedicated to 1 queue. This will need to be expanded in the
future to support multiple queues per core.
2.1 Power state definition:
LOW: Not currently used, reserved for future use.
MED: the frequency is used to process modest traffic workload.
HIGH: the frequency is used to process busy traffic workload.
2.2 There are two phases to establish the power management system:
a.Initialization/Training phase. The training phase is necessary
in order to figure out the system polling baseline numbers from
idle to busy. The highest poll count will be during idle, where
all polls are empty. These poll counts will be different between
systems due to the many possible processor micro-arch, cache
and device configurations, hence the training phase.
In the training phase, traffic is blocked so the training
algorithm can average the empty-poll numbers for the LOW, MED and
HIGH power states in order to create a baseline.
The core's counter are collected every 10ms, and the Training
phase will take 2 seconds.
Training is disabled as default configuration. The default
parameter is applied. Sample App still can trigger training
if that's needed. Once the training phase has been executed once on
a system, the application can then be started with the relevant
thresholds provided on the command line, allowing the application
to start passing start traffic immediately
b.Normal phase. Traffic starts immediately based on the default
thresholds, or based on the user supplied thresholds via the
command line parameters. The run-time poll counts are compared with
the baseline and the decision will be taken to move to MED power
state or HIGH power state. The counters are calculated every 10ms.
3. Proposed API
1. rte_power_empty_poll_stat_init(struct ep_params **eptr,
uint8_t *freq_tlb, struct ep_policy *policy);
which is used to initialize the power management system.
2. rte_power_empty_poll_stat_free(void);
which is used to free the resource hold by power management system.
3. rte_power_empty_poll_stat_update(unsigned int lcore_id);
which is used to update specific core empty poll counter, not thread safe
4. rte_power_poll_stat_update(unsigned int lcore_id, uint8_t nb_pkt);
which is used to update specific core valid poll counter, not thread safe
5. rte_power_empty_poll_stat_fetch(unsigned int lcore_id);
which is used to get specific core empty poll counter.
6. rte_power_poll_stat_fetch(unsigned int lcore_id);
which is used to get specific core valid poll counter.
7. rte_empty_poll_detection(struct rte_timer *tim, void *arg);
which is used to detect empty poll state changes then take action.
Signed-off-by: Liang Ma <liang.j.ma@intel.com>
Reviewed-by: Lei Yao <lei.a.yao@intel.com>
Acked-by: David Hunt <david.hunt@intel.com>
This commit changes the hashing mechanism to "partial-key
hashing" to calculate bucket index and signature of key.
This is proposed in Bin Fan, et al's paper
"MemC3: Compact and Concurrent MemCache with Dumber Caching
and Smarter Hashing". Basically the idea is to use "xor" to
derive alternative bucket from current bucket index and
signature.
With "partial-key hashing", it reduces the bucket memory
requirement from two cache lines to one cache line, which
improves the memory efficiency and thus the lookup speed.
Signed-off-by: Yipeng Wang <yipeng1.wang@intel.com>
Reviewed-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
Acked-by: Dharmik Thakkar <dharmik.thakkar@arm.com>
Acked-by: Bruce Richardson <bruce.richardson@intel.com>
