Include efx_mcdi.h from main internal header efx_impl.h directly
instead of indirect inclusion via family-specific header.
It avoids duplication and fixes Medford and Medford2 cases where
the header inclusion is lost.
Correctness is still guaranteed by checks in efx_check.h which
require EFSYS_OPT_MCDI for corresponding families and do not
allow to enable the option if no family requires it.
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
net/sfc/base: avoid division by 0 if no event queue timers
If event queue timers are not supported, enc_evq_timer_quantum_ns and
enc_evq_timer_max_us should be set to 0. Make sure that division by 0
does not happen in libefx, if public function efx_ev_usecs_to_ticks()
is used in this case.
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
net/sfc/base: pass Rx buffer size to RxQ init if supported
Riverhead requires Rx buffer size to be specified in INIT_RXQ.
If the parameter is not supported (e.g. on EF10 family adapters),
zero must be used on INIT_RXQ.
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Mark Spender [Thu, 7 Feb 2019 16:29:33 +0000 (16:29 +0000)]
net/sfc/base: support the Rx event mode w/o continue
The recently added NO_CONT_EV mode is recommended for when looking for
maximum throughput on 100G links as it allows the firmware to operate
more efficiently. The biggest benefit is when using scatter and jumbo
frames, but it is also necessary to achieve line rate in other cases.
Support for NO_CONT_EV when scatter is disabled is simple - the main
datapath change is to always read the packet length from the prefix,
not the event. This requires storing a flag with the event queue so
the event handler knows NO_CONT_EV mode is in use.
Supporting NO_CONT_EV with scatter would require an API change.
(Now the ee_flags field in efx_evq_t is used on the datapath, move it
next to other fields which are also read on the datapath to try to avoid
reading an additional cache line.)
Signed-off-by: Mark Spender <mspender@solarflare.com> Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Igor Romanov [Thu, 7 Feb 2019 16:29:29 +0000 (16:29 +0000)]
net/sfc: use NIC event descriptor size instead of common
Use of the macro for calculating a Rx queue size is deprecated. Replace
it with a call to a function that uses descriptor size specified for
every NIC.
Signed-off-by: Igor Romanov <igor.romanov@oktetlabs.ru> Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Igor Romanov [Thu, 7 Feb 2019 16:29:28 +0000 (16:29 +0000)]
net/sfc: use NIC Rx descriptor size instead of common
Use of the macro for calculating a Rx queue size is deprecated. Replace
it with a call to a function that uses descriptor size specified for
every NIC.
Signed-off-by: Igor Romanov <igor.romanov@oktetlabs.ru> Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Igor Romanov [Thu, 7 Feb 2019 16:29:27 +0000 (16:29 +0000)]
net/sfc: use NIC Tx descriptor size instead of common
Use of the macro for calculating a Tx queue size is deprecated. Replace
it with a call to a function that uses descriptor size specified for
every NIC.
Signed-off-by: Igor Romanov <igor.romanov@oktetlabs.ru> Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Igor Romanov [Thu, 7 Feb 2019 16:29:26 +0000 (16:29 +0000)]
net/sfc/base: round number of queue buffers up
Functions efx_*q_nbufs return the number of a queue buffers by dividing
the queue size by page size. If minimum size of a queue is smaller than
one page, we still need the page and number of buffers should not be 0.
Signed-off-by: Igor Romanov <igor.romanov@oktetlabs.ru> Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Igor Romanov [Thu, 7 Feb 2019 16:29:25 +0000 (16:29 +0000)]
net/sfc/base: support different event descriptor sizes
For consistency with the size of Tx descriptors, the size of event
descriptors should be a part of NIC config, not a macro that is
common for all NIC families. Also, add a max number of EvQ buffers
define which is needed to initialize an MCDI buffer at compile time.
Signed-off-by: Igor Romanov <igor.romanov@oktetlabs.ru> Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Igor Romanov [Thu, 7 Feb 2019 16:29:24 +0000 (16:29 +0000)]
net/sfc/base: support different Rx descriptor sizes
For consistency with the size of Tx descriptors, the size of Rx
descriptors should be a part of NIC config, not a macro that is
common for all NIC families.
Signed-off-by: Igor Romanov <igor.romanov@oktetlabs.ru> Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Igor Romanov [Thu, 7 Feb 2019 16:29:21 +0000 (16:29 +0000)]
net/sfc/base: make max number of TxQ bufs EF10-specific
The define EFX_TXQ_MAX_BUFS is used only by EF10 family implementation.
Since the macro is not used by any drivers, it does not require
exposed libefx interface.
Make the define EF10-specific.
Signed-off-by: Igor Romanov <igor.romanov@oktetlabs.ru> Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Igor Romanov [Thu, 7 Feb 2019 16:29:20 +0000 (16:29 +0000)]
net/sfc/base: move a macro to Siena implementation header
The macro EFX_RXQ_DC_NDESCS() is used only by Siena implementation.
Since the macro is not used by any drivers, it does not require
exposed libefx interface.
Make the macro Siena-specific.
Signed-off-by: Igor Romanov <igor.romanov@oktetlabs.ru> Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Igor Romanov [Thu, 7 Feb 2019 16:29:19 +0000 (16:29 +0000)]
net/sfc/base: move EVQ descs number check to generic place
Now we have min/max limits in NIC config, so we can do check against
min/max in a generic place instead of NIC family specific functions.
Check that the descriptors number is a power of 2 is also can be
made common. It removes code duplication and makes NIC family
specific functions a bit shorter.
Signed-off-by: Igor Romanov <igor.romanov@oktetlabs.ru> Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Igor Romanov [Thu, 7 Feb 2019 16:29:18 +0000 (16:29 +0000)]
net/sfc/base: move Rx descs number check to generic place
Now we have min/max limits in NIC config, so we can do check against
min/max in a generic place instead of NIC family specific functions.
Check that the descriptors number is a power of 2 is also can be
made common. It removes code duplication and makes NIC family
specific functions a bit shorter.
Signed-off-by: Igor Romanov <igor.romanov@oktetlabs.ru> Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Igor Romanov [Thu, 7 Feb 2019 16:29:17 +0000 (16:29 +0000)]
net/sfc/base: move Tx descs number check to generic place
Now we have min/max limits in NIC config, so we can do check against
min/max in a generic place instead of NIC family specific functions.
Check that the descriptors number is a power of 2 is also can be made
common. It removes code duplication and makes NIC family
specific functions a bit shorter.
Signed-off-by: Igor Romanov <igor.romanov@oktetlabs.ru> Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Igor Romanov [Thu, 7 Feb 2019 16:29:10 +0000 (16:29 +0000)]
net/sfc/base: remove min define for number of Tx descs
EF100/Riverhead has different min limit. So, this limit should
be a part of NIC config, not define common for all NIC families.
Define maximum Tx descriptor number for Siena in the same way as
minimum for consistency.
Signed-off-by: Igor Romanov <igor.romanov@oktetlabs.ru> Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Ivan Malov [Thu, 7 Feb 2019 16:29:09 +0000 (16:29 +0000)]
net/sfc/base: report support for Tx checksum op descriptors
FreeBSD driver needs a patch to provide a means for packets
which do not need checksum offload but have flow ID set
to avoid hitting only the first Tx queue (which has been used
for packets not needing checksum offload).
This should be possible on Huntington, Medford or Medford2 chips
since these support toggling checksum offload on any given queue
dynamically by means of pushing option descriptors.
The patch for FreeBSD driver will then need a means to figure out
whether the feature can be used, and testing adapter family might
not be a good solution.
This patch adds a feature bit specifically to indicate support
for checksum option descriptors. The new feature bits may have
more users in future, apart from the mentioned FreeBSD patch.
Signed-off-by: Ivan Malov <ivan.malov@oktetlabs.ru> Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
net/sfc/base: update external port number calculation
Revise the external port calculation to support all
X2 port modes. The previous algorithm could not
handle different port numbering schemes on each cage.
Signed-off-by: Richard Houldsworth <rhouldsworth@solarflare.com> Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
net/sfc: make main logging macro reusable in secondary
Logging should be supported in secondary process and now we have
everything required to share at least main logging macro.
sfc_adapter is going to become primary process only.
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
net/sfc: start to factor out multi-process shared data
sfc_adapter structure will become primary process only private data.
Start to factor out shared data into dedicated structure which will
become separate structure finally.
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
net/sfc: make it simpler to change datapath ops location
Rx and Tx datapath ops should be stored in process private
data since these structures are located in the data section
(not allocated from shared memory).
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Ivan Malov [Thu, 7 Feb 2019 12:17:24 +0000 (12:17 +0000)]
net/sfc: log port ID as 16-bit unsigned integer on panic
Fixes: f8244c6399d9 ("ethdev: increase port id range") Cc: stable@dpdk.org Signed-off-by: Ivan Malov <ivan.malov@oktetlabs.ru> Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
test_create_bonded_device is failing due to improper initialisation in
bonded device configuration. Which leads to crash while setting up
queues.
The value of nb_rx_desc is checked if it is not in range of rx_desc_lim
of bonded device which fails.
This is due to "rx_desc_lim" is set to 0 as default value of bonded
device during bond_alloc().
Hence nb_rx_desc (1024) is > 0 and test fails.
Fix is to set the default values of rx_desc_lim of bonded device to
appropriate value.
Receive the values from slaves configuration like done for other
existing slave configuration
Fixes: 2efb58cbab6e ("bond: new link bonding library") Cc: stable@dpdk.org Signed-off-by: Hari Kumar Vemula <hari.kumarx.vemula@intel.com> Acked-by: Chas Williams <chas3@att.com>
Nikhil Rao [Tue, 22 Jan 2019 05:44:26 +0000 (11:14 +0530)]
doc: announce parameter change of a service API function
rte_service_attr_get() is passed a uint32_t * to retrieve
an attribute value, this will be changed to uin64_t * as per
patch posted at http://patchwork.dpdk.org/patch/49968/
Signed-off-by: Nikhil Rao <nikhil.rao@intel.com> Acked-by: Harry van Haaren <harry.van.haaren@intel.com> Acked-by: Kevin Traynor <ktraynor@redhat.com> Acked-by: Andrew Rybchenko <arybchenko@solarflare.com> Acked-by: David Marchand <david.marchand@redhat.com>
Bruce Richardson [Wed, 16 Jan 2019 11:37:13 +0000 (11:37 +0000)]
doc: announce meson version upgrade
Add a notice in the deprecation section of the release notes to call out
the fact that the minimum supported meson version for DPDK will change
from 19.05 onwards.
Signed-off-by: Bruce Richardson <bruce.richardson@intel.com> Acked-by: Luca Boccassi <bluca@debian.org> Acked-by: Timothy Redaelli <tredaelli@redhat.com> Acked-by: Harry van Haaren <harry.van.haaren@intel.com> Acked-by: Thomas Monjalon <thomas@monjalon.net>
Bruce Richardson [Wed, 30 Jan 2019 16:43:44 +0000 (16:43 +0000)]
doc: remove note about meson being experimental
At this stage, meson builds are widely supported for DPDK, and so
the build system should be no longer called out as experimental.
NOTE: this does not imply it's the primary build system, just that
it's safe to use for day-to-day work and for packaging if so
desired.
Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
Anatoly Burakov [Thu, 31 Jan 2019 17:05:51 +0000 (17:05 +0000)]
doc: clarify libnuma requirement for NUMA systems
Since 18.05, libnuma is pretty much required on Linux when using
non-legacy mode, because without it, we cannot know where our
hugepages are located [1].
In legacy mode, libnuma is not required because we can still sort
pages by sockets, as we use pagemap lookup method to figure out
socket ID's for pages.
So, document libnuma as required for NUMA systems and non-legacy
mode.
Declan Doherty [Mon, 28 Jan 2019 12:57:19 +0000 (12:57 +0000)]
maintainers: resign from maintainership of link bonding
I've been unable to dedicate enough time to actively help in the
maintainership of the link bonding PMD, and as Chas is now actively
maintaining this PMD, I'm removing my name against it in the
MAINTAINERS file.
Thomas Monjalon [Wed, 30 Jan 2019 11:20:19 +0000 (12:20 +0100)]
doc: add references to flow isolated mode in NICs guide
Some drivers (mlx, mvpp2, sfc) support the flow isolated mode,
but the feature was not advertised.
A reference to the feature description is added for each driver.
Signed-off-by: Thomas Monjalon <thomas@monjalon.net> Acked-by: Shahaf Shuler <shahafs@mellanox.com>
Rami Rosen [Sun, 20 Jan 2019 18:29:47 +0000 (20:29 +0200)]
doc: add dependency for PDF in contributing guide
There is a missing depenency for building docs with "make doc-guides-pdf".
This causes it to break with "make[3]: latexmk: Command not found". This
was observed and reported in https://bugs.dpdk.org/show_bug.cgi?id=182
This patch fixes this issue by adding the latexmk package dependency in
sub-section 4.3.1 of the contributing guide ("Dependencies").
Rafal Kozik [Fri, 25 Jan 2019 08:10:28 +0000 (09:10 +0100)]
doc: add instructions for WC in ENAv2
ENAv2 hardware provides Low Latency Queue v2 (LLQv2). It needs kernel
PCI driver to support write combining (WC). This patch add information
how to use it with igb_uio and vfio-pci drivers.
Signed-off-by: Rafal Kozik <rk@semihalf.com> Acked-by: Michal Krawczyk <mk@semihalf.com>
git.droids-corp.org / dpdk.git / log