> -----Original Message----- > From: Abdullah Sevincer <abdullah.sevin...@intel.com> > Sent: Monday, September 9, 2024 9:35 PM > To: dev@dpdk.org > Cc: Jerin Jacob <jer...@marvell.com>; bruce.richard...@intel.com; > pravin.pat...@intel.com; mattias.ronnb...@ericsson.com; > manish.aggar...@intel.com; Abdullah Sevincer <abdullah.sevin...@intel.com> > Subject: [EXTERNAL] [PATCH v12 2/3] event/dlb2: add support for independent > enqueue > > DLB devices need events to be enqueued in the same order they are dequeued. > Applications are not suppose to change event order between dequeue and to > enqueue. Since Eventdev standard does not add such restrictions independent > enqueue support > DLB devices need events to be enqueued in the same order they are dequeued. > Applications are not suppose to change event order between dequeue and to > enqueue. Since Eventdev standard does not add such restrictions independent > enqueue support is needed for DLB PMD so that it restores dequeue order on > enqueue if applications happen to change it. It also adds missing releases in > places where events are dropped by the application and it expects implicit > release to handle it. > > By default the feature will be off on all DLB ports and they will behave the > same > as older releases. To enable reordering feature, applications need to add the > flag RTE_EVENT_PORT_CFG_INDEPENDENT_ENQ > to port configuration if only the device advertises the capability > RTE_EVENT_DEV_CAP_INDEPENDENT_ENQ. > > Signed-off-by: Abdullah Sevincer <abdullah.sevin...@intel.com> > Acked-by: Mattias Rönnblom <mattias.ronnb...@ericsson.com> > --- > doc/guides/eventdevs/dlb2.rst | 41 +++ > doc/guides/rel_notes/release_24_11.rst | 6 + > drivers/event/dlb2/dlb2.c | 492 ++++++++++++++++--------- > drivers/event/dlb2/dlb2_avx512.c | 27 +- > drivers/event/dlb2/dlb2_inline_fns.h | 8 + > drivers/event/dlb2/dlb2_priv.h | 25 +- > drivers/event/dlb2/rte_pmd_dlb2.h | 10 + > 7 files changed, 419 insertions(+), 190 deletions(-) > > diff --git a/doc/guides/eventdevs/dlb2.rst b/doc/guides/eventdevs/dlb2.rst > index 2532d92888..8b973cf81e 100644 > --- a/doc/guides/eventdevs/dlb2.rst > +++ b/doc/guides/eventdevs/dlb2.rst > @@ -456,6 +456,47 @@ Example command to enable QE Weight feature: > > --allow ea:00.0,enable_cq_weight=<y/Y> > > +Independent Enqueue Capability > +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ > + > +DLB2 hardware device expects all forwarded events to be enqueued in the > +same order as they are dequeued. For dropped events, their releases > +should come at the same location as the original event was expected. > +Hardware has this restriction as it uses the order to retrieve > +information about the original event that was sent to the CPU. This > +contains information like atomic flow ID to release the flow lock and > +ordered events sequence number to restore the original order. > + > +Some applications, like those based on the DPDK dispatcher library, > +want enqueue order independence. To support this, DLB2 PMD supports the > +``RTE_EVENT_DEV_CAP_INDEPENDENT_ENQ`` capability. > + > +This capability applies to Eventdevs supporting burst mode. On ports > +where the application is going to change enqueue order, > +``RTE_EVENT_PORT_CFG_INDEPENDENT_ENQ`` support should be enabled. > + > +Example code to inform PMD that the application plans to use > +independent enqueue order on a port: > + > + .. code-block:: c > + > + if (capability & RTE_EVENT_DEV_CAP_INDEPENDENT_ENQ) > + port_config = port_config | > + RTE_EVENT_PORT_CFG_INDEPENDENT_ENQ;
A lot of thing in this is generic, update and move that to doc/guides/prog_guide/eventdev/eventdev.rst And add DLB2 specific items here. > + > +This code example enables enqueue event reordering inside DLB2 PMD > +before the events are sent to the DLB2 hardware. If the application is > +not going to change the enqueue order, this flag should not be enabled > +to get better performance. DLB2 PMD saves ordering information inside > +the impl_opaque field of the event, and this field should be preserved > +for all FORWARD or RELEASE events. Following MACROs are provided to get > +and set this field inside the event in case the same event is not used > +for forwarding (e.g., a new RELEASE event is created when the original event > is > dropped instead of reusing the same event). > + > + .. code-block:: c > + > + #define RTE_EVENT_GET_IMPL_OPAQUE(ev) (ev->impl_opaque) > + #define RTE_EVENT_SET_IMPL_OPAQUE(ev, val) (ev->impl_opaque = > + val) Don’t add public RTE_EVENT_ symbol as example macro, use some other prefix like DLB2. > + > Running Eventdev Applications with DLB Device > --------------------------------------------- > > diff --git a/doc/guides/rel_notes/release_24_11.rst > b/doc/guides/rel_notes/release_24_11.rst > index 89c6a67e6a..61796f24e6 100644 > --- a/doc/guides/rel_notes/release_24_11.rst > +++ b/doc/guides/rel_notes/release_24_11.rst > @@ -60,6 +60,12 @@ New Features > * Added support for independent enqueue feature. Updated Event Device and > PMD feature list. > > +* **Updated DLB2 Driver for independent enqueue feature** Driver->driver > + > + Added support for DLB independent enqueue feature. Applications > + should use ``RTE_EVENT_PORT_CFG_INDEPENDENT_ENQ`` to enable the > + feature if the capability ``RTE_EVENT_DEV_CAP_INDEPENDENT_ENQ`` exists. Next patch add one more item to release notes. Instead, IMO, it better to update the generic section. for this PMD support. I.e. add one item under "*Updated eventdev to support independent enqueue feature" section. As add DLB2 support. > > #include <rte_compat.h> > > +/** > + * Macros to get/set QID depth and QE weight from rte_event metadata. > + * Currently 'rsvd' field is used for these. Lower 2 bits are used to > +store > + * QID depth while the upper 2 bits are used for QER weight. > + */ > +#define RTE_PMD_DLB2_GET_QID_DEPTH(x) ((x)->rsvd & 0x3) #define > +RTE_PMD_DLB2_SET_QID_DEPTH(x, v) ((x)->rsvd = ((x)->rsvd & ~0x3) | (v & > +0x3)) #define RTE_PMD_DLB2_GET_QE_WEIGHT(x) (((x)->rsvd >> 2) & 0x3) > +#define RTE_PMD_DLB2_SET_QE_WEIGHT(x, v) ((x)->rsvd = ((x)->rsvd & 0x3) > +| ((v & 0x3) << 2)) All of the above MACROs are public symbols, Add Doxygen comment for each of them. > + > /** > * @warning > * @b EXPERIMENTAL: this API may change, or be removed, without prior > notice > -- > 2.25.1
