On 5/7/2021 4:22 PM, Anatoly Burakov wrote:
Currently, there is a hard limitation on the PMD power management
support that only allows it to support a single queue per lcore. This is
not ideal as most DPDK use cases will poll multiple queues per core.
The PMD power management mechanism relies on ethdev Rx callbacks, so it
is very difficult to implement such support because callbacks are
effectively stateless and have no visibility into what the other ethdev
devices are doing. This places limitations on what we can do within the
framework of Rx callbacks, but the basics of this implementation are as
follows:
- Replace per-queue structures with per-lcore ones, so that any device
polled from the same lcore can share data
- Any queue that is going to be polled from a specific lcore has to be
added to the list of queues to poll, so that the callback is aware of
other queues being polled by the same lcore
- Both the empty poll counter and the actual power saving mechanism is
shared between all queues polled on a particular lcore, and is only
activated when all queues in the list were polled and were determined
to have no traffic.
- The limitation on UMWAIT-based polling is not removed because UMWAIT
is incapable of monitoring more than one address.
Also, while we're at it, update and improve the docs.
Signed-off-by: Anatoly Burakov <anatoly.bura...@intel.com>
---
Notes:
v6:
- Track each individual queue sleep status (Konstantin)
- Fix segfault (Dave)
v5:
- Remove the "power save queue" API and replace it with mechanism
suggested by
Konstantin
v3:
- Move the list of supported NICs to NIC feature table
v2:
- Use a TAILQ for queues instead of a static array
- Address feedback from Konstantin
- Add additional checks for stopped queues
doc/guides/nics/features.rst | 10 +
doc/guides/prog_guide/power_man.rst | 65 ++--
doc/guides/rel_notes/release_21_08.rst | 3 +
lib/power/rte_power_pmd_mgmt.c | 452 +++++++++++++++++++------
4 files changed, 394 insertions(+), 136 deletions(-)
--snip--
+static inline void
+queue_reset(struct pmd_core_cfg *cfg, struct queue_list_entry *qcfg)
+{
+ const bool is_ready_to_sleep = qcfg->n_empty_polls > EMPTYPOLL_MAX;
+
+ /* reset empty poll counter for this queue */
+ qcfg->n_empty_polls = 0;
+ /* reset the queue sleep counter as well */
+ qcfg->n_sleeps = 0;
+ /* remove the queue from list of cores ready to sleep */
+ if (is_ready_to_sleep)
+ cfg->n_queues_ready_to_sleep--;
Hi Anatoly,
I don't think the logic around this is bulletproof yet, in my
testing I'm seeing n_queues_ready_to_sleep wrap around (i.e. decremented
while already zero).
Rgds,
Dave.
--snip--
