nvme: Support Autonomous Power State Transition (APST)

APST is an optional NVMe power-saving feature that allows devices
to autonomously enter higher non-operational power states after a
certain amount of idle time, reducing the controller's overall power
consumption.

Signed-off-by: Alexey Sukhoguzov <sap@eseipi.net>
Reviewed by: imp
Pull Request: https://github.com/freebsd/freebsd-src/pull/1444

share/man/man4/nvme.4

@@ -128,6 +128,15 @@ hw.nvme.hmb_max
 .Pp
 The default value is 5% of physical memory size per device.
 .Pp
+To enable Autonomous Power State Transition (APST), set the following
+tunable value in
+.Xr loader.conf 5 :
+.Bd -literal -offset indent
+hw.nvme.apst_enable=1
+.Ed
+.Pp
+The default vendor-provided settings, if any, will be applied.
+.Pp
 The
 .Xr nvd 4
 driver is used to provide a disk driver to the system by default.

sys/dev/nvme/nvme_ctrlr.c

@@ -781,6 +781,41 @@ nvme_ctrlr_configure_aer(struct nvme_controller *ctrlr)
 	}
 }
 
+static void
+nvme_ctrlr_configure_apst(struct nvme_controller *ctrlr)
+{
+	struct nvme_completion_poll_status status;
+	uint64_t *data;
+	int data_size;
+	bool enable, error = true;
+
+	if (TUNABLE_BOOL_FETCH("hw.nvme.apst_enable", &enable) == 0 ||
+	    ctrlr->cdata.apsta == 0)
+		return;
+
+	data_size = 32 * sizeof(*data);
+	data = malloc(data_size, M_NVME, M_WAITOK | M_ZERO);
+
+	status.done = 0;
+	nvme_ctrlr_cmd_get_feature(ctrlr,
+	    NVME_FEAT_AUTONOMOUS_POWER_STATE_TRANSITION, 0,
+	    data, data_size, nvme_completion_poll_cb, &status);
+	nvme_completion_poll(&status);
+	if (nvme_completion_is_error(&status.cpl))
+		goto out;
+
+	status.done = 0;
+	nvme_ctrlr_cmd_set_feature(ctrlr,
+	    NVME_FEAT_AUTONOMOUS_POWER_STATE_TRANSITION, enable, 0, 0,
+	    0, 0, data, data_size, nvme_completion_poll_cb, &status);
+	nvme_completion_poll(&status);
+	error = nvme_completion_is_error(&status.cpl);
+out:
+	if (error && bootverbose)
+		nvme_printf(ctrlr, "failed to configure APST\n");
+	free(data, M_NVME);
+}
+
 static void
 nvme_ctrlr_configure_int_coalescing(struct nvme_controller *ctrlr)
 {
@@ -1047,6 +1082,7 @@ nvme_ctrlr_start(void *ctrlr_arg, bool resetting)
 	}
 
 	nvme_ctrlr_configure_aer(ctrlr);
+	nvme_ctrlr_configure_apst(ctrlr);
 	nvme_ctrlr_configure_int_coalescing(ctrlr);
 
 	for (i = 0; i < ctrlr->num_io_queues; i++)