linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Gwendal Grignou <gwendal@chromium.org>
To: briannorris@chromium.org, jic23@kernel.org, knaack.h@gmx.de,
	lars@metafoo.de, pmeerw@pmeerw.net, lee.jones@linaro.org,
	bleung@chromium.org, enric.balletbo@collabora.com,
	dianders@chromium.org, groeck@chromium.org,
	fabien.lahoudere@collabora.com
Cc: linux-kernel@vger.kernel.org, linux-iio@vger.kernel.org,
	Gwendal Grignou <gwendal@chromium.org>
Subject: [PATCH v2 05/18] platform: chrome: cros-ec: record event timestamp in the hard irq
Date: Sun, 20 Oct 2019 22:53:50 -0700	[thread overview]
Message-ID: <20191021055403.67849-6-gwendal@chromium.org> (raw)
In-Reply-To: <20191021055403.67849-1-gwendal@chromium.org>

To improve sensor timestamp precision, given EC and AP are in
different time domains, the AP needs to try to record the exact
moment an event was signalled to the AP by the EC as soon as
possible after it happens.

First thing in the hard irq is the best place for this.

Signed-off-by: Gwendal Grignou <gwendal@chromium.org>
---
Changes in v2:
  Make cros_ec_get_time_ns inline.
  Using ktime_t instead of s64 when dealing with time.
  Added code in ishtp to gather timestamp.

 drivers/platform/chrome/cros_ec.c           | 12 +++++++++++-
 drivers/platform/chrome/cros_ec_ishtp.c     | 17 +++++++++++++++--
 drivers/platform/chrome/cros_ec_lpc.c       |  2 ++
 include/linux/platform_data/cros_ec_proto.h | 16 ++++++++++++++++
 4 files changed, 44 insertions(+), 3 deletions(-)

diff --git a/drivers/platform/chrome/cros_ec.c b/drivers/platform/chrome/cros_ec.c
index 9b2d07422e175..739f3cffe26e3 100644
--- a/drivers/platform/chrome/cros_ec.c
+++ b/drivers/platform/chrome/cros_ec.c
@@ -31,6 +31,15 @@ static struct cros_ec_platform pd_p = {
 	.cmd_offset = EC_CMD_PASSTHRU_OFFSET(CROS_EC_DEV_PD_INDEX),
 };
 
+static irqreturn_t ec_irq_handler(int irq, void *data)
+{
+	struct cros_ec_device *ec_dev = data;
+
+	ec_dev->last_event_time = cros_ec_get_time_ns();
+
+	return IRQ_WAKE_THREAD;
+}
+
 static irqreturn_t ec_irq_thread(int irq, void *data)
 {
 	struct cros_ec_device *ec_dev = data;
@@ -141,7 +150,8 @@ int cros_ec_register(struct cros_ec_device *ec_dev)
 	}
 
 	if (ec_dev->irq) {
-		err = devm_request_threaded_irq(dev, ec_dev->irq, NULL,
+		err = devm_request_threaded_irq(
+				dev, ec_dev->irq, ec_irq_handler,
 				ec_irq_thread, IRQF_TRIGGER_LOW | IRQF_ONESHOT,
 				"chromeos-ec", ec_dev);
 		if (err) {
diff --git a/drivers/platform/chrome/cros_ec_ishtp.c b/drivers/platform/chrome/cros_ec_ishtp.c
index 25ca2c894b4de..5c848f22b44b4 100644
--- a/drivers/platform/chrome/cros_ec_ishtp.c
+++ b/drivers/platform/chrome/cros_ec_ishtp.c
@@ -200,13 +200,14 @@ static int ish_send(struct ishtp_cl_data *client_data,
  * process_recv() - Received and parse incoming packet
  * @cros_ish_cl: Client instance to get stats
  * @rb_in_proc: Host interface message buffer
+ * @timestamp: Timestamp of when parent callback started
  *
  * Parse the incoming packet. If it is a response packet then it will
  * update per instance flags and wake up the caller waiting to for the
  * response. If it is an event packet then it will schedule event work.
  */
 static void process_recv(struct ishtp_cl *cros_ish_cl,
-			 struct ishtp_cl_rb *rb_in_proc)
+			 struct ishtp_cl_rb *rb_in_proc, ktime_t timestamp)
 {
 	size_t data_len = rb_in_proc->buf_idx;
 	struct ishtp_cl_data *client_data =
@@ -295,6 +296,11 @@ static void process_recv(struct ishtp_cl *cros_ish_cl,
 		break;
 
 	case CROS_MKBP_EVENT:
+		/*
+		 * Set timestamp from beginning of function since we actually
+		 * got an incoming MKBP event
+		 */
+		client_data->ec_dev->last_event_time = timestamp;
 		/* The event system doesn't send any data in buffer */
 		schedule_work(&client_data->work_ec_evt);
 
@@ -322,10 +328,17 @@ static void ish_event_cb(struct ishtp_cl_device *cl_device)
 {
 	struct ishtp_cl_rb *rb_in_proc;
 	struct ishtp_cl	*cros_ish_cl = ishtp_get_drvdata(cl_device);
+	ktime_t timestamp;
+
+	/*
+	 * Take timestamp as close to hardware interrupt as possible for sensor
+	 * timestamps.
+	 */
+	timestamp = cros_ec_get_time_ns();
 
 	while ((rb_in_proc = ishtp_cl_rx_get_rb(cros_ish_cl)) != NULL) {
 		/* Decide what to do with received data */
-		process_recv(cros_ish_cl, rb_in_proc);
+		process_recv(cros_ish_cl, rb_in_proc, timestamp);
 	}
 }
 
diff --git a/drivers/platform/chrome/cros_ec_lpc.c b/drivers/platform/chrome/cros_ec_lpc.c
index 7d10d909435ff..3c77496e164da 100644
--- a/drivers/platform/chrome/cros_ec_lpc.c
+++ b/drivers/platform/chrome/cros_ec_lpc.c
@@ -313,6 +313,8 @@ static void cros_ec_lpc_acpi_notify(acpi_handle device, u32 value, void *data)
 {
 	struct cros_ec_device *ec_dev = data;
 
+	ec_dev->last_event_time = cros_ec_get_time_ns();
+
 	if (ec_dev->mkbp_event_supported &&
 	    cros_ec_get_next_event(ec_dev, NULL) > 0)
 		blocking_notifier_call_chain(&ec_dev->event_notifier, 0,
diff --git a/include/linux/platform_data/cros_ec_proto.h b/include/linux/platform_data/cros_ec_proto.h
index 691f9e953a96a..b183024fef1f6 100644
--- a/include/linux/platform_data/cros_ec_proto.h
+++ b/include/linux/platform_data/cros_ec_proto.h
@@ -122,6 +122,8 @@ struct cros_ec_command {
  * @event_data: Raw payload transferred with the MKBP event.
  * @event_size: Size in bytes of the event data.
  * @host_event_wake_mask: Mask of host events that cause wake from suspend.
+ * @last_event_time: exact time from the hard irq when we got notified of
+ *     a new event.
  * @ec: The platform_device used by the mfd driver to interface with the
  *      main EC.
  * @pd: The platform_device used by the mfd driver to interface with the
@@ -162,6 +164,7 @@ struct cros_ec_device {
 	int event_size;
 	u32 host_event_wake_mask;
 	u32 last_resume_result;
+	ktime_t last_event_time;
 
 	/* The platform devices used by the mfd driver */
 	struct platform_device *ec;
@@ -210,4 +213,17 @@ int cros_ec_check_features(struct cros_ec_dev *ec, int feature);
 
 int cros_ec_get_sensor_count(struct cros_ec_dev *ec);
 
+/**
+ * cros_ec_get_time_ns - Return time in ns.
+ *
+ * This is the function used to record the time for last_event_time in struct
+ * cros_ec_device during the hard irq.
+ *
+ * Return: ktime_t format since boot.
+ */
+static inline ktime_t cros_ec_get_time_ns(void)
+{
+	return ktime_get_boottime_ns();
+}
+
 #endif /* __LINUX_CROS_EC_PROTO_H */
-- 
2.23.0.866.gb869b98d4c-goog


  parent reply	other threads:[~2019-10-21  5:54 UTC|newest]

Thread overview: 42+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-10-21  5:53 [PATCH v2 00/18] cros_ec: Add sensorhub driver and FIFO processing*** SUBJECT HERE Gwendal Grignou
2019-10-21  5:53 ` [PATCH v2 01/18] platform: chrome: Put docs with the code Gwendal Grignou
2019-10-21 15:31   ` Jonathan Cameron
2019-10-24 17:03     ` Enric Balletbo i Serra
2019-10-21  5:53 ` [PATCH v2 02/18] mfd: cros_ec: Add sensor_count and make check_features public Gwendal Grignou
2019-10-21 15:12   ` Enric Balletbo i Serra
2019-10-21 15:49   ` Jonathan Cameron
2019-11-01  9:00   ` Lee Jones
2019-10-21  5:53 ` [PATCH v2 03/18] platform: cros_ec: Add cros_ec_sensor_hub driver Gwendal Grignou
2019-10-21 15:59   ` Jonathan Cameron
2019-10-22  8:32     ` Enric Balletbo i Serra
2019-10-21  5:53 ` [PATCH v2 04/18] platform/mfd:iio: cros_ec: Register sensor through sensorhub Gwendal Grignou
2019-10-21 16:00   ` Jonathan Cameron
2019-10-22  9:39     ` Enric Balletbo i Serra
2019-10-21  5:53 ` Gwendal Grignou [this message]
2019-10-22 10:03   ` [PATCH v2 05/18] platform: chrome: cros-ec: record event timestamp in the hard irq Enric Balletbo i Serra
2019-10-21  5:53 ` [PATCH v2 06/18] platform: chrome: cros_ec: Do not attempt to register a non-positive IRQ number Gwendal Grignou
2019-10-22 10:05   ` Enric Balletbo i Serra
2019-10-21  5:53 ` [PATCH v2 07/18] platform: chrome: cros_ec: handle MKBP more events flag Gwendal Grignou
2019-10-21 16:07   ` Jonathan Cameron
2019-10-21  5:53 ` [PATCH v2 08/18] Revert "Input: cros_ec_keyb - add back missing mask for event_type" Gwendal Grignou
2019-10-21  5:53 ` [PATCH v2 09/18] Revert "Input: cros_ec_keyb: mask out extra flags in event_type" Gwendal Grignou
2019-10-21  5:53 ` [PATCH v2 10/18] platform: chrome: sensorhub: Add FIFO support Gwendal Grignou
2019-10-21 16:27   ` Jonathan Cameron
2019-10-21 22:09     ` Gwendal Grignou
2019-10-22 10:42       ` Jonathan Cameron
2019-10-21  5:53 ` [PATCH v2 11/18] platform: chrome: sensorhub: Add code to spread timestmap Gwendal Grignou
2019-10-21  5:53 ` [PATCH v2 12/18] platform: chrome: sensorhub: Add median filter Gwendal Grignou
2019-10-21  5:53 ` [PATCH v2 13/18] iio: cros_ec: Move function description to .c file Gwendal Grignou
2019-10-21 16:33   ` Jonathan Cameron
2019-10-21  5:53 ` [PATCH v2 14/18] iio: cros_ec: Register to cros_ec_sensorhub when EC supports FIFO Gwendal Grignou
2019-10-21 16:38   ` Jonathan Cameron
2019-10-21  5:54 ` [PATCH v2 15/18] iio: cros_ec: Remove pm function Gwendal Grignou
2019-10-21 16:39   ` Jonathan Cameron
2019-10-21  5:54 ` [PATCH v2 16/18] iio: cros_ec: Expose hwfifo_timeout Gwendal Grignou
2019-10-21 16:42   ` Jonathan Cameron
2019-10-21  5:54 ` [PATCH v2 17/18] iio: cros_ec: Report hwfifo_watermark_max Gwendal Grignou
2019-10-21 16:42   ` Jonathan Cameron
2019-10-21  5:54 ` [PATCH v2 18/18] iio: cros_ec: Use Hertz as unit for sampling frequency Gwendal Grignou
2019-10-21 16:45   ` Jonathan Cameron
2019-10-22 19:11     ` Gwendal Grignou
2019-10-21 15:29 ` [PATCH v2 00/18] cros_ec: Add sensorhub driver and FIFO processing*** SUBJECT HERE Jonathan Cameron

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20191021055403.67849-6-gwendal@chromium.org \
    --to=gwendal@chromium.org \
    --cc=bleung@chromium.org \
    --cc=briannorris@chromium.org \
    --cc=dianders@chromium.org \
    --cc=enric.balletbo@collabora.com \
    --cc=fabien.lahoudere@collabora.com \
    --cc=groeck@chromium.org \
    --cc=jic23@kernel.org \
    --cc=knaack.h@gmx.de \
    --cc=lars@metafoo.de \
    --cc=lee.jones@linaro.org \
    --cc=linux-iio@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=pmeerw@pmeerw.net \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).