Skip to content

Commit

Permalink
hid-xpadneo, rumble: Throttle reprogramming of rumble motors
Browse files Browse the repository at this point in the history
Thanks to @medusalix great findings that the controller crashes are
most probably indeed a timing issue, this commit introduces a
modified rumble protocol which throttles rumble commands sent to the
controller. Also, it seems we cannot rely on `ff-memless` to only send
us commands once per 50ms although that's the proposed time resolution
of the protocol. It may depend on kernel HZ setting.

This commit migrates the rumble worker into a delayed worker. When
programming a rumble motor, we now remember the timing and allow the
next reprogramming only 10ms+ in the future. Further requests will be
buffered and the final command will be replayed to the controller when
the delayed worker finally runs. Care was taken that we do not
accidentally introduce a fixed lag of 10ms for rumble events.

Maybe-fixes: atar-axis#180
Fixes: atar-axis#189
Signed-off-by: Kai Krakow <[email protected]>
  • Loading branch information
kakra committed Jun 27, 2020
1 parent e37ea45 commit 5b8ffcf
Showing 1 changed file with 57 additions and 7 deletions.
64 changes: 57 additions & 7 deletions hid-xpadneo/src/hid-xpadneo.c
Original file line number Diff line number Diff line change
Expand Up @@ -13,6 +13,7 @@
#include <linux/module.h>
#include <linux/power_supply.h>
#include <linux/slab.h>
#include <linux/time.h>
#include "hid-ids.h" /* VENDOR_ID... */

#define DEBUG
Expand Down Expand Up @@ -81,6 +82,9 @@ MODULE_PARM_DESC(quriks,

static DEFINE_IDA(xpadneo_device_id_allocator);

#define XPADNEO_RUMBLE_THROTTLE_DELAY (10L * HZ / 1000)
#define XPADNEO_RUMBLE_THROTTLE_JIFFIES (jiffies + XPADNEO_RUMBLE_THROTTLE_DELAY)

enum {
FF_RUMBLE_NONE = 0x00,
FF_RUMBLE_WEAK = 0x01,
Expand Down Expand Up @@ -135,7 +139,9 @@ struct xpadneo_devdata {

/* buffer for ff_worker */
spinlock_t ff_lock;
struct work_struct ff_worker;
struct delayed_work ff_worker;
unsigned long ff_throttle_until;
bool ff_scheduled;
struct ff_data ff;
struct ff_data ff_shadow;
void *output_report_dmabuf;
Expand Down Expand Up @@ -204,7 +210,8 @@ static const struct usage_map xpadneo_usage_maps[] = {

static void xpadneo_ff_worker(struct work_struct *work)
{
struct xpadneo_devdata *xdata = container_of(work, struct xpadneo_devdata, ff_worker);
struct xpadneo_devdata *xdata =
container_of(to_delayed_work(work), struct xpadneo_devdata, ff_worker);
struct hid_device *hdev = xdata->hdev;
struct ff_report *r = xdata->output_report_dmabuf;
int ret;
Expand All @@ -231,6 +238,9 @@ static void xpadneo_ff_worker(struct work_struct *work)

spin_lock_irqsave(&xdata->ff_lock, flags);

/* let our scheduler know we've been called */
xdata->ff_scheduled = false;

if (unlikely(xdata->quirks & XPADNEO_QUIRK_NO_TRIGGER_RUMBLE)) {
/* do not send these bits if not supported */
r->ff.enable &= ~FF_RUMBLE_TRIGGERS;
Expand Down Expand Up @@ -263,6 +273,12 @@ static void xpadneo_ff_worker(struct work_struct *work)
/* shadow our current rumble values for the next cycle */
memcpy(&xdata->ff_shadow, &xdata->ff, sizeof(xdata->ff));

/*
* throttle next command submission, the firmware doesn't like us to
* send rumble data any faster
*/
xdata->ff_throttle_until = XPADNEO_RUMBLE_THROTTLE_JIFFIES;

spin_unlock_irqrestore(&xdata->ff_lock, flags);

/* do not send these bits if not supported */
Expand All @@ -285,7 +301,8 @@ static int xpadneo_ff_play(struct input_dev *dev, void *data, struct ff_effect *
QUARTER = DIRECTION_LEFT,
};

unsigned long flags;
unsigned long flags, ff_run_at, ff_throttle_until;
long delay_work;
int fraction_TL, fraction_TR, fraction_MAIN;
s32 weak, strong, direction, max_damped, max_unscaled;

Expand Down Expand Up @@ -391,10 +408,40 @@ static int xpadneo_ff_play(struct input_dev *dev, void *data, struct ff_effect *
xdata->ff.magnitude_left = (u8)((max_damped * fraction_TL) / U16_MAX);
xdata->ff.magnitude_right = (u8)((max_damped * fraction_TR) / U16_MAX);

spin_unlock_irqrestore(&xdata->ff_lock, flags);
/* synchronize: is our worker still scheduled? */
if (xdata->ff_scheduled) {
/* the worker is still guarding rumble programming */
hid_notice_once(hdev, "throttling rumble reprogramming\n");
goto unlock_and_return;
}

/* we want to run now but may be throttled */
ff_run_at = jiffies;
ff_throttle_until = xdata->ff_throttle_until;
if (time_before(ff_run_at, ff_throttle_until)) {
/* last rumble was recently executed */
delay_work = (long)ff_throttle_until - (long)ff_run_at;
} else {
/* the firmware is ready */
delay_work = 0;
}

/*
* sanitize: If 0 > delay > 1000ms, something is weird: this
* may happen if the delay between two rumble requests is
* several weeks long
*/
delay_work = min((long)HZ, delay_work);
delay_work = max(0L, delay_work);

/* schedule writing a rumble report to the controller */
schedule_work(&xdata->ff_worker);
if (schedule_delayed_work(&xdata->ff_worker, delay_work))
xdata->ff_scheduled = true;
else
hid_err(hdev, "lost rumble packet\n");

unlock_and_return:
spin_unlock_irqrestore(&xdata->ff_lock, flags);
return 0;
}

Expand Down Expand Up @@ -489,7 +536,7 @@ static int xpadneo_init_ff(struct hid_device *hdev)
struct xpadneo_devdata *xdata = hid_get_drvdata(hdev);
struct input_dev *idev = xdata->idev;

INIT_WORK(&xdata->ff_worker, xpadneo_ff_worker);
INIT_DELAYED_WORK(&xdata->ff_worker, xpadneo_ff_worker);
xdata->output_report_dmabuf = devm_kzalloc(&hdev->dev,
sizeof(struct ff_report), GFP_KERNEL);
if (xdata->output_report_dmabuf == NULL)
Expand All @@ -501,6 +548,9 @@ static int xpadneo_init_ff(struct hid_device *hdev)
if (param_ff_connect_notify)
xpadneo_welcome_rumble(hdev);

/* initialize our rumble command throttle */
xdata->ff_throttle_until = XPADNEO_RUMBLE_THROTTLE_JIFFIES;

input_set_capability(idev, EV_FF, FF_RUMBLE);
return input_ff_create_memless(idev, NULL, xpadneo_ff_play);
}
Expand Down Expand Up @@ -1001,7 +1051,7 @@ static void xpadneo_remove(struct hid_device *hdev)

hid_hw_close(hdev);

cancel_work_sync(&xdata->ff_worker);
cancel_delayed_work_sync(&xdata->ff_worker);

xpadneo_release_device_id(xdata);
hid_hw_stop(hdev);
Expand Down

0 comments on commit 5b8ffcf

Please sign in to comment.