Reduce memory requirement for MT syncing

Changes the algorithm: before we'd ioctl all axes for all slots, then generate
events for all slots one-by-one.

Now we ioctl the slot state for each axis, copy the new event value into
the device and mark a bitfield that we've updated the value. Then loop through
the slots and generate events where changed.

Side-effect: this makes it easy to check if anything in the slot has updated,
so we can skip empty slot events during sync.

Min memory requirement for the state storage was:
  MAX_SLOTS  * (ABS_MT_CNT + 1) * sizeof(int) = 1980
Min memory requirement now:
  (ABS_MT_CNT + 1) * sizeof(int) + NLONGS((MAX_SLOTS * ABS_MT_CNT) bits) = 544

This is sigsafe code, so this was stack memory. Reducing the requirement
allows us to up MAX_SLOTS in the future if we need to.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>
This commit is contained in:
Peter Hutterer 2014-02-27 13:10:35 +10:00
parent 89e7998fee
commit 24c9ca81f0

View file

@ -552,26 +552,26 @@ static int
sync_mt_state(struct libevdev *dev, int create_events)
{
int rc;
int i;
int axis, slot;
int ioctl_success = 0;
struct mt_state {
int code;
int val[MAX_SLOTS];
} mt_state[ABS_MT_CNT];
} mt_state;
unsigned long slot_update[NLONGS(MAX_SLOTS * ABS_MT_CNT)] = {0};
memset(&mt_state, 0, sizeof(mt_state));
#define AXISBIT(_slot, _axis) (_slot * ABS_MT_CNT + _axis - ABS_MT_MIN)
for (i = ABS_MT_MIN; i <= ABS_MT_MAX; i++) {
int idx;
if (i == ABS_MT_SLOT)
for (axis = ABS_MT_MIN; axis <= ABS_MT_MAX; axis++) {
if (axis == ABS_MT_SLOT)
continue;
if (!libevdev_has_event_code(dev, EV_ABS, i))
if (!libevdev_has_event_code(dev, EV_ABS, axis))
continue;
idx = i - ABS_MT_MIN;
mt_state[idx].code = i;
rc = ioctl(dev->fd, EVIOCGMTSLOTS(sizeof(struct mt_state)), &mt_state[idx]);
memset(&mt_state, 0, sizeof(mt_state));
mt_state.code = axis;
rc = ioctl(dev->fd, EVIOCGMTSLOTS(sizeof(struct mt_state)), &mt_state);
if (rc < 0) {
/* if the first ioctl fails with -EINVAL, chances are the kernel
doesn't support the ioctl. Simply continue */
@ -579,39 +579,49 @@ sync_mt_state(struct libevdev *dev, int create_events)
rc = 0;
} else /* if the second, ... ioctl fails, really fail */
goto out;
} else if (ioctl_success == 0)
ioctl_success = 1;
} else {
if (ioctl_success == 0)
ioctl_success = 1;
for (slot = 0; slot < min(dev->num_slots, MAX_SLOTS); slot++) {
if (*slot_value(dev, slot, axis) == mt_state.val[slot])
continue;
*slot_value(dev, slot, axis) = mt_state.val[slot];
set_bit(slot_update, AXISBIT(slot, axis));
/* note that this slot has updates */
set_bit(slot_update, AXISBIT(slot, ABS_MT_SLOT));
}
}
}
for (i = 0; i < min(dev->num_slots, MAX_SLOTS); i++) {
int j;
for (slot = 0; create_events && slot < min(dev->num_slots, MAX_SLOTS); slot++) {
struct input_event *ev;
if (create_events) {
ev = queue_push(dev);
init_event(dev, ev, EV_ABS, ABS_MT_SLOT, i);
}
if (!bit_is_set(slot_update, AXISBIT(slot, ABS_MT_SLOT)))
continue;
for (j = ABS_MT_MIN; j <= ABS_MT_MAX; j++) {
int jdx = j - ABS_MT_MIN;
ev = queue_push(dev);
init_event(dev, ev, EV_ABS, ABS_MT_SLOT, slot);
if (j == ABS_MT_SLOT)
for (axis = ABS_MT_MIN; axis <= ABS_MT_MAX; axis++) {
if (axis == ABS_MT_SLOT ||
!libevdev_has_event_code(dev, EV_ABS, axis))
continue;
if (!libevdev_has_event_code(dev, EV_ABS, j))
continue;
if (*slot_value(dev, i, j) == mt_state[jdx].val[i])
continue;
if (create_events) {
if (bit_is_set(slot_update, AXISBIT(slot, axis))) {
ev = queue_push(dev);
init_event(dev, ev, EV_ABS, j, mt_state[jdx].val[i]);
init_event(dev, ev, EV_ABS, axis, *slot_value(dev, slot, axis));
}
*slot_value(dev, i, j) = mt_state[jdx].val[i];
}
}
#undef AXISBIT
rc = 0;
out:
return rc ? -errno : 0;