Re: [PATCH 4/4] input: evdev: only wake poll on EV_SYN
From: Henrik Rydberg <hidden>
Date: 2011-03-25 08:31:10
Also in:
lkml
On Tue, Mar 22, 2011 at 06:04:04PM -0700, Jeff Brown wrote:
quoted hunk ↗ jump to hunk
On SMP systems, it is possible for an evdev client blocked on poll() to wake up and read events from the evdev ring buffer at the same rate as they are enqueued. This can result in high CPU usage, particularly for MT devices, because the client ends up reading events one at a time instead of reading complete packets. This patch ensures that the client only wakes from poll() when a complete packet is ready to be read. Signed-off-by: jeffbrown@android.com --- drivers/input/evdev.c | 5 +++-- 1 files changed, 3 insertions(+), 2 deletions(-)diff --git a/drivers/input/evdev.c b/drivers/input/evdev.c index 203ed70..7b6770d 100644 --- a/drivers/input/evdev.c +++ b/drivers/input/evdev.c@@ -73,7 +73,7 @@ static void evdev_pass_event(struct evdev_client *client, } spin_unlock(&client->buffer_lock); - if (event->type == EV_SYN) + if (event->type == EV_SYN && event->code != SYN_MT_REPORT)
It is not clear what should happen at the other SYN events. Maybe event->code == SYN_REPORT instead?
quoted hunk ↗ jump to hunk
kill_fasync(&client->fasync, SIGIO, POLL_IN); }@@ -103,7 +103,8 @@ static void evdev_event(struct input_handle *handle, rcu_read_unlock(); - wake_up_interruptible(&evdev->wait); + if (type == EV_SYN && code != SYN_MT_REPORT) + wake_up_interruptible(&evdev->wait);
Ah, this is a good one. Since the code depends on the same logic being applied in evdev_pass_event as well, a boolean argument to that function would be good. Thanks, Henrik