Re: [PATCH 2/2] xen/input: add multi-touch support

From: Oleksandr Andrushchenko
Date: Mon Jun 19 2017 - 06:00:50 EST


Ping

On 06/08/2017 09:45 AM, Oleksandr Andrushchenko wrote:
Hi, Dmitry!

On 06/07/2017 07:56 PM, Dmitry Torokhov wrote:
On Wed, May 31, 2017 at 12:06:56PM +0300, Oleksandr Andrushchenko wrote:
Hi, Dmitry!

On 05/30/2017 07:37 PM, Dmitry Torokhov wrote:
On Tue, May 30, 2017 at 03:50:20PM +0300, Oleksandr Andrushchenko wrote:
Hi, Dmitry!

On 05/30/2017 08:51 AM, Dmitry Torokhov wrote:
On Fri, Apr 21, 2017 at 09:40:36AM +0300, Oleksandr Andrushchenko wrote:
Hi, Dmitry!

On 04/21/2017 05:10 AM, Dmitry Torokhov wrote:
Hi Oleksandr,

On Thu, Apr 13, 2017 at 02:38:04PM +0300, Oleksandr Andrushchenko wrote:
From: Oleksandr Andrushchenko <oleksandr_andrushchenko@xxxxxxxx>

Extend xen_kbdfront to provide multi-touch support
to unprivileged domains.

Signed-off-by: Oleksandr Andrushchenko <oleksandr_andrushchenko@xxxxxxxx>
---
drivers/input/misc/xen-kbdfront.c | 142 +++++++++++++++++++++++++++++++++++++-
1 file changed, 140 insertions(+), 2 deletions(-)

diff --git a/drivers/input/misc/xen-kbdfront.c b/drivers/input/misc/xen-kbdfront.c
index 01c27b4c3288..e5d064aaa237 100644
--- a/drivers/input/misc/xen-kbdfront.c
+++ b/drivers/input/misc/xen-kbdfront.c
@@ -17,6 +17,7 @@
#include <linux/errno.h>
#include <linux/module.h>
#include <linux/input.h>
+#include <linux/input/mt.h>
#include <linux/slab.h>
#include <asm/xen/hypervisor.h>
@@ -34,11 +35,14 @@
struct xenkbd_info {
struct input_dev *kbd;
struct input_dev *ptr;
+ struct input_dev *mtouch;
struct xenkbd_page *page;
int gref;
int irq;
struct xenbus_device *xbdev;
char phys[32];
+ /* current MT slot/contact ID we are injecting events in */
+ int mtouch_cur_contact_id;
};
enum { KPARAM_X, KPARAM_Y, KPARAM_CNT };
@@ -47,6 +51,12 @@ module_param_array(ptr_size, int, NULL, 0444);
MODULE_PARM_DESC(ptr_size,
"Pointing device width, height in pixels (default 800,600)");
+enum { KPARAM_MT_X, KPARAM_MT_Y, KPARAM_MT_CNT };
+static int mtouch_size[KPARAM_MT_CNT] = { XENFB_WIDTH, XENFB_HEIGHT };
+module_param_array(mtouch_size, int, NULL, 0444);
+MODULE_PARM_DESC(ptr_size,
+ "Multi-touch device width, height in pixels (default 800,600)");
+
Why do you need separate module parameters for multi-touch device?
please see below
static int xenkbd_remove(struct xenbus_device *);
static int xenkbd_connect_backend(struct xenbus_device *, struct xenkbd_info *);
static void xenkbd_disconnect_backend(struct xenkbd_info *);
@@ -100,6 +110,60 @@ static irqreturn_t input_handler(int rq, void *dev_id)
input_report_rel(dev, REL_WHEEL,
-event->pos.rel_z);
break;
+ case XENKBD_TYPE_MTOUCH:
+ dev = info->mtouch;
+ if (unlikely(!dev))
+ break;
+ if (unlikely(event->mtouch.contact_id !=
+ info->mtouch_cur_contact_id)) {
Why is this unlikely? Does contact ID changes once in 1000 packets or
even less?
Mu assumption was that regardless of the fact that we are multi-touch
device still single touches will come in more frequently
But I can remove *unlikely* if my assumption is not correct
I think the normal expectation is that "unlikely" is supposed for
something that happens once in a blue moon, so I'd rather remove it.

agree, removed "unlikely"
+ info->mtouch_cur_contact_id =
+ event->mtouch.contact_id;
+ input_mt_slot(dev, event->mtouch.contact_id);
+ }
+ switch (event->mtouch.event_type) {
+ case XENKBD_MT_EV_DOWN:
+ input_mt_report_slot_state(dev, MT_TOOL_FINGER,
+ true);
Should we establish tool event? We have MT_TOOL_PEN, etc.
I think that for multi-touch MT_TOOL_FINGER is enough
any reason we would also want MT_TOOL_PEN here?
Why would not you? Let's say you have a drawing application running in
guest that can make use of tool types. Why would not you want to tell it
that the tool user is currently using is in fact a pen and not finger?
But it is a finger :) we are multi-touch, not multi pen
So for tablets that support both touch and stylus you would export them
as 2 separate devices?
this could be done in different ways, but please see on
pen support below
Besides, that, if I am about to implement pen support
(which I still not convinced we really need), how will I
do that?
I do not know what you have on the backend side, but roughly speaking if
you detect a pen/stylus you let your guest know that the contact is not
a finger, but pen. How you plumb it through is up to you.
we do not detect pen, only finger at the moment
and the existing protocol has no means to tell
type of the tool used, everything is supposed to
be "finger", so front-end has no possibility to
tell one tool from another
My understanding is that I need 2 different slots to report
the same coordinates for finger and pen. This is because
input_mt_report_slot_state has a check that if tool has
changed for the current slot then a new tracking ID is set.
Do I also need to allocate twice more slots, so I can
report 2 * num_contacts events simultaneously (one for finger
and another for pen)?
That said, I believe we can start with multi-touch support
and if need be then add pen support as a separate change,
does that make sense for you?
(I guess MT_TOOL_PALM is not appropriate anyways)
Depends on if you do straight pass-through from the host side or not. If
you stack does palm rejection before passing the data through then you
would not see MT_TOOL_PALM in guest.
the protocol used between guest and host is a generic one,
not using Linux types/constants/events.
It does not have to use Linux types to support the concept of different
tools.
agree
So, no PALM/TOOL support is in place
OK, that is fair. The pen support is definitely not a hard requirement.
so, can we live with finger at this point, no pen?
I was just wondering if you considered or have plans for adding that.
well, honestly, we do not need pen at the moment,
but we did add multi-touch support into Xen protocols
and if pen required it can be done as a separate
change to both protocol and front/back ends
Or
if you want to review the protocol so it can be easily added in the
future. For example you could have tool type to be part of
XENKBD_MT_EV_DOWN event.
protocol did a long way to get into Xen/Kernel... :)
of course, this can be done, but I would prefer it is
added when it is needed, not only that we have this functionality
+ input_event(dev, EV_ABS, ABS_MT_POSITION_X,
+ event->mtouch.u.pos.abs_x);
+ input_event(dev, EV_ABS, ABS_MT_POSITION_Y,
+ event->mtouch.u.pos.abs_y);
+ input_event(dev, EV_ABS, ABS_X,
+ event->mtouch.u.pos.abs_x);
+ input_event(dev, EV_ABS, ABS_Y,
+ event->mtouch.u.pos.abs_y);
+ break;
+ case XENKBD_MT_EV_UP:
+ input_mt_report_slot_state(dev, MT_TOOL_FINGER,
+ false);
+ break;
+ case XENKBD_MT_EV_MOTION:
+ input_event(dev, EV_ABS, ABS_MT_POSITION_X,
+ event->mtouch.u.pos.abs_x);
+ input_event(dev, EV_ABS, ABS_MT_POSITION_Y,
+ event->mtouch.u.pos.abs_y);
+ input_event(dev, EV_ABS, ABS_X,
+ event->mtouch.u.pos.abs_x);
+ input_event(dev, EV_ABS, ABS_Y,
+ event->mtouch.u.pos.abs_y);
+ break;
+ case XENKBD_MT_EV_SYN:
+ input_mt_sync_frame(dev);
+ break;
+ case XENKBD_MT_EV_SHAPE:
+ input_event(dev, EV_ABS, ABS_MT_TOUCH_MAJOR,
+ event->mtouch.u.shape.major);
+ input_event(dev, EV_ABS, ABS_MT_TOUCH_MINOR,
+ event->mtouch.u.shape.minor);
+ break;
+ case XENKBD_MT_EV_ORIENT:
+ input_event(dev, EV_ABS, ABS_MT_ORIENTATION,
+ event->mtouch.u.orientation);
+ break;
+ }
+ /* only report syn when requested */
+ if (event->mtouch.event_type != XENKBD_MT_EV_SYN)
+ dev = NULL;
}
if (dev)
input_sync(dev);
@@ -115,9 +179,9 @@ static int xenkbd_probe(struct xenbus_device *dev,
const struct xenbus_device_id *id)
{
int ret, i;
- unsigned int abs;
+ unsigned int abs, touch;
struct xenkbd_info *info;
- struct input_dev *kbd, *ptr;
+ struct input_dev *kbd, *ptr, *mtouch;
info = kzalloc(sizeof(*info), GFP_KERNEL);
if (!info) {
@@ -152,6 +216,17 @@ static int xenkbd_probe(struct xenbus_device *dev,
}
}
+ touch = xenbus_read_unsigned(dev->nodename,
+ XENKBD_FIELD_FEAT_MTOUCH, 0);
+ if (touch) {
+ ret = xenbus_write(XBT_NIL, dev->nodename,
+ XENKBD_FIELD_REQ_MTOUCH, "1");
+ if (ret) {
+ pr_warning("xenkbd: can't request multi-touch");
+ touch = 0;
+ }
+ }
+
/* keyboard */
kbd = input_allocate_device();
if (!kbd)
@@ -208,6 +283,67 @@ static int xenkbd_probe(struct xenbus_device *dev,
}
info->ptr = ptr;
+ /* multi-touch device */
+ if (touch) {
+ int num_cont, width, height;
+
+ mtouch = input_allocate_device();
+ if (!mtouch)
+ goto error_nomem;
+
+ num_cont = xenbus_read_unsigned(info->xbdev->nodename,
+ XENKBD_FIELD_MT_NUM_CONTACTS,
+ 1);
Should we refuse MT devices with number of contacts less than 2?
we can, but I see no harm in 1. what is more, this may
allow guests to emulate more pointing devices
but, if you insist, then I will add appropriate code to
reject if number of contacts is less then 2
The question is if you are planning to keep the single-touch interface
or you will migrate everything to multi-touch.
I will keep single touch as legacy, but in our use-cases
we are more focusing on using multi-touch devices.
but even with number of contacts == 1 it can still be
useful as it gives more flexibility in configuring guest OS
If you insist that num contacts == 1 must be removed,
please let me know and I will handle that
+ width = xenbus_read_unsigned(info->xbdev->nodename,
+ XENKBD_FIELD_MT_WIDTH,
+ XENFB_WIDTH);
+ height = xenbus_read_unsigned(info->xbdev->nodename,
+ XENKBD_FIELD_MT_HEIGHT,
+ XENFB_HEIGHT);
Curious why you need separate parameters here too...
This is because mt parameters are different from ptr
in a way that they are configurable per front driver's
instance rather than per backend, e.g. in XenStore:

/local/domain/0/backend/vkbd/1/0/width = "1920"
/local/domain/0/backend/vkbd/1/0/height = "1080"

/local/domain/1/device/vkbd/0/multi-touch-width = "1920"
/local/domain/1/device/vkbd/0/multi-touch-height = "1080"
/local/domain/1/device/vkbd/0/multi-touch-num-contacts = "10"

/local/domain/1/device/vkbd/1/multi-touch-width = "800"
/local/domain/1/device/vkbd/1/multi-touch-height = "600"
/local/domain/1/device/vkbd/1/multi-touch-num-contacts = "3"

The main reason for such configuration is that you can
configure multiple mt input devices even for the same guest
with different resolutions which may not match those
configured for ptr.
(In my use-case I use new displif protocol [1] in conjunction
with mt input devices and the corresponding backend is not
QEMU's xenfb)
I see.

As to module parameters, I added those to be consistent with
ptr device. Do you think we can live without them and
do you want me to remove them?
Yes, I think we better. I am also confused by the way you are handling
the module parameters. It looks to me you update them with data passed
>from the backend, but never use the data...
I have removed module parameters (the only use of those
was to be able to see configured width and height on
guest side, but this is minor)
evtest would show it to you. Or you can query input device yourself
(EVIOCGABS iotcl).
yes, if embedded system (which is my target) has evtest
but it definitely does have ioctl though
+
+ mtouch->name = "Xen Virtual Multi-touch";
+ mtouch->phys = info->phys;
+ mtouch->id.bustype = BUS_PCI;
+ mtouch->id.vendor = 0x5853;
+ mtouch->id.product = 0xfffd;
+
+ __set_bit(EV_ABS, mtouch->evbit);
+ __set_bit(EV_KEY, mtouch->evbit);
+ __set_bit(BTN_TOUCH, mtouch->keybit);
Please make it
input_set_capability(mtouch, EV_KEY, BTN_TOUCH);

and drop all __set_bit()s.
done, thank you
+
+ input_set_abs_params(mtouch, ABS_X,
+ 0, width, 0, 0);
+ input_set_abs_params(mtouch, ABS_Y,
+ 0, height, 0, 0);
+ input_set_abs_params(mtouch, ABS_PRESSURE,
+ 0, 255, 0, 0);
+
+ input_set_abs_params(mtouch, ABS_MT_TOUCH_MAJOR,
+ 0, 255, 0, 0);
+ input_set_abs_params(mtouch, ABS_MT_POSITION_X,
+ 0, width, 0, 0);
+ input_set_abs_params(mtouch, ABS_MT_POSITION_Y,
+ 0, height, 0, 0);
+ input_set_abs_params(mtouch, ABS_MT_PRESSURE,
+ 0, 255, 0, 0);
+
+ input_mt_init_slots(mtouch, num_cont, 0);
We need error handling here.
done
Also, it would be nice if we set INPUT_MT_*
flags here, so that userspace had better chance of figuring how to
handle the device.
done, I will use INPUT_MT_DIRECT | INPUT_MT_DROP_UNUSED
Does that mean that your backend does not reliably report release of
contacts?
there is a ring buffer between host and guest,
so there is always a possibility (rather small I believe)
that the buffer overruns. Do you think I need INPUT_MT_DROP_UNUSED or
we can live without it?
Again, it depends on your backend behavior. Do you report all slots
repeatedly for every packet or you report only changed slots?
we report events repeatedly, so I think we can live
w/o _DROP_UNUSED
Thanks.

Thanks.

+
+ mtouch_size[KPARAM_MT_X] = width;
+ mtouch_size[KPARAM_MT_Y] = height;
+ info->mtouch_cur_contact_id = -1;
+
+ ret = input_register_device(mtouch);
+ if (ret) {
+ input_free_device(mtouch);
+ xenbus_dev_fatal(info->xbdev, ret,
+ "input_register_device(mtouch)");
+ goto error;
+ }
+ info->mtouch_cur_contact_id = -1;
+ info->mtouch = mtouch;
+ }
+
ret = xenkbd_connect_backend(dev, info);
if (ret < 0)
goto error;
@@ -240,6 +376,8 @@ static int xenkbd_remove(struct xenbus_device *dev)
input_unregister_device(info->kbd);
if (info->ptr)
input_unregister_device(info->ptr);
+ if (info->mtouch)
+ input_unregister_device(info->mtouch);
free_page((unsigned long)info->page);
kfree(info);
return 0;
--
2.7.4

Thanks.

For your convenience I am attaching the changes I am about
to put into v1 of the series:
- remove unlikely
- remove module parameters
- error handling for input_mt_init_slots
- let userspace better chance of figuring how to handle the device

Thank you,
Oleksandr
From e76506c55846e2bb4ccbafa430642e368643e51d Mon Sep 17 00:00:00 2001
From: Oleksandr Andrushchenko <oleksandr_andrushchenko@xxxxxxxx>
Date: Tue, 30 May 2017 14:49:58 +0300
Subject: [PATCH] Fix: remove unlikely Fix: remove module paramters Fix: error
handling for input_mt_init_slots Fix: let userspace better chance of figuring
how to handle the device

Signed-off-by: Oleksandr Andrushchenko <oleksandr_andrushchenko@xxxxxxxx>
---
drivers/input/misc/xen-kbdfront.c | 21 ++++++++++-----------
1 file changed, 10 insertions(+), 11 deletions(-)

diff --git a/drivers/input/misc/xen-kbdfront.c b/drivers/input/misc/xen-kbdfront.c
index 8266ef948a06..273d786a19cd 100644
--- a/drivers/input/misc/xen-kbdfront.c
+++ b/drivers/input/misc/xen-kbdfront.c
@@ -51,12 +51,6 @@ module_param_array(ptr_size, int, NULL, 0444);
MODULE_PARM_DESC(ptr_size,
"Pointing device width, height in pixels (default 800,600)");
-enum { KPARAM_MT_X, KPARAM_MT_Y, KPARAM_MT_CNT };
-static int mtouch_size[KPARAM_MT_CNT] = { XENFB_WIDTH, XENFB_HEIGHT };
-module_param_array(mtouch_size, int, NULL, 0444);
-MODULE_PARM_DESC(ptr_size,
- "Multi-touch device width, height in pixels (default 800,600)");
-
static int xenkbd_remove(struct xenbus_device *);
static int xenkbd_connect_backend(struct xenbus_device *, struct xenkbd_info *);
static void xenkbd_disconnect_backend(struct xenkbd_info *);
@@ -114,8 +108,8 @@ static irqreturn_t input_handler(int rq, void *dev_id)
dev = info->mtouch;
if (unlikely(!dev))
break;
- if (unlikely(event->mtouch.contact_id !=
- info->mtouch_cur_contact_id)) {
+ if (event->mtouch.contact_id !=
+ info->mtouch_cur_contact_id) {
info->mtouch_cur_contact_id =
event->mtouch.contact_id;
input_mt_slot(dev, event->mtouch.contact_id);
@@ -327,10 +321,15 @@ static int xenkbd_probe(struct xenbus_device *dev,
input_set_abs_params(mtouch, ABS_MT_PRESSURE,
0, 255, 0, 0);
- input_mt_init_slots(mtouch, num_cont, 0);
+ ret = input_mt_init_slots(mtouch, num_cont,
+ INPUT_MT_DIRECT | INPUT_MT_DROP_UNUSED);
+ if (ret) {
+ input_free_device(mtouch);
+ xenbus_dev_fatal(info->xbdev, ret,
+ "input_mt_init_slots");
+ goto error;
+ }
- mtouch_size[KPARAM_MT_X] = width;
- mtouch_size[KPARAM_MT_Y] = height;
info->mtouch_cur_contact_id = -1;
ret = input_register_device(mtouch);
--
2.7.4

Thank you,
Oleksandr
Dmitry, thank you for comments
The bottom line I see is:
- no support for PEN tool at the moment
- num contacts == 1 is OK
- I will not use INPUT_MT_DROP_UNUSED

If the above is ok to you, then I will send another version of the
series (BTW, can I use your RB for the first patch which
removes hard-codes?)

Thank you,
Oleksandr