On Tue, Jan 5, 2010 at 1:29 PM, hedwin <hedwin.kon...@gmail.com> wrote:
>
> So I know that a working mouse depends on being able to open the files in
> '/sys/power'.
>
No, the mouse is always working. A correct PM state is based on the
access to /sys/power. The reason for the mouse events get dropped is
the PM said screen is off and WM drops the mouse event based on the
predefined policy. It has nothing to do with mouse. If your screen is
off, there are not reason to process the mouse event.
> After further looking a QEMU I noticed that the ownership of the /sys/power
> files are changed to what is indicated in init.rc but if I monitor these
> over
> time that after a while all of the ownership of the '/sys/power' files
> change to root:root ownership thus preventing android from being allowed to
> read
> these files and as a result the mouse doesn't work anymore.
>
Sounds like a driver refreshed these file entries. Maybe your QEMU
faked some ACPI or PM events that triggered the problem.
> Also tried VirtualBox and this one doesn't seem to have this problem mouse
> works fine.
As expected.
>
> Hedwin
>
> On Wed, Dec 30, 2009 at 9:47 PM, hedwin <hedwin.kon...@gmail.com> wrote:
>>
>> No, don't see OLD_PATHS either. Also searched the kernel nothing there as
>> well. Guess power.c never got cleaned up
>>
>> Would also assume it to work as well. Will try VirtualBox see if it does
>> the same or different.
>>
>> Hedwin
>>
>> On Wed, Dec 30, 2009 at 9:24 PM, Yi Sun <beyo...@gmail.com> wrote:
>>>
>>> > When I change the sys file properties of the files in OLD_PATHS and
>>> > NEW_PATHS to 666 mouse movement events are processed. What??? yes
>>> > (OLD_PATHS and NEW_PATHS are defined
>>> > in /hardware/libhardware_legacy/power/power.c)
>>> >
>>> The power.c runs as system I think, and the files in /sys/power has rw
>>> for the system group. It should be able to be accessed by power.c
>>>
>>> BTW-- on all the x86 systems I ever used, I did not see OLD_PATHS in
>>> sysfs. Do you see it in your env?
>>> > So I wonder how can this work anyway on any machine or is it just
>>> > QEMU ..............
>>> No, it seems that only you have the problem.
>>> Really, I only seen the same problem when we did not enable the wakelock
>>> driver in the kernel. And that was 10 months ago.
>>>
>>> Yi
>>> On Wed, 2009-12-30 at 20:49 +0100, hedwin wrote:
>>> > Hi,
>>> >
>>> > Dug a little deeper now.
>>> >
>>> > Handling mouse movement indirectly depends on the power state of
>>> > android. Eh what ????
>>> >
>>> > Mouse events are handled by the InputDeviceReader. This thread calls
>>> > method 'preprocessEvent' in the WindowManagerService. The method
>>> > itself then calls mPowerManager.screenIsOn. This method returns false
>>> > for reason I explain later. This eventually results in preprocessEvent
>>> > to return 0 and as a result the mouse movement events are not handled.
>>> > Therefor not moving mouse pointer.
>>> >
>>> > Ok
>>> >
>>> > Now to the why screenIsOn return false. When android starts it will
>>> > eventually execute initInThread in PowerManagementService. At the end
>>> > setPowerState(ALL_BRIGHT) is called. So far
>>> > so good. The method setPowerState will call
>>> > Power.setScreenState(true). Still going ok. Power.setScreenState is a
>>> > function implemented in /base/core/jni/android_os_power.cpp.
>>> > The function Power.setScreenState will eventually call function
>>> > set_screen_state implemented
>>> > in /hardware/libhardware_legacy/power/power.c. The function
>>> > set_screen_state will at the begining
>>> > call initialize_fds. And here is where it goes wrong because
>>> > initialize_fds cannot open the files mentioned in the array OLD_PATHS
>>> > and NEW_PATHS. The reason why it cannot
>>> > open the files is because the filesystem properties are set to 660 in
>>> > init.rc. For this reason the power change never takes places and
>>> > indirectly prevents the move movement events from
>>> > being handled.
>>> >
>>> > When I change the sys file properties of the files in OLD_PATHS and
>>> > NEW_PATHS to 666 mouse movement events are processed. What??? yes
>>> > (OLD_PATHS and NEW_PATHS are defined
>>> > in /hardware/libhardware_legacy/power/power.c)
>>> >
>>> > So I wonder how can this work anyway on any machine or is it just
>>> > QEMU ..............
>>> >
>>> > Hedwin
>>> >
>>> > On Sat, Dec 26, 2009 at 8:55 PM, Yi Sun <beyo...@gmail.com> wrote:
>>> > If you use the latest qemu, would you still need these
>>> > changes?
>>> > Yi
>>> >
>>> > On Sat, 2009-12-26 at 20:14 +0100, hedwin wrote:
>>> > > The serial is just added to see the kernel boot an have
>>> > access to the
>>> > > console later on.
>>> > > I use a slight older qemu, the one supplied with fedora.
>>> > Newer qemu
>>> > > versions use usb per
>>> > > default. That might be the reason why you did no have to
>>> > provide the
>>> > > usb arguments.
>>> > >
>>> > > Hedwin
>>> > >
>>> > > On Sat, Dec 26, 2009 at 7:39 PM, Yi Sun <beyo...@gmail.com>
>>> > wrote:
>>> > > Thanks for the help.
>>> > > FYI-- in my env, I did not really add "-serial stdio
>>> > -usb
>>> > > -usbdevice
>>> > > mouse" for my usb mouse to work with qemu. I'm not
>>> > sure why
>>> > > you need it.
>>> > > I may have to check on the qemu manual.
>>> > > Yi
>>> > >
>>> > > On Sat, 2009-12-26 at 19:32 +0100, hedwin wrote:
>>> > > > agree, so consider it a workaround. I will take a
>>> > further
>>> > > look into
>>> > > > this
>>> > > >
>>> > > > Hedwin
>>> > > >
>>> > > > On Sat, Dec 26, 2009 at 7:29 PM, Yi Sun
>>> > <beyo...@gmail.com>
>>> > > wrote:
>>> > > > BTW-- the patch does not look like a fix
>>> > but a hack
>>> > > when we
>>> > > > don't know
>>> > > > what really went wrong. Would it possible
>>> > for you to
>>> > > find out
>>> > > > why the
>>> > > > Android thinks that the screen is not on
>>> > in your
>>> > > case. We may
>>> > > > have a bug
>>> > > > somewhere else.
>>> > > > boolean screenIsOn() {
>>> > > > synchronized (mLocks) {
>>> > > > + mPowerState |= SCREEN_ON_BIT;
>>> > > > return (mPowerState &
>>> > SCREEN_ON_BIT) !=
>>> > > 0;
>>> > > > }
>>> > > > }
>>> > > >
>>> > > >
>>> > > > On Sat, 2009-12-26 at 19:13 +0100, hedwin
>>> > wrote:
>>> > > >
>>> > > >
>>> > > > > www.android-x86.org
>>> > > > >
>>> > > > > Reason it doesn't work is that android
>>> > thinks the
>>> > > screen is
>>> > > > off.
>>> > > > > Issue 62 contains the solution to get it
>>> > working.
>>> > > > >
>>> > > > > Hedwin
>>> > > > >
>>> > > > > On Sat, Dec 26, 2009 at 6:52 PM, Yi Sun
>>> > > <beyo...@gmail.com>
>>> > > > wrote:
>>> > > > > The mouse should work on
>>> > Android-x86.org
>>> > > out of box.
>>> > > > Where did
>>> > > > > you get
>>> > > > > your source code for your
>>> > android-x86?
>>> > > > > Yi
>>> > > > >
>>> > > > > On Wed, 2009-12-23 at 12:44
>>> > +0100, hedwin
>>> > > wrote:
>>> > > > > > Meant to say is that it did
>>> > not work
>>> > > because
>>> > > > android thinks
>>> > > > > the screen is off.
>>> > > > > >
>>> > > > > > Will provide the details when
>>> > I am back
>>> > > home.
>>> > > > > >
>>> > > > > > Hedwin
>>> > > > > >
>>> > > > > > On Wed, Dec 23, 2009 at 12:43
>>> > PM, hedwin
>>> > > > > <hedwin.kon...@gmail.com> wrote:
>>> > > > > > > Got the mouse working on
>>> > android-x86
>>> > > donut.
>>> > > > > > >
>>> > > > > > > Reason it did work is
>>> > because android
>>> > > thinks the
>>> > > > screen is
>>> > > > > off.
>>> > > > > > >
>>> > > > > > > Hedwin
>>> > > > > > >
>>> > > > > > > On Sat, Dec 19, 2009 at 9:20
>>> > AM, jian
>>> > > qin
>>> > > > > <qinjiana0...@gmail.com> wrote:
>>> > > > > > >> yeah,I want do this,because
>>> > I havn't
>>> > > any touch
>>> > > > > screen,thanks you guys,I will
>>> > > > > > >> try.
>>> > > > > > >>
>>> > > > > > >> 2009/12/18 hedwin
>>> > > <hedwin.kon...@gmail.com>
>>> > > > > > >>>
>>> > > > > > >>> Working on that, were
>>> > would you
>>> > > expect the
>>> > > > mouse
>>> > > > > handling to take
>>> > > > > > >>> place in android?
>>> > > > > > >>>
>>> > > > > > >>> On Thu, Dec 17, 2009 at
>>> > 8:32 AM, Yi
>>> > > Sun
>>> > > > > <beyo...@gmail.com> wrote:
>>> > > > > > >>> > Yes, that is mips, so do
>>> > check the
>>> > > upstream
>>> > > > otherwise
>>> > > > > it is confusing.
>>> > > > > > >>> >
>>> > > > > > >>> > On Wed, Dec 16, 2009 at
>>> > 11:07 PM,
>>> > > hedwin
>>> > > > > <hedwin.kon...@gmail.com>
>>> > > > > > >>> > wrote:
>>> > > > > > >>> >> USB mouse and keyboard
>>> > do work,
>>> > > only
>>> > > > enabled them in
>>> > > > > the kernel and it
>>> > > > > > >>> >> worked without any
>>> > change to
>>> > > android.
>>> > > > > > >>> >> But then again was
>>> > using the mips
>>> > > android
>>> > > > so not sure
>>> > > > > yet if specific
>>> > > > > > >>> >> changes were made there
>>> > that
>>> > > makes it work.
>>> > > > > > >>> >>
>>> > > > > > >>> >> Hedwin
>>> > > > > > >>> >>
>>> > > > > > >>> >> On Wed, Dec 16, 2009 at
>>> > 7:09 PM,
>>> > > Yi Sun
>>> > > > > <beyo...@gmail.com> wrote:
>>> > > > > > >>> >>> Android does not
>>> > support mouse
>>> > > by default.
>>> > > > But you
>>> > > > > can pick up all the
>>> > > > > > >>> >>> changes from
>>> > android-x86.org for
>>> > > the mouse
>>> > > > support.
>>> > > > > Check the git logs
>>> > > > > > >>> >>> in frameworks/base and
>>> > > > frameworks/policy/base. You
>>> > > > > will be abel to
>>> > > > > > >>> >>> find
>>> > > > > > >>> >>> the changes
>>> > > > > > >>> >>> And, yes, you need to
>>> > enabled
>>> > > your mouse
>>> > > > driver in
>>> > > > > the kernel first.
>>> > > > > > >>> >>> Yi
>>> > > > > > >>> >>> On Wed, 2009-12-16 at
>>> > 14:56
>>> > > +0100, hedwin
>>> > > > wrote:
>>> > > > > > >>> >>>> You need to enable
>>> > this in the
>>> > > linux
>>> > > > kernel.
>>> > > > > > >>> >>>>
>>> > > > > > >>> >>>> Hedwin
>>> > > > > > >>> >>>>
>>> > > > > > >>> >>>> On Wed, Dec 16, 2009
>>> > at 4:02
>>> > > AM, jian qin
>>> > > > > <qinjiana0...@gmail.com>
>>> > > > > > >>> >>>> wrote:
>>> > > > > > >>> >>>> > I want know how to
>>> > use mouse
>>> > > in
>>> > > > Android,please
>>> > > > > help me.
>>> > > > > > >>> >>>> >
>>> > > > > > >>> >>>> > --
>>> > > > > > >>> >>>> > unsubscribe:
>>> > android-porting
>>> > > > > +unsubscr...@googlegroups.com
>>> > > > > > >>> >>>> > website:
>>> > > > >
>>> > > http://groups.google.com/group/android-porting
>>> > > > > > >>> >>>>
>>> > > > > > >>> >>>> --
>>> > > > > > >>> >>>> unsubscribe:
>>> > android-porting
>>> > > > > +unsubscr...@googlegroups.com
>>> > > > > > >>> >>>> website:
>>> > > > >
>>> > > http://groups.google.com/group/android-porting
>>> > > > > > >>> >>>
>>> > > > > > >>> >>>
>>> > > > > > >>> >>> --
>>> > > > > > >>> >>> Android-x86
>>> > > > > > >>> >>> www.android-x86.org
>>> > > > > > >>> >>>
>>> > > > > > >>> >>> --
>>> > > > > > >>> >>> unsubscribe:
>>> > android-porting
>>> > > > > +unsubscr...@googlegroups.com
>>> > > > > > >>> >>> website:
>>> > > > >
>>> > > http://groups.google.com/group/android-porting
>>> > > > > > >>> >>
>>> > > > > > >>> >> --
>>> > > > > > >>> >> unsubscribe:
>>> > android-porting
>>> > > > > +unsubscr...@googlegroups.com
>>> > > > > > >>> >> website:
>>> > > > >
>>> > > http://groups.google.com/group/android-porting
>>> > > > > > >>> >
>>> > > > > > >>> >
>>> > > > > > >>> >
>>> > > > > > >>> > --
>>> > > > > > >>> > Android-x86
>>> > > > > > >>> >
>>> > http://www.android-x86.org
>>> > > > > > >>> >
>>> > > > > > >>> > --
>>> > > > > > >>> > unsubscribe:
>>> > android-porting
>>> > > > > +unsubscr...@googlegroups.com
>>> > > > > > >>> > website:
>>> > > > >
>>> > > http://groups.google.com/group/android-porting
>>> > > > > > >>>
>>> > > > > > >>> --
>>> > > > > > >>> unsubscribe:
>>> > android-porting
>>> > > > > +unsubscr...@googlegroups.com
>>> > > > > > >>> website:
>>> > > >
>>> > http://groups.google.com/group/android-porting
>>> > > > > > >>
>>> > > > > > >> --
>>> > > > > > >> unsubscribe:
>>> > android-porting
>>> > > > +unsubscr...@googlegroups.com
>>> > > > > > >> website:
>>> > > >
>>> > http://groups.google.com/group/android-porting
>>> > > > > > >
>>> > > > > >
>>> > > > >
>>> > > > > > --
>>> > > > >
>>> > > > > > unsubscribe: android-porting
>>> > > > +unsubscr...@googlegroups.com
>>> > > > > > website:
>>> > > >
>>> > http://groups.google.com/group/android-porting
>>> > > > >
>>> > > > >
>>> > > > > --
>>> > > > > Android-x86
>>> > > > > www.android-x86.org
>>> > > > >
>>> > > > > --
>>> > > > > unsubscribe: android-porting
>>> > > > +unsubscr...@googlegroups.com
>>> > > > > website:
>>> > > >
>>> > http://groups.google.com/group/android-porting
>>> > > > >
>>> > > > >
>>> > > > >
>>> > > >
>>> > > > > --
>>> > > >
>>> > > > > unsubscribe: android-porting
>>> > > +unsubscr...@googlegroups.com
>>> > > > > website:
>>> > > http://groups.google.com/group/android-porting
>>> > > >
>>> > > >
>>> > > > --
>>> > > > Android-x86
>>> > > > www.android-x86.org
>>> > > >
>>> > > > --
>>> > > > unsubscribe: android-porting
>>> > > +unsubscr...@googlegroups.com
>>> > > > website:
>>> > > http://groups.google.com/group/android-porting
>>> > > >
>>> > > >
>>> > > >
>>> > >
>>> > > > --
>>> > >
>>> > > > unsubscribe: android-porting
>>> > +unsubscr...@googlegroups.com
>>> > > > website:
>>> > http://groups.google.com/group/android-porting
>>> > >
>>> > >
>>> > > --
>>> > > Android-x86
>>> > > www.android-x86.org
>>> > >
>>> > > --
>>> > > unsubscribe: android-porting
>>> > +unsubscr...@googlegroups.com
>>> > > website:
>>> > http://groups.google.com/group/android-porting
>>> > >
>>> > >
>>> > >
>>> >
>>> > > --
>>> >
>>> > > unsubscribe: android-porting+unsubscr...@googlegroups.com
>>> > > website: http://groups.google.com/group/android-porting
>>> >
>>> >
>>> > --
>>> > Android-x86
>>> > www.android-x86.org
>>> >
>>> > --
>>> > unsubscribe: android-porting+unsubscr...@googlegroups.com
>>> > website: http://groups.google.com/group/android-porting
>>> >
>>> >
>>> >
>>> > --
>>> > unsubscribe: android-porting+unsubscr...@googlegroups.com
>>> > website: http://groups.google.com/group/android-porting
>>>
>>>
>>> --
>>> Android-x86
>>> www.android-x86.org
>>>
>>> --
>>> unsubscribe: android-porting+unsubscr...@googlegroups.com
>>> website: http://groups.google.com/group/android-porting
>
>
> --
> unsubscribe: android-porting+unsubscr...@googlegroups.com
> website: http://groups.google.com/group/android-porting
>
--
Android-x86
http://www.android-x86.org
--
unsubscribe: android-porting+unsubscr...@googlegroups.com
website: http://groups.google.com/group/android-porting
