new version of zmplayer: v1_1 optimized mplayer and frontend

[bragon]

Please find attached to this post a new version of the zmplayer package that
should work now on all the latest firmwares for A1200 and A780/E680.

The Changelog includes the following:

* Update the menu.conf file to correct the key mappings. Should all work.
* Run-time selection of E680/A780 key bindings based on/etc/modelnum.txt or /etc/ap_version.txt. This basically assumes that on a E680 the first 4 letters in /etc/ap_version.txt are "E680", and so might fail. This probing of the phone type can be overridden by setting the environment variable ZMPLAYER_PLATFORM to A780, E680, or anything else for the default bindings.
* Added code to menu to probe whether the menu was on the screen. This removes certain bugs with the menu code
* Re-enable the power management on the A780 when flip-closed to allow battery save when listening to audio playlists (feature already in A1200 version). For A780 due to missing functionality this was done by trapping the FocusIn and FocusOut events and switching power management on them. In general use these events should only happen on a flip, and so this should be ok, though is a bit ugly.
* Changes to allow playlists and in particular audio playlists to work correctly.
- Added a "-bgvideo" option from geexbox to allow a background video for audio only tracks
- Save and restore volume across elements of the playlist
- Fix framedrop issue in geexbox bgvideo patch
- Don't use high-res mp3 seeking as it doesn't work across elements on a playlist
* Do the environment initialization in the zmplayer binary itself, removing the need for basename and sed. This also addresses the SE Linux security of A1200 China Mobile firmware's as the QT application can now be directly called from the application manager.

The original post for zmplayer is there: http://www.motorolafans.com/index.ph...ewtopic&t=4941
Please do read this post for detailed installation instructions.

Enjoy!

Authors: dbateman and bragon

[tarwin]

Hi, I didn't know whether to reply here or in the e680i forum so I chose here.

One bug I have noticed is that on the e680i there is no sound. To get sound you have to plug in the headset (and then unplug them if you want the sound to come through the speakers). Also once sound has been activated if you advance or rewind (a little bit or to the end/beginning, it doesn't matter) of a file it is disabled again. With eplayer this was resolved by just pressing the mute button. Unfortunately activating and disabling mute on zmplayer doesn't work. Nor does increasing or decreasing the volume.

Also what does the "speed" option do? If I try to do anything with it the video freezes (I have only tried zmplayer with mp4's so I know not how it works with other files).

The player is otherwise great, it gives you lots of options and control. I just have to test it against files that give eplayer difficulty to see how it compares.

Out of curiousity is zmplayer overclocked? Because if it isn't that would affect the results of comparison testing (after all, if it's not overclocked and it is "just" equal to eplayer 1.5 then it would be superior as eplayer 1.5 is overclocked).

And out of curiousity, does it use bluetooth stereo sound? I can't test that due to the audio glitch I mentioned earlier... (it would be great if it did but I understand if it doesn't).

Thanks =)

P.S. What are you doing your testing on? an e680 or an e680i. You keep on saying e680 and if it's not an "i" then it would be nice to know as the two are not entirely compatible and sometimes require remapping of the keys (and if what I understand is correct the program can't differentiate between an e680 and an e680i as it only looks at the first 4 letters).

Keep up the great work, we totally appreciate it

[dbateman]

Hi, I didn't know whether to reply here or in the e680i forum so I chose here.

One bug I have noticed is that on the e680i there is no sound. To get sound you have to plug in the headset (and then unplug them if you want the sound to come through the speakers). Also once sound has been activated if you advance or rewind (a little bit or to the end/beginning, it doesn't matter) of a file it is disabled again. With eplayer this was resolved by just pressing the mute button. Unfortunately activating and disabling mute on zmplayer doesn't work. Nor does increasing or decreasing the volume.

Yes I saw the code to fix this in cmplayer, but considered this crufty as essentially to does a ioctl on the mixer on most operations to mplayer. Also we don't seem to need this code for the A1200, and so I'd hoped not to have to port it.

I also have another issue in that we can not make the QT frontend open source yet, and so strictly speaking I can't copy the code from cmplayer. The concept is fairly simple, so I'd have no problems duplicating it, but as I can't copy it will slow me down. Ok, I've added the need for this audio fix to my todo list.

Also what does the "speed" option do? If I try to do anything with it the video freezes (I have only tried zmplayer with mp4's so I know not how it works with other files).

"speed" allow rapid advance through the video will still displaying the video. As it essentially means all frames have to be decoded before displaying them after about a fact of x2 it gets difficult for the poor pxa270 to keep up.

The player is otherwise great, it gives you lots of options and control. I just have to test it against files that give eplayer difficulty to see how it compares.

Out of curiousity is zmplayer overclocked? Because if it isn't that would affect the results of comparison testing (after all, if it's not overclocked and it is "just" equal to eplayer 1.5 then it would be superior as eplayer 1.5 is overclocked).

zmplayer is not overclocked, and i'd prefer not to do this. What I'd prefer to do and have already first proof of concepts is to use the yuv colour space conversion of the pxa270 to allow sufficient acceleration to not need overclocking. This should allow even large divx files to be displayed. The basic code is in the yarus bvdd driver, however this can not be used as the yarus doesn't use the SRAM of the pxa270 at all, whereas the EZX uses the SRAM for the main framebuffer. Also the yarus doean't use any of the pxa270 overlays, whereas the EZX seems to use OVERLAY1 on the A1200 to allow RGB16 to RGB19 conversion. Therefore the complexity of such an effort is to prevent the pxafb of the kernel to not tread on the feet of an EZX driver. It would probably be easier to write a new pxafb driver, but that means essentially a custom kernel .....Yuk...

And out of curiousity, does it use bluetooth stereo sound? I can't test that due to the audio glitch I mentioned earlier... (it would be great if it did but I understand if it doesn't).

Don't know, I don't have a bluetooth stereo headset.

Thanks =)

P.S. What are you doing your testing on? an e680 or an e680i. You keep on saying e680 and if it's not an "i" then it would be nice to know as the two are not entirely compatible and sometimes require remapping of the keys (and if what I understand is correct the program can't differentiate between an e680 and an e680i as it only looks at the first 4 letters).

Keep up the great work, we totally appreciate it

We tested breifly on a E680i, essential to show it worked and grab the keymapping. We can't claim extensive testing. The keymapping is therefore an E680i. The program is not setup to tell the difference between an E680i and an E680. If you want to send me the keymapping (ie the keycodes QT sees) for an E680 and the means to identify an E680 form an E680i (eg string in a file in /etc as I currently use) then I'd be happy to add this feature.


D.

[tarwin]

thanks for your response

and I only have an e680i, and it worked fine for me. I was just for clarification whether you'd tested on e680 or e680i since in your posts mentioned e680, not i.

But thanks

And though I can't wait for the "fixes" I shall wait patiently and eagerly for your next release. (and though I would love to see bluetooth stereo support I won't hold my breath of pester you for it)

And I thinkit's great you're able to get such great quality without overclocking

[dbateman]

thanks for your response

and I only have an e680i, and it worked fine for me. I was just for clarification whether you'd tested on e680 or e680i since in your posts mentioned e680, not i.

But thanks

And though I can't wait for the "fixes" I shall wait patiently and eagerly for your next release. (and though I would love to see bluetooth stereo support I won't hold my breath of pester you for it)

And I thinkit's great you're able to get such great quality without overclocking

Tarwin, as eplayer, cmplayer and zmplayer all use an mplayer backend there is nothing to stop you copying our mplayer binary and replacing the one in eplayer or cmplayer. There are two features that give a speed boost to our mplayer binaries for the A780/E680 (though is the same mplayer binary for all phones we support and only the zmplayer frontend changes). These are that we added code for YUV422 to RGB16 colour space conversion using the Intel IPP libraries. This gives about a 25% speed-up as far as I can tell. We also added a 8x8 block rotation code that is optimal to avoid cache misses on the arm chips, so that if you are using rotation of a 320x240 video to fit the 240x320 screen then this is faster. I observed that the rotation has twice as faster for an improvement of another 10% or so in overall speed. These numbers are with a certain test video I had, so of course your luck may vary, but I estimate our mplayer binary as about 30% faster at the same clock speed than the ones in cmplayer and eplayer. Unfortunately only the rotation is useful for the A1200 due to the ARGB19 mode it uses on the screen. The intel engineer who had the great idea for that this 19-bit mode were the bits are packed should be taken out and shot. The cost this implies in manipulation of the bits prior to writing them to the screen is horrid.

D.

[tarwin]

I know I can just replace the mplayer file, but your front end is also quite good, with the ability to control brightness and contrast as well as other things. Options other players don't have. And though CMplayer can do these things I for some reason cannot get it to run mp4 files, something your player does without any problems.

And the speed boost is great, now there's no need to overclock at my current video settings. Now to test the limit with higher bitrates

[neutrino]

I test it on E680 (not i) and it seems that there are some error?
Not working...
I try to see the content of ap_version.txt, and there is firmware code (You can get the same by using "WWW008W" - It's something like R51_....
There is no identification string "E680" or "E680i" ?
From the other hand ./mplayer binary does work, because I test it manually using qonsole application and trying to play something.
Thank You

[Marecki]

Hi Bragon, I've heard that a new version of MPlayer has been released. Are you going to recompile it for EZX?

[blackhawk]

@bragon
Your mplayer ogg format support is broken. I can't play all my .ogg file collections. Look like use regular ogg/vorbis library, not fixed-point version (tremor). Can you fix it in the next version? Thank you.

[blackhawk]

@bragon
Have found mplayer's configure file have tremor support option. I hope easy for you to fix .ogg support problem. Linux media player incomplete without .ogg support Can you make an update? Thank you.