Active Topics
-
Porting apps to Leste (29)
to Maemo 7 / Leste by Arno_11 - 1 day, 17 hrs ago -
Install Fonts in N900 (68)
to Maemo 5 / Fremantle by teroyk - 2 days, 20 hrs ago - more...
|
2014-05-04
, 14:57
|
|
Posts: 8 |
Thanked: 1 time |
Joined on Jun 2013
|
#402
|
Hello, i'm having some problems with cameraplus. Basically, when the app starts, i have to wait 1 minute until i can take any photos. Any solutions?
 |
|
2014-05-04
, 16:02
|
|
Posts: 605 |
Thanked: 1,778 times |
Joined on Feb 2008
@ Helsinki
|
#403
|
Originally Posted by ignaciospaventa
Why do you have to wait? Does the viewfinder become visible? What exactly happens when you launch the app?
Hello, i'm having some problems with cameraplus. Basically, when the app starts, i have to wait 1 minute until i can take any photos. Any solutions?
__________________
We need a coffee machine that doesn't ask for coffee beans every 10 hours
We need a coffee machine that doesn't ask for coffee beans every 10 hours
|
|
2014-05-04
, 16:08
|
|
Posts: 605 |
Thanked: 1,778 times |
Joined on Feb 2008
@ Helsinki
|
#404
|
Originally Posted by filip.pz
I think it might be some change in the media controller API (might be ioctls renamed or changed struct definitions).
I understand - it's probably just some minor glitch, but w/o some debug info from libomap3camd itself it's kind a hard to figure out what's going on. My best guess is that V4L2 stuff behaves different between two kernels (I already cracked one of those bugs that prevented first usage of libomap3camd under 3.5.3 due to this, and subsequent uses worked - you can imagine debugging multithreaded lib with gdb and w/o debugging info that fails only for first use - I had to reboot about 1037 times). Fcam is probably better option in the long run.
As for the kernel itself, here is the link to it: https://github.com/filippz/kernel-adaptation-n950-n9 and instructions for building: https://wiki.merproject.org/wiki/N9_...tions_from_git. If necessary I can provide my own build (it takes tome time to set everything, ie: MerSDK, SB2...), but it would be hard to play with it w/o the ability to compile it. I've tested it using ubiboot on nemomobile, but it works under sailfish too. For preview I also had to build waylandsink (I can send compiled lib if needed)
There are guys on #nemomobile IRC channel that are better than me in setting the enviroment for compiling, so they can help with kernel building and setting up nemomobile if needed.
Anyway thanks for your app, and you taking the time to take a look at this issue!
omap3camd is a piece of crap. I know
I don't know if yavta http://git.ideasonboard.org/yavta.git would help or not.
I'd personally use filesink and dump the raw frames to a file and inspect them just to be sure that the issue is not a wayland issue.
I will try to have a look and see. I personally prefer using nemo than Sailfish but I will see what's flashed on my spare N9
P.S. also note that the nemo version of omap3camd is crippled. The harmattan version is better.
__________________
We need a coffee machine that doesn't ask for coffee beans every 10 hours
We need a coffee machine that doesn't ask for coffee beans every 10 hours
 |
|
2014-05-04
, 17:18
|
|
Posts: 108 |
Thanked: 579 times |
Joined on Feb 2013
@ Požega, Croatia
|
#405
|
Originally Posted by MSameer
I think that the change to ioctls/structs would cause exceptions in code, rather than misbehaving, but anything is possible.
I think it might be some change in the media controller API (might be ioctls renamed or changed struct definitions).
omap3camd is a piece of crap. I know
I don't know if yavta http://git.ideasonboard.org/yavta.git would help or not.
I'd personally use filesink and dump the raw frames to a file and inspect them just to be sure that the issue is not a wayland issue.
I will try to have a look and see. I personally prefer using nemo than Sailfish but I will see what's flashed on my spare N9
P.S. also note that the nemo version of omap3camd is crippled. The harmattan version is better.
I used default nemo kernel to test waylandsink and it works OK. To be safe I tried filesink, and app.c from https://gitorious.org/msameer/omap3camd/source/master: Raw image data contains occasional ones - not just entire zeroes - so maybe liboamp3camd has trouble with "brightness" setting (wild guess). I also tried setting smiapp driver to test pattern to exclude problems with smiapp (without using libomap3camd I get test pattern to display - but with wrong colors). Since I had to reconfigure kernel memory to be used for zero-copy buffering - I'm not sure that the issue isn't there.
"First use" bug was due to the fact that V4L2 in older kernels would make ctrl->val same as ctrl->cur.val when setting v4l2_ctrl struct for smiapp driver (fix is here: https://github.com/filippz/kernel-ad...a78a8413378f69) - on the second run ctrl->val would be OK. Since there are ctrl->val, ctr->val64, ctrl->cur.val and ctrl->cur.val64 for every v4l2_ctrl it's possible that libomap3camd looks at (for example) ctrl->cur.val, but in newer kernels it should look at cur->val64. This doesn't cause it to break, but it reads default value and produces wrong result.
I also prefer nemo on my spare N9 - the guys there are doing a fantastic job. I'm not sure what you mean when you say that nemo version is crippled - I was under impression that the versions are the same?
|
|
2014-05-04
, 17:28
|
|
Posts: 8 |
Thanked: 1 time |
Joined on Jun 2013
|
#406
|
Originally Posted by MSameer
The viewfinder becomes visible, but only for 1 second, then it freezes and it shows "cameraplus not responding". If i wait a minute it starts working. It's not the hardware, because if I use the native camera it works perfectly well
Why do you have to wait? Does the viewfinder become visible? What exactly happens when you launch the app?
|
|
2014-05-04
, 17:49
|
|
Posts: 605 |
Thanked: 1,778 times |
Joined on Feb 2008
@ Helsinki
|
#407
|
Originally Posted by ignaciospaventa
Could you launch it from the terminal and paste anything printed there?
The viewfinder becomes visible, but only for 1 second, then it freezes and it shows "cameraplus not responding". If i wait a minute it starts working. It's not the hardware, because if I use the native camera it works perfectly well
This needs debugging with gdb apparently but let's see.
__________________
We need a coffee machine that doesn't ask for coffee beans every 10 hours
We need a coffee machine that doesn't ask for coffee beans every 10 hours
|
|
2014-05-04
, 17:55
|
|
Posts: 8 |
Thanked: 1 time |
Joined on Jun 2013
|
#408
|
Originally Posted by MSameer
Could you launch it from the terminal and paste anything printed there?
This needs debugging with gdb apparently but let's see.
Code:
~ $ cameraplus Found SGX/MBX driver, enabling FullClearOnEveryFrame Found v1.4 driver, enabling brokenTexSubImage Loading plugin: true Found SGX/MBX driver, enabling FullClearOnEveryFrame Found v1.4 driver, enabling brokenTexSubImage libomap3camd 1.136 CC reset, prev_cc: 0 new_cc -2259980 new offset: 2259980 DBus error: org.freedesktop.DBus.Error.NoReply Did not receive a reply. Possible causes include: the remote application did not send a reply, the message bus security policy blocked the reply, the reply timeout expired, or the networkconnection was broken. Error occured while reading data from socket:"Unknown error" libomap3camd 1.136 Warning Internal data flow problem. : gstbasesink.c(3678): gst_base_sink_chain_unlocked (): /GstCameraBin2:QtCameraCameraBin/GstViewfinderBin:vf-bin/GstGLTextureSink:QtCamViewfinderRendererMeeGoSink: Received buffer without a new-segment. Assuming timestamps start from 0. libomap3camd 1.136 Warning Internal data flow problem. : gstbasesink.c(3678): gst_base_sink_chain_unlocked (): /GstCameraBin2:QtCameraCameraBin/GstViewfinderBin:vf-bin/GstGLTextureSink:QtCamViewfinderRendererMeeGoSink: Received buffer without a new-segment. Assuming timestamps start from 0. CC reset, prev_cc: 0 new_cc -2266580 new offset: 2266580 Too many samples waiting in socket. Flushing it to empty
|
|
2014-05-04
, 17:56
|
|
Posts: 605 |
Thanked: 1,778 times |
Joined on Feb 2008
@ Helsinki
|
#409
|
Originally Posted by filip.pz
changes in ioctls can exhibit different behavior depending on the ioctl itself and its usage, Anyway I am just guessing as I have not yet checked.
I think that the change to ioctls/structs would cause exceptions in code, rather than misbehaving, but anything is possible.
I used default nemo kernel to test waylandsink and it works OK. To be safe I tried filesink, and app.c from https://gitorious.org/msameer/omap3camd/source/master: Raw image data contains occasional ones - not just entire zeroes - so maybe liboamp3camd has trouble with "brightness" setting (wild guess). I also tried setting smiapp driver to test pattern to exclude problems with smiapp (without using libomap3camd I get test pattern to display - but with wrong colors). Since I had to reconfigure kernel memory to be used for zero-copy buffering - I'm not sure that the issue isn't there.
"First use" bug was due to the fact that V4L2 in older kernels would make ctrl->val same as ctrl->cur.val when setting v4l2_ctrl struct for smiapp driver (fix is here: https://github.com/filippz/kernel-ad...a78a8413378f69) - on the second run ctrl->val would be OK. Since there are ctrl->val, ctr->val64, ctrl->cur.val and ctrl->cur.val64 for every v4l2_ctrl it's possible that libomap3camd looks at (for example) ctrl->cur.val, but in newer kernels it should look at cur->val64. This doesn't cause it to break, but it reads default value and produces wrong result.
I also prefer nemo on my spare N9 - the guys there are doing a fantastic job. I'm not sure what you mean when you say that nemo version is crippled - I was under impression that the versions are the same?
I remember I came across an old debug binary for omap3camd while trying to reverse engineer its header. I can't remember where I got it from
Have you tried using the N9 driver instead of the smiapp one?
I am trying to ask around about the driver state but no replies so far.
As for omap3camd. My understanding is that the public nemo version was crippled by Nokia. I am not entirely sure but that's another topic
__________________
We need a coffee machine that doesn't ask for coffee beans every 10 hours
We need a coffee machine that doesn't ask for coffee beans every 10 hours
|
|
2014-05-04
, 18:19
|
|
Posts: 605 |
Thanked: 1,778 times |
Joined on Feb 2008
@ Helsinki
|
#410
|
Code:
DBus error: org.freedesktop.DBus.Error.NoReply Did not receive a reply. Possible causes include: the remote application did not send a reply, the message bus security policy blocked the reply, the reply timeout expired, or the networkconnection was broken.
Can you try after disabling GPS and geotags?
__________________
We need a coffee machine that doesn't ask for coffee beans every 10 hours
We need a coffee machine that doesn't ask for coffee beans every 10 hours
 |
| Tags |
| ?where is nemo?, thank you! |
«
Previous Thread
|
Next Thread
»
|
All times are GMT. The time now is 13:18.
As for the kernel itself, here is the link to it: https://github.com/filippz/kernel-adaptation-n950-n9 and instructions for building: https://wiki.merproject.org/wiki/N9_...tions_from_git. If necessary I can provide my own build (it takes tome time to set everything, ie: MerSDK, SB2...), but it would be hard to play with it w/o the ability to compile it. I've tested it using ubiboot on nemomobile, but it works under sailfish too. For preview I also had to build waylandsink (I can send compiled lib if needed)
There are guys on #nemomobile IRC channel that are better than me in setting the enviroment for compiling, so they can help with kernel building and setting up nemomobile if needed.
Anyway thanks for your app, and you taking the time to take a look at this issue!