pulseaudio makes system unstable

pulseaudio makes system unstable

[Alexander Shashkevych]

Hello,

Maybe this is known issue, but I have not found any answers by
googling. I have two boards and two systems. One board is wandboard
quad with imx6 and second is custom board based on wandboard quad.
Also I have Ubuntu 13.10 (baked with my own recipes) and I have Yocto
1.6.2.

On both systems and both devices (in any combination) all graphic
stuff is working fine: gles2.0 demos, gstreamer with imx plugins.
Everything is running with h/w acceleration. But once I start to play
audio with pulseaudio (for example using gstreamer) system becomes
pretty unstable: gstreamer could crash with 'illegal instruction'
message or with 'pulseaudio socket write error'. Sometimes even
desktop crashes and throws me out to login screen. This happens on all
devices with all systems.

I need pulseaudio to play and capture desktop sounds, but it looks
like pulseaudio even can't play without problems. At the moment I'm
using alsa device for playing and everything is OK, but capture of
desktop sounds from alsa device, afaik, could be tricky. I'm not sure
what I should do: fix pulseaudio or implement workaround with alsa.
Not sure which one will require more efforts.

Is this known issue with pulseaudio?

Regards,
Alexander

Re: pulseaudio makes system unstable

[Marco Trillo]

Hi,

On 11/24/2014 01:12 PM, Alexander Shashkevych wrote:
> Hello,
>
> Maybe this is known issue, but I have not found any answers by
> googling. I have two boards and two systems. One board is wandboard
> quad with imx6 and second is custom board based on wandboard quad.
> Also I have Ubuntu 13.10 (baked with my own recipes) and I have Yocto
> 1.6.2.
>
> On both systems and both devices (in any combination) all graphic
> stuff is working fine: gles2.0 demos, gstreamer with imx plugins.
> Everything is running with h/w acceleration. But once I start to play
> audio with pulseaudio (for example using gstreamer) system becomes
> pretty unstable: gstreamer could crash with 'illegal instruction'

Which audio device are you using? The HDMI port (imx-hdmi-soc-audio) or 
the analog port (sgtl5000) on the Wandboard? If you are indeed using the 
HDMI audio, could you try using the analog audio and see if the problem 
persists?

I reproduced the 'Illegal instruction' crashes you mention when using 
HDMI audio -- not only with `pulsesink' but also with `alsasink' 
depending on other unrelated particularities of the test-- but the 
problem disappeared as soon as I switched to the sgtl5000 audio.

Kind regards,
Marco.

Status of imx28evk recipes/images

[Simone]

Dear All,
In my company we are starting a new project (a low-cost industrial controller) and we are thinking about use the iMX286 CPU because it offer all what we need. Of course we would start from imx28evk for testing, then we will develop our custom board based on that one.
But before all I want to know (from who is already using/mantaining it), how is the status of support (recipes, images, kernel) for such board in meta-fsl-arm.

If possible, I want to know if there are known problems or missing things, or if this cpu is not suggested for a new project.. 
Any opinion you want to share could be very useful to me.

Thank you all for every suggestion or information you will provide

Regards
Simone

Re: Status of imx28evk recipes/images

[Daiane Angolini]

On Mon, Nov 24, 2014 at 6:03 PM, Simone <cjb.sw.nospam@gmail.com> wrote:
> Dear All,
> In my company we are starting a new project (a low-cost industrial controller) and we are thinking about use the iMX286 CPU because it offer all what we need. Of course we would start from imx28evk for testing, then we will develop our custom board based on that one.
> But before all I want to know (from who is already using/mantaining it), how is the status of support (recipes, images, kernel) for such board in meta-fsl-arm.
>
> If possible, I want to know if there are known problems or missing things, or if this cpu is not suggested for a new project..
> Any opinion you want to share could be very useful to me.

imx28evk is a beautiful board. I´ve been testing it here and there and
I cannot remember serious problems. Of course, there are some:

* touch screen (with x11 at least) is not completely calibrated
* you can choose between linux-fslc or linux-imx
* you can choose between gstreamer mainline or the one provided by
freescale (see RN [1] in order to understand the HW/SW correlation)
* there is a pin conflict in imx28evk board, that I don´t remember
from top of my mind (I think it´s NAND + Ethernet cannot be used
together)
* NAND is already supported in meta-fsl-arm
* mfgtools is not already supported in meta-fsl-arm, but it´s easy to
be included and your patch is appreciated.
* pendrive is not automatically mounted, but once you mount it
everything works fine (I have never stopped to fix that)


that´s all I could remember

[1]http://freescale.github.io/doc/release-notes/1.7/index.html#document-fsl-userspace-pkg


Otavio, could some of this items be considered "known issue" to be
included in release notes? What do you think?

Daiane
>
> Thank you all for every suggestion or information you will provide
>
> Regards
> Simone
> --
> _______________________________________________
> meta-freescale mailing list
> meta-freescale@yoctoproject.org
> https://lists.yoctoproject.org/listinfo/meta-freescale

Re: Status of imx28evk recipes/images

[Otavio Salvador]

On Tue, Nov 25, 2014 at 8:58 AM, Daiane Angolini <daiane.list@gmail.com> wrote:
> On Mon, Nov 24, 2014 at 6:03 PM, Simone <cjb.sw.nospam@gmail.com> wrote:
>> Dear All,
>> In my company we are starting a new project (a low-cost industrial controller) and we are thinking about use the iMX286 CPU because it offer all what we need. Of course we would start from imx28evk for testing, then we will develop our custom board based on that one.
>> But before all I want to know (from who is already using/mantaining it), how is the status of support (recipes, images, kernel) for such board in meta-fsl-arm.
>>
>> If possible, I want to know if there are known problems or missing things, or if this cpu is not suggested for a new project..
>> Any opinion you want to share could be very useful to me.
>
> imx28evk is a beautiful board. I´ve been testing it here and there and
> I cannot remember serious problems. Of course, there are some:
>
> * touch screen (with x11 at least) is not completely calibrated
> * you can choose between linux-fslc or linux-imx
> * you can choose between gstreamer mainline or the one provided by
> freescale (see RN [1] in order to understand the HW/SW correlation)
> * there is a pin conflict in imx28evk board, that I don´t remember
> from top of my mind (I think it´s NAND + Ethernet cannot be used
> together)
> * NAND is already supported in meta-fsl-arm
> * mfgtools is not already supported in meta-fsl-arm, but it´s easy to
> be included and your patch is appreciated.
> * pendrive is not automatically mounted, but once you mount it
> everything works fine (I have never stopped to fix that)
>
>
> that´s all I could remember
>
> [1]http://freescale.github.io/doc/release-notes/1.7/index.html#document-fsl-userspace-pkg
>
>
> Otavio, could some of this items be considered "known issue" to be
> included in release notes? What do you think?

Yes for sure :) Maybe board specific issues?

-- 
Otavio Salvador                             O.S. Systems
http://www.ossystems.com.br        http://code.ossystems.com.br
Mobile: +55 (53) 9981-7854            Mobile: +1 (347) 903-9750

Re: Status of imx28evk recipes/images

[Simone]

Hello Daiane,
Thanks so much for your answer.

Which Kernel version is actually supported by this board?
What is the difference between linux-fslc and linux-imx ?

Thanks so much
Simone



-----Messaggio originale-----
Da: angolini@gmail.com [mailto:angolini@gmail.com] Per conto di Daiane Angolini
Inviato: martedì 25 novembre 2014 11:58
A: Simone; Otavio Salvador
Cc: meta-freescale@yoctoproject.org
Oggetto: Re: [meta-freescale] Status of imx28evk recipes/images

On Mon, Nov 24, 2014 at 6:03 PM, Simone <cjb.sw.nospam@gmail.com> wrote:
> Dear All,
> In my company we are starting a new project (a low-cost industrial controller) and we are thinking about use the iMX286 CPU because it offer all what we need. Of course we would start from imx28evk for testing, then we will develop our custom board based on that one.
> But before all I want to know (from who is already using/mantaining it), how is the status of support (recipes, images, kernel) for such board in meta-fsl-arm.
>
> If possible, I want to know if there are known problems or missing things, or if this cpu is not suggested for a new project..
> Any opinion you want to share could be very useful to me.

imx28evk is a beautiful board. I´ve been testing it here and there and I cannot remember serious problems. Of course, there are some:

* touch screen (with x11 at least) is not completely calibrated
* you can choose between linux-fslc or linux-imx
* you can choose between gstreamer mainline or the one provided by freescale (see RN [1] in order to understand the HW/SW correlation)
* there is a pin conflict in imx28evk board, that I don´t remember from top of my mind (I think it´s NAND + Ethernet cannot be used
together)
* NAND is already supported in meta-fsl-arm
* mfgtools is not already supported in meta-fsl-arm, but it´s easy to be included and your patch is appreciated.
* pendrive is not automatically mounted, but once you mount it everything works fine (I have never stopped to fix that)


that´s all I could remember

[1]http://freescale.github.io/doc/release-notes/1.7/index.html#document-fsl-userspace-pkg


Otavio, could some of this items be considered "known issue" to be included in release notes? What do you think?

Daiane
>
> Thank you all for every suggestion or information you will provide
>
> Regards
> Simone
> --
> _______________________________________________
> meta-freescale mailing list
> meta-freescale@yoctoproject.org
> https://lists.yoctoproject.org/listinfo/meta-freescale

Re: Status of imx28evk recipes/images

[Daiane Angolini]

On Tue, Nov 25, 2014 at 11:24 AM, Simone <cjb.sw.nospam@gmail.com> wrote:
> Hello Daiane,
> Thanks so much for your answer.
>
> Which Kernel version is actually supported by this board?

define supported.

linux-imx is supported by freescale
linux-fslc is from kernel.org

> What is the difference between linux-fslc and linux-imx ?

huge

Please, test and taste both, then decide which one is the one for your
application.

(this question is really tricky, by definition, the difference between
2 kernels is defined by the git diff of them, however, the kernel
version in this case differs, causing a huge (huge) gap. I cannot say
to you what are the differences that probably can impact in your case,
so I prefer to say nothing)


Daiane
>
> Thanks so much
> Simone
>
>
>
> -----Messaggio originale-----
> Da: angolini@gmail.com [mailto:angolini@gmail.com] Per conto di Daiane Angolini
> Inviato: martedì 25 novembre 2014 11:58
> A: Simone; Otavio Salvador
> Cc: meta-freescale@yoctoproject.org
> Oggetto: Re: [meta-freescale] Status of imx28evk recipes/images
>
> On Mon, Nov 24, 2014 at 6:03 PM, Simone <cjb.sw.nospam@gmail.com> wrote:
>> Dear All,
>> In my company we are starting a new project (a low-cost industrial controller) and we are thinking about use the iMX286 CPU because it offer all what we need. Of course we would start from imx28evk for testing, then we will develop our custom board based on that one.
>> But before all I want to know (from who is already using/mantaining it), how is the status of support (recipes, images, kernel) for such board in meta-fsl-arm.
>>
>> If possible, I want to know if there are known problems or missing things, or if this cpu is not suggested for a new project..
>> Any opinion you want to share could be very useful to me.
>
> imx28evk is a beautiful board. I´ve been testing it here and there and I cannot remember serious problems. Of course, there are some:
>
> * touch screen (with x11 at least) is not completely calibrated
> * you can choose between linux-fslc or linux-imx
> * you can choose between gstreamer mainline or the one provided by freescale (see RN [1] in order to understand the HW/SW correlation)
> * there is a pin conflict in imx28evk board, that I don´t remember from top of my mind (I think it´s NAND + Ethernet cannot be used
> together)
> * NAND is already supported in meta-fsl-arm
> * mfgtools is not already supported in meta-fsl-arm, but it´s easy to be included and your patch is appreciated.
> * pendrive is not automatically mounted, but once you mount it everything works fine (I have never stopped to fix that)
>
>
> that´s all I could remember
>
> [1]http://freescale.github.io/doc/release-notes/1.7/index.html#document-fsl-userspace-pkg
>
>
> Otavio, could some of this items be considered "known issue" to be included in release notes? What do you think?
>
> Daiane
>>
>> Thank you all for every suggestion or information you will provide
>>
>> Regards
>> Simone
>> --
>> _______________________________________________
>> meta-freescale mailing list
>> meta-freescale@yoctoproject.org
>> https://lists.yoctoproject.org/listinfo/meta-freescale
>

Re: pulseaudio makes system unstable

[Alexander Shashkevych]

>> Maybe this is known issue, but I have not found any answers by
>> googling. I have two boards and two systems. One board is wandboard
>> quad with imx6 and second is custom board based on wandboard quad.
>> Also I have Ubuntu 13.10 (baked with my own recipes) and I have Yocto
>> 1.6.2.
>>
>> On both systems and both devices (in any combination) all graphic
>> stuff is working fine: gles2.0 demos, gstreamer with imx plugins.
>> Everything is running with h/w acceleration. But once I start to play
>> audio with pulseaudio (for example using gstreamer) system becomes
>> pretty unstable: gstreamer could crash with 'illegal instruction'
>
>
> Which audio device are you using? The HDMI port (imx-hdmi-soc-audio) or the
> analog port (sgtl5000) on the Wandboard? If you are indeed using the HDMI
> audio, could you try using the analog audio and see if the problem persists?
>
> I reproduced the 'Illegal instruction' crashes you mention when using HDMI
> audio -- not only with `pulsesink' but also with `alsasink' depending on
> other unrelated particularities of the test-- but the problem disappeared as
> soon as I switched to the sgtl5000 audio.

I've tried to use different pulse sinks and I see that it behaves a
bit differently, but still unstable. Sometimes pavucontrol closes
unexpectedly, sometimes desktop closes, sometimes pulse server stops
and once pulse server stops, gstreamer can't connect again to newly
started instance of server and so on. Very strange and unpredictable.

But once I'm starting to use alsa device from gstreamer - everything
plays. Looks like pulseaudio is incompatible with hardware. Not sure
where to move further.

Regards,
Alexander

Re: pulseaudio makes system unstable

[Daiane Angolini]

On Tue, Nov 25, 2014 at 2:05 PM, Alexander Shashkevych <alex@stunpix.com> wrote:
>>> Maybe this is known issue, but I have not found any answers by
>>> googling. I have two boards and two systems. One board is wandboard
>>> quad with imx6 and second is custom board based on wandboard quad.
>>> Also I have Ubuntu 13.10 (baked with my own recipes) and I have Yocto
>>> 1.6.2.
>>>
>>> On both systems and both devices (in any combination) all graphic
>>> stuff is working fine: gles2.0 demos, gstreamer with imx plugins.
>>> Everything is running with h/w acceleration. But once I start to play
>>> audio with pulseaudio (for example using gstreamer) system becomes
>>> pretty unstable: gstreamer could crash with 'illegal instruction'
>>
>>
>> Which audio device are you using? The HDMI port (imx-hdmi-soc-audio) or the
>> analog port (sgtl5000) on the Wandboard? If you are indeed using the HDMI
>> audio, could you try using the analog audio and see if the problem persists?
>>
>> I reproduced the 'Illegal instruction' crashes you mention when using HDMI
>> audio -- not only with `pulsesink' but also with `alsasink' depending on
>> other unrelated particularities of the test-- but the problem disappeared as
>> soon as I switched to the sgtl5000 audio.
>
> I've tried to use different pulse sinks and I see that it behaves a
> bit differently, but still unstable. Sometimes pavucontrol closes
> unexpectedly, sometimes desktop closes, sometimes pulse server stops
> and once pulse server stops, gstreamer can't connect again to newly
> started instance of server and so on. Very strange and unpredictable.
>
> But once I'm starting to use alsa device from gstreamer - everything
> plays. Looks like pulseaudio is incompatible with hardware. Not sure
> where to move further.


May [1] be related with your problems?

[1] http://git.freescale.com/git/cgit.cgi/imx/linux-2.6-imx.git/tree/sound/soc/fsl/imx-hdmi-dma.c?h=imx_3.10.31_1.1.0_beta2#n289


Daiane
>
> Regards,
> Alexander
> --
> _______________________________________________
> meta-freescale mailing list
> meta-freescale@yoctoproject.org
> https://lists.yoctoproject.org/listinfo/meta-freescale

Re: pulseaudio makes system unstable

[Marco Trillo]

Hi,

On 11/25/2014 05:05 PM, Alexander Shashkevych wrote:
>> Which audio device are you using? The HDMI port (imx-hdmi-soc-audio) or the
>> analog port (sgtl5000) on the Wandboard? If you are indeed using the HDMI
>> audio, could you try using the analog audio and see if the problem persists?
>>
>> I reproduced the 'Illegal instruction' crashes you mention when using HDMI
>> audio -- not only with `pulsesink' but also with `alsasink' depending on
>> other unrelated particularities of the test-- but the problem disappeared as
>> soon as I switched to the sgtl5000 audio.
> I've tried to use different pulse sinks and I see that it behaves a
> bit differently, but still unstable. Sometimes pavucontrol closes
> unexpectedly, sometimes desktop closes, sometimes pulse server stops
> and once pulse server stops, gstreamer can't connect again to newly
> started instance of server and so on. Very strange and unpredictable.

Yeah, I understand. I also remember these issues with `pulseaudio' when 
I did my tests. ISTR that starting the daemon with `--exit-idle-time=-1' 
and `--allow-exit=false' avoided the daemon to stop as you mention.

However I think these issues are not related to the `Illegal 
instruction' crashes. Did these crashes still occur when using the 
analog device as a sink?  For me, these crashes were linked to the HDMI 
audio output device.

Kind regards,
Marco

Re: pulseaudio makes system unstable

[Alexander Shashkevych]

>>> I reproduced the 'Illegal instruction' crashes you mention when using HDMI
>>> audio -- not only with `pulsesink' but also with `alsasink' depending on
>>> other unrelated particularities of the test-- but the problem disappeared as
>>> soon as I switched to the sgtl5000 audio.
>>
>> I've tried to use different pulse sinks and I see that it behaves a
>> bit differently, but still unstable. Sometimes pavucontrol closes
>> unexpectedly, sometimes desktop closes, sometimes pulse server stops
>> and once pulse server stops, gstreamer can't connect again to newly
>> started instance of server and so on. Very strange and unpredictable.
>>
>> But once I'm starting to use alsa device from gstreamer - everything
>> plays. Looks like pulseaudio is incompatible with hardware. Not sure
>> where to move further.
>
>
> May [1] be related with your problems?
>
> [1] http://git.freescale.com/git/cgit.cgi/imx/linux-2.6-imx.git/tree/sound/soc/fsl/imx-hdmi-dma.c?h=imx_3.10.31_1.1.0_beta2#n289

Interesting. I've found same #ifdefs in yocto branch I'm using with
kernel 3.10.17, but there is no #define that disables  neon's dma for
hdmi, so I'll define it and recompile.

Thank you!

-- 
Regards,
Alexander

Re: pulseaudio makes system unstable

[Alexander Shashkevych]

>> I've tried to use different pulse sinks and I see that it behaves a
>> bit differently, but still unstable. Sometimes pavucontrol closes
>> unexpectedly, sometimes desktop closes, sometimes pulse server stops
>> and once pulse server stops, gstreamer can't connect again to newly
>> started instance of server and so on. Very strange and unpredictable.
>
>
> Yeah, I understand. I also remember these issues with `pulseaudio' when I
> did my tests. ISTR that starting the daemon with `--exit-idle-time=-1' and
> `--allow-exit=false' avoided the daemon to stop as you mention.
>
> However I think these issues are not related to the `Illegal instruction'
> crashes. Did these crashes still occur when using the analog device as a
> sink?  For me, these crashes were linked to the HDMI audio output device.

Since yesterday I haven't observed crashes with 'illegal insruction',
so seems they really hdmi specific. For that I'll disable neon
optimizations in kernel, as Marco Diane suggested, and check this.

Also thank you for hints about additional args for pulse server. I
definitely will check this.

-- 
Regards,
Alexander

Re: pulseaudio makes system unstable

[Alexander Shashkevych]

>> May [1] be related with your problems?
>>
>> [1] http://git.freescale.com/git/cgit.cgi/imx/linux-2.6-imx.git/tree/sound/soc/fsl/imx-hdmi-dma.c?h=imx_3.10.31_1.1.0_beta2#n289
>
> Interesting. I've found same #ifdefs in yocto branch I'm using with
> kernel 3.10.17, but there is no #define that disables  neon's dma for
> hdmi, so I'll define it and recompile.

Yeah! That's it! No more crashes, no more unexpected exits from apps
and even more, earlier sound was always choppy and now it is not.
Through hdmi.

I also added --disallow-exit and --exit-idle-time to pulse server and
now it doesn't exit unexpectedly.

Thank you guys! You really helped me.

-- 
Regards,
Alexander

Re: Status of imx28evk recipes/images

[Simone]

[-- Attachment #1: Type: text/plain, Size: 4122 bytes --]


> Il giorno 25/nov/2014, alle ore 15:44, Daiane Angolini <daiane.list@gmail.com> ha scritto:
> 
>> On Tue, Nov 25, 2014 at 11:24 AM, Simone <cjb.sw.nospam@gmail.com> wrote:
>> Hello Daiane,
>> Thanks so much for your answer.
>> 
>> Which Kernel version is actually supported by this board?
> 
> define supported.

I mean, we work also on imx53qsb board and we know that the last working kernel is 2.6.35.3
So, let's redefine the question: which is the latest working kernel for this board?

> 
> linux-imx is supported by freescale
> linux-fslc is from kernel.org
> 
>> What is the difference between linux-fslc and linux-imx ?
> 
> huge
> 
> Please, test and taste both, then decide which one is the one for your
> application.
> 
> (this question is really tricky, by definition, the difference between
> 2 kernels is defined by the git diff of them, however, the kernel
> version in this case differs, causing a huge (huge) gap. I cannot say
> to you what are the differences that probably can impact in your case,
> so I prefer to say nothing)
> 

Thank you, but I'm still missing something.. I suppose that linux-imx is a patched version of the kernel.org one, that include the support the specific freescale (imx) boards. Isn't it?
If yes, then I should prefer linux-imx because the freescale device drivers should not be included in fslc kernel...
Another question, how I should compare the imx kernel with the fslc one using the evk board?

Sorry to ask you such stupid things, but I feel a bit confused...

Thank you
Simone


> 
> Daiane
>> 
>> Thanks so much
>> Simone
>> 
>> 
>> 
>> -----Messaggio originale-----
>> Da: angolini@gmail.com [mailto:angolini@gmail.com] Per conto di Daiane Angolini
>> Inviato: martedì 25 novembre 2014 11:58
>> A: Simone; Otavio Salvador
>> Cc: meta-freescale@yoctoproject.org
>> Oggetto: Re: [meta-freescale] Status of imx28evk recipes/images
>> 
>>> On Mon, Nov 24, 2014 at 6:03 PM, Simone <cjb.sw.nospam@gmail.com> wrote:
>>> Dear All,
>>> In my company we are starting a new project (a low-cost industrial controller) and we are thinking about use the iMX286 CPU because it offer all what we need. Of course we would start from imx28evk for testing, then we will develop our custom board based on that one.
>>> But before all I want to know (from who is already using/mantaining it), how is the status of support (recipes, images, kernel) for such board in meta-fsl-arm.
>>> 
>>> If possible, I want to know if there are known problems or missing things, or if this cpu is not suggested for a new project..
>>> Any opinion you want to share could be very useful to me.
>> 
>> imx28evk is a beautiful board. I´ve been testing it here and there and I cannot remember serious problems. Of course, there are some:
>> 
>> * touch screen (with x11 at least) is not completely calibrated
>> * you can choose between linux-fslc or linux-imx
>> * you can choose between gstreamer mainline or the one provided by freescale (see RN [1] in order to understand the HW/SW correlation)
>> * there is a pin conflict in imx28evk board, that I don´t remember from top of my mind (I think it´s NAND + Ethernet cannot be used
>> together)
>> * NAND is already supported in meta-fsl-arm
>> * mfgtools is not already supported in meta-fsl-arm, but it´s easy to be included and your patch is appreciated.
>> * pendrive is not automatically mounted, but once you mount it everything works fine (I have never stopped to fix that)
>> 
>> 
>> that´s all I could remember
>> 
>> [1]http://freescale.github.io/doc/release-notes/1.7/index.html#document-fsl-userspace-pkg
>> 
>> 
>> Otavio, could some of this items be considered "known issue" to be included in release notes? What do you think?
>> 
>> Daiane
>>> 
>>> Thank you all for every suggestion or information you will provide
>>> 
>>> Regards
>>> Simone
>>> --
>>> _______________________________________________
>>> meta-freescale mailing list
>>> meta-freescale@yoctoproject.org
>>> https://lists.yoctoproject.org/listinfo/meta-freescale
>> 

[-- Attachment #2: Type: text/html, Size: 8788 bytes --]

Re: Status of imx28evk recipes/images

[Daiane Angolini]

On Tue, Nov 25, 2014 at 5:47 PM, Simone <cjb.sw.nospam@gmail.com> wrote:

>
> I mean, we work also on imx53qsb board and we know that the last working
> kernel is 2.6.35.3
> So, let's redefine the question: which is the latest working kernel for this
> board?

The latest working kernel for this board is linux-next from
kernel.org. It´s not even supported in FSL Community BSP as we only
include stable versions (but it´s a perfectly working kernel)

The question you want to do is "What is the latest release from
Freescale for imx28evk". And for this question you must go to
freescale.com:
http://www.freescale.com/webapp/sps/site/prod_summary.jsp?code=MCIMX28EVKJ&fpsp=1&tab=Design_Tools_Tab#


And, only to make it clear:
http://freescale.github.io/doc/release-notes/1.7/#what-the-project-name-is-not

>
> Thank you, but I'm still missing something.. I suppose that linux-imx is a
> patched version of the kernel.org one, that include the support the specific
> freescale (imx) boards. Isn't it?

See the description of each kernel provider here:
http://freescale.github.io/doc/release-notes/1.7/index.html#document-kernel-linux


> If yes, then I should prefer linux-imx because the freescale device drivers
> should not be included in fslc kernel...
> Another question, how I should compare the imx kernel with the fslc one
> using the evk board?

I don´t know. It depends on your project.

But, if you don´t know what to use or what you need, start with the
default setup, which is virtual/kernel = linux-imx (see mxs-base.inc)

Daiane

Re: Status of imx28evk recipes/images

[Simone]

Hello Daiane,
Thank you so much for your replies.
I think I will start with linux-imx as you suggested.

Thanks
Simone


-----Messaggio originale-----
Da: angolini@gmail.com [mailto:angolini@gmail.com] Per conto di Daiane Angolini
Inviato: giovedì 27 novembre 2014 12:55
A: Simone
Cc: meta-freescale@yoctoproject.org; Otavio Salvador
Oggetto: Re: [meta-freescale] Status of imx28evk recipes/images

On Tue, Nov 25, 2014 at 5:47 PM, Simone <cjb.sw.nospam@gmail.com> wrote:

>
> I mean, we work also on imx53qsb board and we know that the last 
> working kernel is 2.6.35.3 So, let's redefine the question: which is 
> the latest working kernel for this board?

The latest working kernel for this board is linux-next from kernel.org. It´s not even supported in FSL Community BSP as we only include stable versions (but it´s a perfectly working kernel)

The question you want to do is "What is the latest release from Freescale for imx28evk". And for this question you must go to
freescale.com:
http://www.freescale.com/webapp/sps/site/prod_summary.jsp?code=MCIMX28EVKJ&fpsp=1&tab=Design_Tools_Tab#


And, only to make it clear:
http://freescale.github.io/doc/release-notes/1.7/#what-the-project-name-is-not

>
> Thank you, but I'm still missing something.. I suppose that linux-imx 
> is a patched version of the kernel.org one, that include the support 
> the specific freescale (imx) boards. Isn't it?

See the description of each kernel provider here:
http://freescale.github.io/doc/release-notes/1.7/index.html#document-kernel-linux


> If yes, then I should prefer linux-imx because the freescale device 
> drivers should not be included in fslc kernel...
> Another question, how I should compare the imx kernel with the fslc 
> one using the evk board?

I don´t know. It depends on your project.

But, if you don´t know what to use or what you need, start with the default setup, which is virtual/kernel = linux-imx (see mxs-base.inc)

Daiane

Re: pulseaudio makes system unstable

[Otavio Salvador]

On Tue, Nov 25, 2014 at 4:42 PM, Alexander Shashkevych <alex@stunpix.com> wrote:
>>> May [1] be related with your problems?
>>>
>>> [1] http://git.freescale.com/git/cgit.cgi/imx/linux-2.6-imx.git/tree/sound/soc/fsl/imx-hdmi-dma.c?h=imx_3.10.31_1.1.0_beta2#n289
>>
>> Interesting. I've found same #ifdefs in yocto branch I'm using with
>> kernel 3.10.17, but there is no #define that disables  neon's dma for
>> hdmi, so I'll define it and recompile.
>
> Yeah! That's it! No more crashes, no more unexpected exits from apps
> and even more, earlier sound was always choppy and now it is not.
> Through hdmi.
>
> I also added --disallow-exit and --exit-idle-time to pulse server and
> now it doesn't exit unexpectedly.
>
> Thank you guys! You really helped me.

Not so fast! Now as a demonstration of gratitude you could prepare a
nice patch which do the define and apply it on our linux-imx recipe?
That would be awesome!

-- 
Otavio Salvador                             O.S. Systems
http://www.ossystems.com.br        http://code.ossystems.com.br
Mobile: +55 (53) 9981-7854            Mobile: +1 (347) 903-9750

Re: pulseaudio makes system unstable

[Alexander Shashkevych]

>>>> May [1] be related with your problems?
>>>>
>>>> [1] http://git.freescale.com/git/cgit.cgi/imx/linux-2.6-imx.git/tree/sound/soc/fsl/imx-hdmi-dma.c?h=imx_3.10.31_1.1.0_beta2#n289
>>>
>>> Interesting. I've found same #ifdefs in yocto branch I'm using with
>>> kernel 3.10.17, but there is no #define that disables  neon's dma for
>>> hdmi, so I'll define it and recompile.
>>
>> Yeah! That's it! No more crashes, no more unexpected exits from apps
>> and even more, earlier sound was always choppy and now it is not.
>> Through hdmi.
>>
>> I also added --disallow-exit and --exit-idle-time to pulse server and
>> now it doesn't exit unexpectedly.
>>
>> Thank you guys! You really helped me.
>
> Not so fast! Now as a demonstration of gratitude you could prepare a
> nice patch which do the define and apply it on our linux-imx recipe?
> That would be awesome!

I would be glad to do so, but to be honest, I'm not sure has it any
sense, because instability was detected on imx-linux-3.10.17,
meanwhile imx_3.10.31_1.1.0_beta already has this issue fixed and soon
it will be integrated to yocto.

-- 
Regards,
Alexander

Re: pulseaudio makes system unstable

[Otavio Salvador]

On Fri, Nov 28, 2014 at 3:34 PM, Alexander Shashkevych <alex@stunpix.com> wrote:
>>>>> May [1] be related with your problems?
>>>>>
>>>>> [1] http://git.freescale.com/git/cgit.cgi/imx/linux-2.6-imx.git/tree/sound/soc/fsl/imx-hdmi-dma.c?h=imx_3.10.31_1.1.0_beta2#n289
>>>>
>>>> Interesting. I've found same #ifdefs in yocto branch I'm using with
>>>> kernel 3.10.17, but there is no #define that disables  neon's dma for
>>>> hdmi, so I'll define it and recompile.
>>>
>>> Yeah! That's it! No more crashes, no more unexpected exits from apps
>>> and even more, earlier sound was always choppy and now it is not.
>>> Through hdmi.
>>>
>>> I also added --disallow-exit and --exit-idle-time to pulse server and
>>> now it doesn't exit unexpectedly.
>>>
>>> Thank you guys! You really helped me.
>>
>> Not so fast! Now as a demonstration of gratitude you could prepare a
>> nice patch which do the define and apply it on our linux-imx recipe?
>> That would be awesome!
>
> I would be glad to do so, but to be honest, I'm not sure has it any
> sense, because instability was detected on imx-linux-3.10.17,
> meanwhile imx_3.10.31_1.1.0_beta already has this issue fixed and soon
> it will be integrated to yocto.

It does as 3.10.31 is 1.8 material and 1.7 (Dizzy) will stay on
3.10.17 releases.

-- 
Otavio Salvador                             O.S. Systems
http://www.ossystems.com.br        http://code.ossystems.com.br
Mobile: +55 (53) 9981-7854            Mobile: +1 (347) 903-9750

Re: pulseaudio makes system unstable

[Alexander Shashkevych]

>>>>>> May [1] be related with your problems?
>>>>>>
>>>>>> [1] http://git.freescale.com/git/cgit.cgi/imx/linux-2.6-imx.git/tree/sound/soc/fsl/imx-hdmi-dma.c?h=imx_3.10.31_1.1.0_beta2#n289
>>>>>
>>>>> Interesting. I've found same #ifdefs in yocto branch I'm using with
>>>>> kernel 3.10.17, but there is no #define that disables  neon's dma for
>>>>> hdmi, so I'll define it and recompile.
>>>>
>>>> Yeah! That's it! No more crashes, no more unexpected exits from apps
>>>> and even more, earlier sound was always choppy and now it is not.
>>>> Through hdmi.
>>>>
>>>> I also added --disallow-exit and --exit-idle-time to pulse server and
>>>> now it doesn't exit unexpectedly.
>>>>
>>>> Thank you guys! You really helped me.
>>>
>>> Not so fast! Now as a demonstration of gratitude you could prepare a
>>> nice patch which do the define and apply it on our linux-imx recipe?
>>> That would be awesome!
>>
>> I would be glad to do so, but to be honest, I'm not sure has it any
>> sense, because instability was detected on imx-linux-3.10.17,
>> meanwhile imx_3.10.31_1.1.0_beta already has this issue fixed and soon
>> it will be integrated to yocto.
>
> It does as 3.10.31 is 1.8 material and 1.7 (Dizzy) will stay on
> 3.10.17 releases.

Good.
Last question: as I can see there is no way to create generic patch
that will cover all kernels, because each kernel from arm-extra layer
overrides SRC_URI with its own entries. I'm not sure how to handle
that.

-- 
Regards,
Alexander

Re: pulseaudio makes system unstable

[Otavio Salvador]

On Sat, Nov 29, 2014 at 8:46 AM, Alexander Shashkevych <alex@stunpix.com> wrote:
>>>>>>> May [1] be related with your problems?
>>>>>>>
>>>>>>> [1] http://git.freescale.com/git/cgit.cgi/imx/linux-2.6-imx.git/tree/sound/soc/fsl/imx-hdmi-dma.c?h=imx_3.10.31_1.1.0_beta2#n289
>>>>>>
>>>>>> Interesting. I've found same #ifdefs in yocto branch I'm using with
>>>>>> kernel 3.10.17, but there is no #define that disables  neon's dma for
>>>>>> hdmi, so I'll define it and recompile.
>>>>>
>>>>> Yeah! That's it! No more crashes, no more unexpected exits from apps
>>>>> and even more, earlier sound was always choppy and now it is not.
>>>>> Through hdmi.
>>>>>
>>>>> I also added --disallow-exit and --exit-idle-time to pulse server and
>>>>> now it doesn't exit unexpectedly.
>>>>>
>>>>> Thank you guys! You really helped me.
>>>>
>>>> Not so fast! Now as a demonstration of gratitude you could prepare a
>>>> nice patch which do the define and apply it on our linux-imx recipe?
>>>> That would be awesome!
>>>
>>> I would be glad to do so, but to be honest, I'm not sure has it any
>>> sense, because instability was detected on imx-linux-3.10.17,
>>> meanwhile imx_3.10.31_1.1.0_beta already has this issue fixed and soon
>>> it will be integrated to yocto.
>>
>> It does as 3.10.31 is 1.8 material and 1.7 (Dizzy) will stay on
>> 3.10.17 releases.
>
> Good.
> Last question: as I can see there is no way to create generic patch
> that will cover all kernels, because each kernel from arm-extra layer
> overrides SRC_URI with its own entries. I'm not sure how to handle
> that.

I am more concerned about the base BSP; the board specific kernels are
maintainers responsibility.

-- 
Otavio Salvador                             O.S. Systems
http://www.ossystems.com.br        http://code.ossystems.com.br
Mobile: +55 (53) 9981-7854            Mobile: +1 (347) 903-9750