Posts by RSchubert

Debugging seems to work well now with the 4.7.2 toolchain. But I still have problems with the new toolchain and buildroot.
http://www.forum.fs-net.de/viewtopic.php?f=23&t=3380

Thanks...

Buildroot-2012.11 is not working because of the same problems.

Buildroot-2013.02-rc2 is seems to work. I copied the board/f+s directory and the /config/picomod* files to the new buildroot and configured with the picomod_def_config. I get the image files and all the needed libriaries but I don't know if there are some more special changes/settings in buildroot made by F+S.

Kind regards

Added:

In buildroot-2013.02-rc2 I don't get the initial console working. I see all the information while booting but the login is not running on this serial port. What buildroot version is F+S using together with the new toolchain??

As I found in the web the problem seems to be in the gdb sources:
http://patches.openembedded.org/patch/31227/
Is there a newer F+S version of buildroot that fixes such things ?

Thanks for answers.

Error while compiling the Buildroot-2012.08-f+s-V2.0 with the new toolchain.

Also when I disable the GDB in the buildroot configuration the compiler stops with an "incomplete type" error (but in a different file with a different type).

I dont have the CTNG_LD_IS variable set.

Compiling with the 4.6.3 toolchain always was working without errors.

Thanks for advice.

I cannot acces COM1 (Starterboard) with Picomod7A.

COM3 can be accessed on /dev/ttySAC2

for COM1 I tried /dev/ttySAC0 and /dev/ttySAC1 but I get no communication.

Thanks for any hints.

Dear support-team,

I need to set a custom baudrate to the serial port. I try this with the following code:

When I run the program the first time it's killed in the

command with the message "... sets custom speed on ttySAC1. This is deprecated".

The BaudBase and the Divisor will be printed with value 0. When I run the program a second time this seems to work. Baudbase and Divisor will be printed as 300000 and 6. The program will not be killed.

Why does'nt this work when running the first time? Is this the right way to set a custom baud rate?

On PicoMod7 with kernel 2.6.39 it was not possible to set a custom speed this way. I had to run a module where I could access a special register in "PHKMEM":

But that was not very comfortable.

Thanks for any hints.

Any news about this problem? Using uclib is no option for me because threads are not working as expected there.

Thanks for answers.

Thanks for answering. Is it possible to give us a hint what amplifier and maybe what circut (resistors, capacitators, ...) your're using?

Can anyone confirm a working sound output without noise or has anyone the same effect?

I can confirm problems while debugging with the delivered toolchain.

The GDB shell says:
No line 48 in file "xxxx.cpp".
...

It's not possible to set breakpoints in the source code because the line information in the executable seems to be completely wrong. Also the first stop that should be at "main" is somewhere in the code.

I have turned off all optimisations and tried different debug infos (-g2 -g ...).

I tried gdb 6.8a and 7.4.1 provided by buildroot. The problem stays the same - so maybe it's a problem of the compiler/linker.

It's solved with the patch.

Thanks

Is there a reference hardware design of how to connect some little speakers (e.g. 1W) to the PicoMOD7A (amplifier, resistors, capacitators, speaker)?

The cause of my question is that we have much noise on the speaker. We're using an audio amplifier and a little speaker. We can hear all the sound we're playing on the PicoMOD correctly. But we also hear much noise. This noise seems to depend on memory access. If I don't do anything on the memory bus the noise is very silent. If there's much traffic on the memory bus (also during the complete boot process) the noise can be very loud - it's as loud as playing a soundfile.

Thanks for any idea.

Yes, you're right. My mistake. I did not check this, because this was working on PicoMOD7 with Kernel 2.6.29 in my version too.

Thanks a lot.

When using double buffering and switching between the buffers it's nice to wait for the VSync.

returns 0xFFFFFFFF - EPERM (operation not permitted).

In the kernel sc3_fb.c driver there's a function to handle the VSync but it never gets called:

In assembly debugging ioctl is calling "svc 0x00000000" what is returning 0xFFFFFFF2 (EFAULT - bad address).

Thanks for help.

It's working with:

Thanks.

Yes, it seems this has something to do with using sound. If I start my firmware and initialize sound

this problem occurs. This is even if I don't play anything.

When I'm using reboot and didn't start my firmware with the sound - it's working without the I2C hooking.

I also tried to run "speaker-test" and the problem is the same - I2C hooks on software reboot.

That's working. Thanks.

There's an option in the display settings of kernel configuration. Here you can set the resolution as well as the virtual resolution.

I still have some problems using double buffering. Indeed the kernel allocates 6MB of memory (1024x1536x32 bit) but the framebuffer information is still showing a resolution of 1024x768 and also virtual resolution of 1024x768.

Any Idea? Thanks.

When restarting the PicoMOD7A in software (e.g. by executing "reboot") the I2C driver isn't initing correctly and is blocking the kernel. Here I attached my kernel debug:

When rebooting by power off everything is starting correctly.

Thanks for help.